Download

A Way to Tackle Rising P&C Deductibles

Rising deductibles are shifting more storm damage costs to property owners, creating a gap between coverage and affordable recovery.

Insurance

Wind and hail losses are growing more frequent and more expensive, and the shape of that growth is easy to miss if you're only watching total loss dollars. The more significant shift, from where I sit, is in how much of that cost policyholders are now being asked to retain before coverage responds at all.

The pattern in our own data

That shift starts with how the underlying risk itself is behaving. Tornado damage, for example, is showing up in places it historically was not concentrated, a pattern visible in Adaptive's underwriting data over the past several years. Large hail has followed a similar trajectory. Through the first several months of 2026, our data shows events involving hail two inches or greater in diameter, which can damage even newer roofs, have occurred at roughly three times their historical rate.

As those loss patterns change, carriers are adjusting how wind/hail risk is shared with policyholders. One of the clearest results has been the growing use of percentage-based deductibles. A structure that was largely based on a flat dollar amount a decade ago is now commonly 2%, with 5% or higher increasingly standard in higher-risk markets and some coastal and catastrophe-exposed segments exceeding 10%.

The Midwest illustrates this well. Markets that historically carried flat-dollar $1,000 deductibles now commonly see 3% to 5% wind and hail deductibles instead. State Farm's minimum wind and hail deductible in Texas, for example, moved from 1% to 2% deductibles specifically in response to loss frequency in the Dallas-Fort Worth area, a documented, public example of the broader trend.

Run the math on a property insured for $1 million: a 2% deductible means $20,000 out-of-pocket before the primary policy pays a dollar. At 5%, that's $50,000. For larger commercial assets, the absolute numbers scale accordingly. Chicago's Dearborn Station, for example, carries a wind and hail deductible of more than $400,000 under its primary policy.

What retained risk actually does to behavior

The size of the deductible matters. But the more revealing question is what happens to decision-making after a loss, once a deductible is large enough to strain a property owner's finances.

I've watched this play out directly in my own building, a high-rise of more than 30 floors in downtown Chicago. Over the past three years, the property has sustained repeated wind damage to its garage door, with each incident landing just under the policy's deductible threshold. After a significant wind event in March, the HOA changed the building's operating procedure. Instead of filing a claim, it left the garage door open during the day and closed it only at night to reduce further wind exposure.

That may be the most practical financial decision, but it also creates new concerns around issues like security and access. It is a good example of something that broader loss data may not show: a property can be insured, but the deductible may still be high enough that filing a claim does not make sense.

That single building isn't an isolated case. Estimates commonly cited from FEMA put the share of small businesses that don't reopen after a disaster at around 40%. Separately, a 2026 Housecall Pro survey found 77% of homeowners are delaying or scaling back home projects due to rising costs, and 41% report having delayed a repair that ultimately cost more as a result.

Higher deductibles help carriers manage loss volatility, but they also require policyholders to retain more of the financial risk. When a business or homeowner cannot realistically cover that upfront cost, repairs may be postponed. The original damage can worsen, operations can remain disrupted, and the eventual cost of recovery may continue to grow.

For businesses, the effects can extend beyond the property itself to employees, customers, lenders, the surrounding community, and, ultimately, the insurance providers that serve them.

Where deductible buy-back fits into the picture

Wind and hail deductible buy-back coverage is designed to address this specific exposure. It is a supplemental layer that sits alongside the primary policy and reduces what a policyholder ultimately pays after an eligible loss.

Two examples show how the coverage can work across different property sizes:

A small commercial policyholder, such as a children's gymnastics studio carrying a $20,000 wind and hail deductible, could purchase coverage that reduces its ultimate out-of-pocket responsibility to $5,000. The supplemental policy could reimburse up to $15,000 of the deductible, subject to its terms and conditions, for an annual premium of approximately $450.

At the other end of the spectrum, a multiuse commercial building that sustains $1 million in storm damage and carries a $500,000 deductible under its primary policy could see that retained exposure reduced to as little as $10,000 with buy-back coverage in place.

The mechanism is the same in both cases: the coverage reduces the amount the policyholder ultimately retains after an eligible loss.

Depending on the timing and terms of the claim, the property owner may still need to fund some repair costs before reimbursement is received. Once paid, however, the deductible buy-back coverage can help restore working capital and reduce the longer-term financial and operational impact of the loss.

What this means going forward

A new gap is emerging. It is different from the traditional insured-versus-uninsured split.

This one sits between coverage that technically exists and coverage a policyholder can realistically use to recover after a loss.

As deductibles rise, more policyholders may find that the amount they are expected to retain is difficult to absorb. The consequences appear in delayed repairs, deferred maintenance and decisions like leaving a garage door open during business hours because the alternative is not financially manageable.

Closing that gap will take traditional and specialty carriers working from different angles. Deductible buy-back coverage is one of them, a way to give policyholders more control over what they retain and to treat the deductible as a variable worth managing, not a fixed cost of doing business.

It will not be the only solution. As loss patterns, property values, and carrier capacity continue to change, the market will need a range of products that distribute risk more effectively while keeping recovery financially achievable.

The goal is to make sure the risk that remains is something an organization can actually afford to act on when it materializes.

Sources: Adaptive Insurance internal underwriting data; State Farm public deductible notices (Dallas-Fort Worth); Housecall Pro 2026 homeowner survey; FEMA

What Insurers Must Get Right on AI

Converged platforms can unify underwriting, claims and policy data in real time, letting AI act on complete context rather than fragmented systems.

Insurance

There is always going to be another new platform, another integration, and, increasingly, another conversation about AI. But the value of technology isn't in simply having more of it. It's how well it connects and whether insurers get better information when they need it.

At a time when AI is redefining how the insurance industry will function for the next 30 years, a converged environment unifies underwriting, policy administration, billing, claims, and customer engagement on one operating platform, rather than a chain of systems handing data to each other. A claims signal reaches pricing while the claim is still open, informing the next risk assessment in real time. An underwriting decision immediately shapes how service and retention teams treat that customer afterward. Information moves in near real time, following what the business needs rather than the limits of the original system architecture.

The Case for Converged Platforms

Across this platform, AI becomes part of how the workflow runs. An agent reviewing a claim has policy history, the underwriting file, and prior customer interactions available at the same moment, because they live in the same context rather than three different logins. That context lets AI move from flagging an exception to resolving it, and from recommending a next step to taking it, with a human reviewing the outcome afterward. The reasoning behind each action stays visible and auditable, so the people accountable for the outcome can see exactly how the system reached it.

The practical benefits show up quickly. Teams configure new products within a shared environment instead of rebuilding them system by system, so what used to take months of coordinated releases now takes days. Renewal pricing, fraud screening, and service routing draw on the same live data rather than nightly batch exports, so all cross-referenced with each other and reflect the real-time state of the business.

The autonomous AI platform we are describing doesn't mean anyone should abandon the judgment that experienced underwriters and claims handlers bring. A converged environment pairs automation with quality gates. An organization can route a routine renewal straight through and hold a complex commercial risk for human review, using the same data for both. The goal is to make sure the people making the decision are working from the full picture, instead of whatever fragment their siloed, legacy systems offer.

Rethinking Process Design in an AI-First World

It is important that companies should think carefully about the shape of their AI models. There is a natural tendency for people to focus on human-driven, effort-based, bottom-up processes and then apply AI efficiency factors to achieve better productivity results. I believe that approach is fundamentally flawed as it quietly assumes the old process as the correct baseline, when the old process is the very thing that should be questioned.

Companies should work back from the assumption that every function is AI-automated and design that as the starting point, not the improvement point. Then justify, line by line, the human effort that is genuinely needed to be added back: the judgment calls, the client-facing decisions, the regulatory sign-offs, the things an autonomous AI agent demonstrably should not be permitted to do.

The best approach is to question the requirement itself and what value it delivers to the business; wherever possible, delete the step entirely rather than automate it, simplify the real requirements that survive, and only then automate the complete end-to-end process with the minimum human intervention possible. There is no business value in automating a step that should have been removed completely, as this is the most expensive mistake that organizations make.

The clearest sign that this transformation is working comes down to a simple test. Can a service representative see a customer's full history the moment the call connects, without transferring them or putting them on hold while three systems are checked in sequence? Can an underwriter check how similar risks have actually performed against what the rate manual predicts? Can a claims handler settle a straightforward loss the same day it's reported, because the verification, the policy check, and the payment authorization all draw from the same source of truth?

That is the standard worth building toward. For companies where coherence is increasingly a transformational priority, my advice is to build an environment where every part of the business can see what every other part already knows, and where AI has enough context to act on that knowledge instead of merely describing it.


James Hannay

Profile picture for user JamesHannay

James Hannay

James Hannay is chief revenue officer at Sapiens.

Previously, he served as chief growth officer at HCL Software and EVP and general manager at Alight Solutions. Prior to that, he held senior roles at Infor.

NAIC AI Risk Ratings Let Errors Slip

Insurers' self-assessed AI risk ratings determine regulatory scrutiny, but rating customer-facing models too low leaves coverage errors undetected until claims arrive.

Insurance

The NAIC's AI Risk Evaluation Supplement is the questionnaire regulators will draw on to examine how insurers govern their AI. It asks whether a model's outputs were tested for accuracy, but only for the models a carrier itself rated high risk. When a customer-facing model gets a coverage question wrong, the policyholder acts on the wrong answer. 

That is the exposure a high inherent risk rating exists to capture. Rate one low instead, and the inquiry ends there. The model keeps answering coverage questions, the answers keep reaching policyholders, and nothing requires the carrier to check whether they were right. Few carriers check on their own.

What a low rating leaves unchecked surfaces later, in a claim file rather than an examination. Misrepresenting policy provisions is a listed unfair trade practice in nearly every state, and a dated coverage statement the policy contradicts is what a misrepresentation claim is built from. The customer who heard "you're covered," paid for the repairs, and then had the claim denied is the one who brings the complaint.

When that complaint arrives, a court will ask how often the model's answers were checked and what the checks found. Errors and omissions underwriters, who now ask about AI at renewal, want the same answers. A carrier that rated its model low will have no file to produce. There is no record of what the model said and no comparison against the forms in effect at the time. 

What it will have is a self-attested risk assessment concluding that the model was low risk. Plaintiffs' counsel look for exactly that pairing: a carrier that never checked, and a document explaining why it did not have to.

How the rating works

The rating is set early, in an inventory. Each carrier lists its AI models, describes what each one does, and assigns each an inherent risk level. Only the models rated high move on, and there the questions turn specific: were the outputs tested for accuracy, how is performance monitored on a continuing basis, and how is the model reviewed against unfair trade practice and claims settlement laws? 

Models rated moderate or low do not reach those questions, and a regulator reading the inventory has discretion to stop at the rating and request nothing further about the model. That is sensible design, since regulators cannot examine everything. It also makes one self-assessed rating the single point of failure for whether a model's accuracy is ever examined, and the carrier is the one holding the pen.

How a customer-facing model goes unexamined

Two judgments decide whether a customer-facing model is ever examined. Does its output reach a consumer, and how much risk does it carry before controls? A carrier can get either one wrong without meaning to.

Destination settles the first question, not who speaks the words. The supplement's background section covers decisions and actions “made or supported by” AI, and a coverage answer that reaches a policyholder is one of them. The pitfall is the supplement's support category, defined as a system that provides information without suggesting a decision or action, which is not counted as having direct consumer impact. A model that drafts a coverage answer is not just providing information. It is supplying the decision the representative delivers, and the support definition excludes exactly that. But because a representative delivers the answer, the model looks on paper like one that only informs an employee, and that resemblance is what puts it in the wrong box. Filing it as support keeps the model out of the count regulators use to decide what to examine, and out of any question about whether its answers were right. Those answers still go out to policyholders, unchecked.

A model that is counted can still be rated low by mistake, and the same representative is usually why. Most carriers put two guardrails around a model's answers. The representative is expected to check it before it goes out, and the platform runs its own internal check against policy language. Both are controls, and because they are built into the process, crediting them when rating the model is an easy mistake to make. The supplement asks for inherent risk, the risk the model carries before any control is applied, and the word "inherent" is easy to read past. Under the supplement's own terms, the representative and the platform check belong in the residual rating, after controls. A carrier that credits them in the inherent rating has rated its model lower than the supplement intends, and taken it out of the testing questions in the process. Both judgments are already recorded, one line per model, in the carrier's own inventory.

Why the cheaper rating costs more

Rating a model low is the cheaper answer. It closes a line on the inventory, raises no question the carrier has to answer, and creates no obligation to find anything out. Rating it high creates recurring work: finding out whether the model's coverage answers were right and continuing to do so while the model runs. The pressure runs one direction, and nothing in the supplement pushes back. The saving lasts a quarter. The gap it leaves is permanent. That work cannot be backfilled later, because the evidence exists only while the conversations are happening, and controls do not close the gap. A representative who catches a wrong answer fixes that conversation and leaves no record of it, so even a carrier with an excellent review process cannot separate an isolated error from a model-wide one. It also cannot see drift. Models do not hold still after launch. They get updated, the policy forms they draw on are amended, and the questions customers ask shift. A test run before deployment describes the model that was deployed, not the one answering calls this quarter. What a low rating costs is a year of conversations nobody measured, and nobody can go back and measure.

I cannot tell you how often AI models give wrong coverage answers, and neither can anyone else, because almost nobody is measuring. The draft is open for comment through Sept. 29, with a vote on a later version expected in November. The questions are being written now, and the evidence they ask for can only be gathered before they are asked. Rating a model low does not change what it told a policyholder. It only changes whether anyone was looking.

Insurance Agentic AI Needs New Business Models

Asian insurers show that agentic AI success requires reimagining market forces, not just chasing productivity gains through isolated use cases.

Insurance

What if your computer didn't just help you type or access YouTube but acted like a smart assistant instead? That's what we broadly call agentic AI. Some might think AI is here to help us improve productivity and reduce costs, but the change is much deeper (Afanasyev, 2026). It defines what the insurance industry would be in five years and how your organization should adapt.

Asia's First-Mover Advantage and Early Missteps

Asia is the first-mover in AI adoption among financial institutions, compared to developed markets often hindered by tech debt and bureaucracy. The learnings here uncover that, for the industry, agentic AI is a change in the "rules of the game."

Asian financial institutions originally rushed into Agentic AI by treating it like a project focused on multiple use cases, prioritizing 'impact vs. feasibility'. This approach targets quick wins, assuming the organization is the only player implementing AI and hence making the journey risky and expensive. No surprise to see both academia, MIT (Challapally et al, 2025) and consultants, BCG (Apotheker et al, 2025), report that 95% of organizations struggle with Agentic AI. Such organizations miss that their customers and partners are also deploying AI Agents (Citrini and Shah, 2026).

When Productivity Gains Backfire: The Email Case Study

What happens when Agentic AI use cases are assessed without context? Take AI for emails, a productivity use case deployed three years ago, which used AI to help write emails or summarise the received emails. It seemed like a win but lacked context; certain employees used AI Agents to imitate work and secure visibility by generating volumes of emails. AI-generated content is hard (with some researchers claiming impossible) for a human or an algorithm to distinguish from genuinely important messages. The recipients would have to apply AI Agents to search for information. Both senders and receivers myopically reported AI bringing gains, although the insurers likely deployed a use case that resulted in miscommunications and productivity drains.

The New Reality: Everyone Has AI Agents

The right approach for an insurer deploying AI is to accept that others are doing the same, so focusing on internal productivity may not be enough. Think about claims. Your policyholders' and providers' AI Agents are scrutinizing policies and preparing documents to maximize approval. Recall your dental insurance policy, which offers reimbursements only if certain symptoms, such as tooth pain, are present. The AI Agent ensures that the symptom is stated, even when the policyholder visits dentists for routine checks.

In underwriting, similar to how Mythos from Anthropic identifies cybersecurity vulnerabilities (Bloomberg, 2026; Azhar and Williams, 2026), your customers' agents might be identifying gaps. Insurers whose underwriting processes do not handle AI-generated applications may end up underpricing risks. For wealth management, why would a customer having access to LLM models pay advisors who use generic LLMs for generating ideas?

Think about distribution, where an insurance broker deploys AI Agents to identify and message prospects. The competing brokers are doing it too, leading to overwhelmed prospects who might avoid purchases altogether.

The latter is highly illustrative and applies in reverse to a financial institution thinking about its agentic strategy. Previous breakthroughs, such as electricity and the internet, accelerated business but the ability to make intelligent decisions broadly remained intact. AI Agents generate volumes of persuasive, credible content but the overload reduces decision-making ability.

The "Business First" Approach: Three Pillars for Success

Asian financial institutions concluded that to succeed, they need to shift to the "Business First" or "Applied AI" approach, which combines three views—Business domain, Transformation, and Technology—under a single leader (Afanasyev and Milind, 2026). McKinsey used to name leaders with such skills as "Advanced Analytics Translators" and, since the early AI days, has been urging organizations to grow leaders who "ensure that organizations achieve real impact from their analytics initiatives" (Henke, 2018).

Pillar 1: Business Domain — Reimagining Market Forces

The Business domain reimagines how market forces will change, how the industry will provide value, and what an organization's role will be in the long run. This is the least-discussed pillar and precisely what puts organizations at risk.

The majority of Asian organizations are privately controlled, meaning limited dependency on reporting cycles and an ability to prioritize long-term needs. Rapid economic changes over the past two decades have led them to embrace practicality and challenge the status quo. For instance, in developed markets with high labor costs, many insurers seem to be prioritizing Agentic AI technology to improve efficiency and cut costs. Most Asian markets are, however, known to have low labor costs, and local institutions have rightly questioned us about deploying AI for just operational efficiency in the regional context. However, young people, the largest segment of Asia's population, adopt AI Agents for their financial needs the fastest. This has accelerated economic changes not yet noticeable in developed markets.

How Generative AI Disrupts Insurance Fundamentals

We have discussed how Generative AI disrupts signaling theory, which is foundational for modern insurance. In some cases, the policy buyers may know more about the likelihood that they will suffer a loss than the insurance company. The 'market for lemons' theory shows that only the ' worst ' buyers (expecting larger, more likely losses) may end up interested in a policy. Signaling theory explains how parties with superior information (actors, for example, policyholders) convey attributes to parties with less information (decision-makers, for example, policy underwriters) to overcome information asymmetry. For the industry to function, credible signals—think medical records for life insurance—must be costly to imitate. Generative AI has empowered any actor to generate strong signals, disrupting the fundamentals of risk-taking, with empirical studies already demonstrating an erosion of trust.

How can actors credibly signal quality when signals can be cheaply fabricated? The answer, we believe, is grounded in mechanism design, the economics discipline concerned with a decision-maker designing the rules so that rational actors reveal truthful attributes even when their interest might be in using Generative AI. The Applied AI leader has to possess deep insurance domain knowledge and also a practical level of academic perspective. Such a leader analyses disruptions to come up with a view on what the organization's role would be. The outcome could be that the value proposition should be in risk prevention instead of risk pricing; changing target customer segments; vertical integration bringing certain services in-house; or leveling AI to address moral hazards.

Pillar 2: Transformation — Rethinking Processes and Culture

Transformation should focus on agility, changes in processes, decision-making, and human roles. "A growing number of studies reveal that human–AI systems do not necessarily achieve better results than the best of humans or AI alone" (Vaccaro et al, 2024). Many organizations ground their AI performance evaluation on efficiency, accuracy, or productivity. However, "nowadays, productivity gain is no longer the single evaluation criterion. In many instances, computer systems are expected to enhance our creativity, reveal opportunities and open new vistas of uncharted frontiers" (Avital et al, 2009).

Legacy institutions have to unwind certain cultures religiously followed for years. Certain advancements have ensured that employees follow rule books and processes prescribed by management. As a result, many organizations operate as factories, with processes equivalent to production lines of the early 1900s. Employees and mid-management have limited or no discretion outside of what is prescribed. This practice might have been appropriate earlier, but the Agentic AI economy is unpredictable; rule books and processes become outdated in the blink of an eye.

Pillar 3: Technology — Building the Infrastructure for AI Agents

The Technology view is most discussed and depicts how non-human resources are augmented by AI Agents. The learning from Asian progress in Agentic AI is a need for Applied AI leaders to take a broad view on financial industry technology and ground hot topics, such as tech debt, data quality, MCP and A2A, Agentic AI workflow designs, etc.

Asia's Unique Technology Landscape

Asia is home to 5 billion people—about 10x of Western Europe and US populations combined—with interactions among various ethnic groups who have their distinct languages, cultures, and business styles. Asia accounts for about 50% of non-cash transactions globally (Capgemini, 2025), with most countries, including India and China, already cash-free and unimaginable GDP growth in developed markets. Asian institutions deal with unprecedented transaction volumes and diversity, so scale and agility have traditionally been prioritized over politics and offering sophisticated products. Moreover, many overly complicated software products adopted in developed markets for the past two decades found little relevance in Asia.

Before LLMs and Agentic AI, the Asian tech ecosystem had been focused on developing local software solutions to handle volumes. Native software providers have mastered agentic workflow designs using Directed Acyclic Graphs with state management, allowing agents to "Plan, Act, Observe and Reflect". MCP, the "USB-C for AI", has landed in Asia as a standardized interface, allowing agents to pull real-time context from sources like SQL databases, CRM systems, and core banking. For developed markets, MCP would solve the chronic pain point of tech debt by creating an abstraction layer over legacy core systems. Complementing this is the A2A protocol, which facilitates peer-to-peer coordination, and AP2, empowering agents to do financial transactions with or without human presence.

Architectural Challenges: Making AI Agents Work Together

Most legacy systems were architected for workflows that functioned adequately for rule-based operations. Meanwhile, Agentic AI demands open interfaces where workflows are not hardcoded and any given process can be decomposed, assigned to an agent, and reassembled. Organizations are adopting AI capabilities in separate departments and in relative isolation. The containment limits risk, allowing teams to demonstrate value; the actual issue is how these modules function as a coherent whole.

Agent loops, where Agent A triggers Agent B which re-triggers Agent A, require modern architectures to include circuit breakers that terminate chains exceeding defined depth or duration thresholds. Agent disagreements surface when two agents with complementary mandates reach different conclusions. Authority boundary violations occur when agents act beyond their defined scope.

Data Governance and Schema Management

A schema-driven model is recommended. Insurers should adopt a master schema configurator that allows users to govern agent behavior without code. Insurance is challenging in the volume of its unstructured data. Decades of paper-based documentation and inconsistent digitization have left organizations with data states that cannot be queried in real time. For Agentic AI to function, three areas are foundational. Vector databases enable semantic search across document corpora, allowing context retrieval. Structured extraction pipelines convert unstructured documents into queryable records. Master data management maintains reference datasets which can be used for real-time lookup, matching and adjudication.

Regulatory Safeguards: Ensuring Accountability and Control

As regulators, we are of the view that autonomous systems making customer-impacting decisions must be auditable, explainable and interruptible (IAIS, 2024). To ensure agents do not overstep authority, safeguards must be included. Confidence scoring requires that every AI output carry a numerical confidence estimate, with automatic escalation to human staff when results are low. Mandatory review workflows gate high-stakes decisions behind human approval that cannot be bypassed. Continuous feedback loops mean that each correction trains the models. Comprehensive audit trails provide thorough records. 

A Five-Step Modernization Roadmap

A five-step modernization approach is required:

Step 1: Start with structured discovery. Most attempts fail because organizations jump to engineering straightaway. Insurers must produce a clear blueprint of their processes. This discovery phase should identify opportunities to separate capabilities into components — units that can become plug-and-play services with open APIs, allowing future agents and partner systems to come in flexibly.

Step 2: Separate design and validation from execution. Insurers should introduce a solutioning phase where a small team defines the future state by determining which capabilities will be rebuilt or replaced, how workflows should operate, etc. Only once the design has been validated should engineering teams begin executing. AI systems can assist by extracting undocumented patterns from interviews, translating requirements into architectures, and simulating potential models to stress-test assumptions.

Step 3: Build in parallel. The target platform should be developed independently while legacy systems continue running operations. A specific sequence is recommended — load non-member-facing master data first, accepting a short window (as brief as possible) during which updates may need to be reflected in both legacy and new systems. Once it's loaded, validate the setup by running test policies.

Step 4: Migrate data through controlled, incremental processes. A single large cutover concentrates risk and creates extended recovery windows if things go south. Hence, a controlled migration strategy is beneficial. Bring the new system online and then migrate data in staggered batches. Not all data needs to migrate; historical data is often better suited for archival storage or data lakes. Before migration, the new platform should be tested in lower environments. Once validated, the system can go live while legacy systems continue managing the existing portfolio. AI systems support this by inferring undocumented schemas, mapping fields, and extracting structured data from records.

Step 5: Retire legacy systems. The final, frequently overlooked step is decommissioning legacy systems. Leaving them partially operational introduces unnecessary cost and risk. Instead, insurers should archive historical data where necessary, ensure compliance for record retention, and fully decommission said infrastructure.

The Kodak Moment: Adapt or Become Obsolete

Organizations focusing on productivity use cases miss the big picture, but the insurance industry will likely not exist in its current form in five years. Think about Kodak 30 years ago: a leader in optimizing film production costs that missed how digitalization changed preferences. In this democratization, insurers need to shift to business models redesigned to cater to change and structure their AI journeys accordingly.

For the full white paper from which this article is adapted, click here.


Maxim Afanasyev

Profile picture for user MaximAfanasyev

Maxim Afanasyev

Maxim Afanasyev is a senior executive in AI, solutions and product management at Google.

He has more than 20 years of experience in academic research, management consulting, technology and financial services and is an advisor to C-levels of major companies and prominent entrepreneurs.

Afansyev has a PhD from Stanford University in operations, information and technology. He is an adjunct associate professor in AI at the National University of Singapore.


Tomas Holub

Profile picture for user TomasHolub

Tomas Holub

Tomas Holub is the CEO and founder of CoverGo, an AI insurance platform for health, life, and P&C. 

Prior to starting CoverGo, he worked as an insurance and banking consultant at PwC London and also head of operations of an insurance technology company in Singapore. Overall, he has worked in 20 countries across three continents, speaks eight languages and obtained four masters degrees in risk management, international business and public and business administration.

Holub is a frequent speaker at conferences.

Insurance Innovation: The View From Asia

While the "not invented here" attitude is well-known, there is an "invented here" bias among us Americans when it comes to innovation. The view from Asia looks rather different. 

Image
Insurance

The general assumption about innovation is that it's top-down. Someone invents something complicated and expensive. As it achieves scale, the innovation becomes simpler and less expensive, making it accessible to broader audiences. Rinse and repeat until the innovation becomes global.

In insurance, that trend has typically meant an innovation in the U.S. or Western Europe that finds its way over time into developing economies.

But there's also an interesting trend known as "reverse innovation" that I've watched play out for some 15 years. If you have a multimillion-dollar medical device, you can take cost out, time and again (and again and again), and still not make it cheap enough to be economic in a poor, rural community. So some smart folks in India working within the severe cost constraints of a poor area invent a version of the medical device — and sometimes that device works well enough that it can reverse the normal flow, migrating from developing areas into the major economies.

To learn more about what reverse innovation is happening — or could be happening — in insurance, I turned to George Kesselman, the founder of InsurTech Asia. He grew up in Canada and went to school there, before spending the last nearly two decades working in Hong Kong and now Singapore, so he has a thoroughly global perspective.

His short answer: Pay very close attention to China.

His longer answer follows.

I met George through a sister organization, the International Insurance Society, for which he has been hosting a working group on innovation that I'm part of. (George will present the findings from the IIS's global survey on insurers' priorities, as well as insights from our working group, at the IIS' annual gathering, the Global Insurance Forum, being held in London at Lloyd's on Nov. 15 and 16.) Here is what he said about innovation in Asia, edited slightly for length and clarity:

KESSELMAN: This year, I had the pleasure of doing a coast-to-coast tour over two weeks in the U.S. and Canada, spending a few days in New York, Toronto, Vancouver, and San Francisco. My read is that innovation is really quite different in North America than what's happening outside. 

The last wave of innovation, from roughly 2014 to 2023—let's call it a 10-year span—was very much an Asia-first wave. It was all this digital insurance that started in China, then Southeast Asia, with a lot of new models when it comes to embedded, small insurance. Then I think the innovation went a little bit more to Europe. In North America, it hasn't really taken off in the same way.

This time, the wave is kind of inverted. With all the AI innovation, all the investments, it's North America first, then Europe. Asia is seeing bits and bobs but not nearly as much as the other two regions.

My theory is that the scale is massively different in the regions, and the problems are also quite different as a result. Asia is very much about the growth of insurance. Insurance wasn't really popular here because the region was a lot less developed. The last 10 to 15 years have seen tremendous economic growth, so innovation and insurance started to come in. Insurers couldn’t scale the same face-to-face distribution, so they found different ways, including digital distribution.

But the scale still remains drastically different. When I was in North America and Canada, everybody talked in terms of billions of dollars. In Asia, people talk about millions, maybe sometimes tens of millions. 

CARROLL: What kinds of embedded insurance innovations emerged in parts of Asia that have filtered out to other parts of the world?

KESSELMAN: In e-commerce, we saw quite a bit of innovation in warranties and the return shipping space. Now we're seeing innovations around trying a new product and getting coverage if you don't like it—you experience an upset stomach after eating something, or face delayed delivery. These are very region-specific and emblematic of the logistics and distribution challenges in Asia.

Travel is another ecosystem that saw significant innovation. Mobility is the third. Then fintech-related insurance is probably the fourth. Each has its unique characteristics. In fintech, for example, we've seen a lot around cyber incidents—if your wallet gets hacked, you get paid—or bill payment protection, where if you get sick, the bill gets paid on your behalf. These are embedded and bundled propositions that are very continuous to the core offering.

These innovations started here because digital leapfrogged in a lot of countries where the infrastructure didn't exist before and really accelerated adoption. Last week I talked to one of the big e-commerce players here, and they estimate about a billion dollars in these micro embedded insurance propositions being conducted yearly now. The growth is definitely there, and the trend has reached scale across the region, though it's still relatively small compared with an average MGA in the States.

As for how this is spreading, some of it is now landing in Europe. Warranty products, lens insurance, hearing aids—these are probably more specific to Europe because the needs and market development are different. In the mobility space—scooters, car insurance, and ride-sharing—I'm hearing that's happening there, as well.

CARROLL: I assume China has played a big role.

KESSELMAN: Embedded and digital insurance really took off from China. The big players were Ping An and ZhongAn—both kind of protection-focused players—and both of them really aggressively invested in innovation. China probably remains three or four years ahead of the rest of the region.

From China, innovation propagated to Southeast Asia. I think India is now also quite a significant market for it, but China was very clearly the first because of the combination of digital payments penetration being super high there, the development of super apps that put everybody—all the distribution—in the same place, and then e-commerce and all the logistics. Everything developed at the same time pretty rapidly.

CARROLL: What are some examples in Asia that other parts of the world could learn from in terms of insurance innovation?

KESSELMAN: Ping An and ZhongAn remain interesting examples. They have been investing very actively in AI, and I think they're still very much focused on distribution and servicing, and developing propositions that are very targeted.

In India, there are some insurers that are starting to do some really interesting stuff. Acko is one that comes to mind in the mobility space that has been quite active. The rest of the region has pockets of innovation, but because the countries are generally smaller in scale, it's very hard for insurers to really grow very rapidly. You know, you start in Singapore, then you need to expand into a neighboring country, and then you need to go and redo the whole growth journey again.

CARROLL: What sorts of interesting startup ideas are you seeing at Insuretech Asia?

KESSELMAN: There have been a couple of waves of startups. The first was really about trying to disrupt distribution. That wave has kind of passed now.

As part of the second wave, we have a couple of really interesting startups. Two are unicorns. Bolttech is one. They focus on distribution and have a couple of different complementary models. They have a very big presence in the U.S. and a global presence, but their headquarters are in Singapore.

Then we have InsureMO, which is like an infrastructure for insurance distribution. They have very big operations in India and Southeast Asia, and now they have growing operations in Europe, the U.S., and Latin America, as well, but they’re also based in Singapore.

CARROLL: What makes those companies different?

KESSELMAN: I think it’s exactly the fact that they started in Southeast Asia, so they needed to do everything very cheaply and fast even though the platforms that are here are very demanding in terms of the operational support that’s needed, in terms of technology. Being incubated here, I think, gave them that natural foundation of really building a scale-up in a demanding environment. They tell me that when they go to the States and talk to insurers there, insurers are very surprised that they can do things for a fraction of the price and at double the speed.

CARROLL: What has surprised you about how AI is being adopted—or not being adopted?

KESSELMAN: The part that surprises me is that it is still very much a technology conversation rather than a business conversation. To give you an example: Recently a couple of insurers in the region came here, and they have flagship programs, but they're all kind of stuck. They don't really know how to solve the problem of creating business value with AI.

This is the same theme we saw in the IIS survey.

Everybody kind of saw this shiny object and had experience with really interesting outputs of this technology, but there's a disconnect between the solution that exists and the problems that insurance companies need to solve. I also think there's a disconnect between the leadership level of the organizations and middle management organizations, which is not as bought into this whole idea of "let's bring AI everywhere, let's augment every job with it."

CARROLL: That's such a common problem—bosses demand change, and underlings nod their heads… but they also find ways to keep doing things the way they always have. Peter Drucker once told me that people don’t change. They just die off, and the next generation then does things differently. But we were talking 25 years ago. I hope we’ve learned a bit since then.

This has been great, but any final thoughts?

KESSELMAN: There are interesting developments in health insurance, again starting with China. Everything is getting so integrated. Hospital data is getting a lot more integrated with insurance data, and it's getting linked up in the hospitals to make it much more proactive—managing health, managing well-being. So rather than putting strain on hospitals, with insurance being really at the back of it, the loop is starting to get closed.

The rest of the world will eventually need to tackle that integration issue as populations age and healthcare systems are probably redesigned. Insurance right now is still very much on the receiving end.

That integration in health insurance is probably one of the most interesting things happening in this part of the world, and, again, I’d use China as the reference market. 

CARROLL: Thanks, George. This is super helpful.

Cheers,

Paul

Why Fear Doesn't Sell Insurance

Insurance marketing's central challenge isn't explaining risk—it's building trust that makes invisible future value feel worth purchasing today.

Insurance

Insurance marketing has a paradox: You can motivate people into action by showing them the scale of a risk, but that does not explain why they should choose you.

In markets where the perspective on insurance differs, explaining the risk is often not the problem itself. As an insurance researcher operating from a brokerage desk, I observe that a significant portion of the Turkish insurance market is not driven solely by consumer demand. Compulsory motor third-party liability insurance, DASK (compulsory earthquake insurance), and various professional liability policies—these are all products mandated by the state.

The fundamental question is this: Does making a risk "mandatory" ensure that the individual truly internalizes it? This calls into question the sector's most basic assumption — that raising awareness, whether through compulsion or fear, is enough to generate demand.

Often, it is not. The core problem is not that people are unaware of risk. The problem runs deeper: the value of insurance is not visible at the moment of purchase. And trying to sell an invisible value through fear alone might explain why someone needs insurance — but not why they should trust that specific insurer.

Fear Belongs to the Category, Not the Brand

Imagine seeing a crash scene in an insurance commercial. Cars collide. A family is worried. An insurance company's logo appears on the screen with the message, "We are by your side."

But what does the consumer remember? Often, the accidents. Perhaps the fear. But the brand?

This is one of the fundamental paradoxes of insurance communication. Because another insurer could use that same fear. Fear can explain why insurance exists, but it cannot explain why one insurer should be chosen over another.

Fear can grab attention. But to create preference, something else is needed: trust.

What Is Insurance Actually Selling?

When you buy a phone, you can experience the product's value. With insurance, the ideal scenario is that you never actually use what you have purchased.

The value of a policy often becomes apparent not at the moment of purchase, but at some uncertain point in the future. For this reason, when evaluating a policy, a consumer looks beyond mere coverage.

Consequently, the primary question in the consumer's mind is often not, "What does this policy cover?" There is a more fundamental question: "If something happens to me one day, will you really be there for me?"

This is precisely the intangible product of insurance: trust regarding a future moment of vulnerability.

Invisible Value Loses to Visible Price

This is quite evident on the brokerage side. In an environment where multiple providers offer similar products, the conversation quickly shifts to price. Instead of asking, "Which one is better?" the customer asks, "Which one is cheaper?"

This is particularly common in the Turkish market regarding compulsory motor third-party liability insurance. According to the official table published by SEDDK (Turkey's Insurance and Private Pension Regulation and Supervision Authority), the coverage limit for property damage per vehicle for 2026 is 400,000 ₺ [1] —identical across all companies. When coverage is standardized, price becomes the easiest signal to compare.

But what is truly interesting is that even with non-compulsory products where coverage does vary, the conversation still boils down to price. For products like comprehensive motor insurance or health insurance, providers offer varying scopes of coverage and services—yet consumers often fail to assess these differences. Price is visible, whereas trust is not. You can read the coverage limit on the policy, but you cannot test how fairly a company will act when a claim arises at the moment of purchase.

This is precisely where the marketing challenge for insurance lies. A marketer's task is not merely to explain the product but to make invisible value visible.

Consumers must place their trust today in a behavior that may occur in the future. Therefore, insurance is not just a risk-transfer product; it is also a product of the trust economy. When making a purchase, consumers cannot directly determine the product's quality; instead, they must rely on indirect signals. At the moment of decision-making, they evaluate factors such as brand reputation, communication style, transparency, and broker recommendations.

And something so difficult to evaluate is often something the consumer chooses to postpone.

We Protect Our Cars More Easily Than Ourselves

I see this most clearly in health insurance. Many consumers find it quite natural to take out comprehensive motor insurance (kasko) for their cars. From a broker's perspective, I observe that people find it easier to protect their cars—external objects—than their own bodies. But why?

Because it is easy to imagine a car suffering damage. Contemplating how our own health might change down the line, however, is far more unsettling.

Purchasing health insurance is not merely a financial decision; it requires a person to envision themselves in the future as someone who is sick, frail, or in need of assistance. This is where optimism bias comes into play: "I am healthy right now. Nothing is likely to happen to me."

When it comes to cars, it is easy to acknowledge that "an accident could happen." When it comes to our own bodies, however, we are more inclined to say, "It won't happen to me."

Therefore, the marketing challenge for health insurance is not simply explaining the coverage; it is getting people to acknowledge their own future vulnerability.

Why Do People Still Delay the Protection They Need?

When faced with a product whose value is invisible and whose benefit is uncertain, people's initial reaction is often to procrastinate. This is where consumer psychology comes into play. Let's consider a mandatory product.

People are required to renew certain insurance products—such as compulsory traffic insurance—every year, even though they have no desire to use them. What is interesting is this: a new policy can be issued 14 days before the old one expires, and there is no downside to early renewal—the price remains the same, and the new policy only kicks in the day the old one ends. In other words, renewing early is entirely cost-free. Despite this, consumers often wait until the very last day.

Why? Because an insurance payment falls into a category of expenditure conceptualized in consumer behavior as grudge spending [2]: an outlay for security that is deemed necessary but yields no pleasure to the consumer and competes with other, more valued expenses.

You are buying something, yet you hope never to use it. You pay today, but the moment its benefit might materialize is uncertain. Paying for insurance sometimes fails to give the consumer the feeling of having "bought something." Instead, it creates the sensation of having "paid for something I hope I never need."

From the perspective of behavioral economics, mechanisms such as "present bias" and the tendency to undervalue uncertain future benefits come into play here. However, the issue is not merely economic; it also has cultural and emotional dimensions. That is why price is not always the problem. Sometimes, the issue is the invisibility of the value received in exchange for the money spent.

So, Should We Abandon Fear Completely?

No. Insurance is built on risk, so risk communication remains essential. The problem is not fear itself, but fear becoming the ultimate goal of communication. The real question is what the consumer should feel in the face of uncertainty.

The Future of Insurance Communication

A small but significant shift is needed in insurance marketing.

From fear to trust. Instead of "Something bad might happen to you" the message becomes: "We help you stay prepared when uncertainty strikes."

From product to relationship. Instead of "Buy this policy" it becomes: "We will be by your side when you need us."

From communicating risk to communicating value. Instead of "You could lose everything" it becomes: "We help you maintain your life in the face of uncertainty."

This is not merely a matter of using more positive advertising language. It is a more fundamental question of branding: why should the consumer trust you?

In the coming period, insurance companies will not differentiate themselves solely through better pricing, broader coverage, or advanced technology. Products are becoming increasingly alike; technology can be replicated, features can be copied, and prices can be compared.

A relationship based on trust, however, cannot be replicated in the same way. Trust is not a feature proclaimed in advertisements; it is an expectation built over time through the relationship the consumer establishes with the company.

Therefore, in insurance, brand value is defined by the gap—or alignment—between the promise made before a loss occurs and the promise kept when that loss actually happens. It is within this space that a brand is either won or lost.

We do not need to constantly remind consumers that they need insurance. The real challenge is convincing them which brand will stand by them on the day that need arises.

After all, insurance does not really sell the future. In a world where the future is uncertain, it sells the possibility of facing it with confidence.

References

1. Karayolları Motorlu Araçlar Zorunlu Mali Sorumluluk Sigortası Teminat Limitleri. (n.d.). https://www.tsb.org.tr/tr/karayollari-motorlu-araclar-zorunlu-mali-sorumluluk-sigortasi-teminat-limitleri

2. Loader, I., Goold, B., & Thumala, A. (2015). Grudge Spending: The Interplay between Markets and Culture in the Purchase of Security. The Sociological Review, 63(4), 858–875. https://doi.org/10.1111/1467-954X.12329

Don't Speed Up Work; Eliminate It

Insurance broker software should be evaluated not by features but by how many redundant data transfers it eliminates across systems.

Insurance

Ask a commercial lines producer where the day went, and the answer is rarely one system. Time gets lost entering client details into a rating portal that already has the same information in the management system. Then there is the quote downloaded as a PDF, the bound policy keyed back into the agency platform, the certificate request sent to service staff who enter the same account details again, and the commission statement that has to be reconciled against records that do not match automatically.

None of these steps happens within a single application. They happen between applications.

That is where much of the waste sits. Yet insurance broker software is often evaluated based on what it can do within its own screens. A faster screen does little for a producer who still has to move among four systems to place one policy.

The better question is not what a platform does. It is what work it removes.

Count Touches Per Policy Before Anything Else

A touch is any point where someone moves, retypes, reformats, or reconciles data that already exists somewhere in the business. It is not a decision that requires judgment. It is a transfer of information from one place to another.

The easiest way to see how much of this work exists is to follow one recent commercial account through its full lifecycle. Mark every time information has to be transferred.

  • Client information is entered into the management system at intake and then entered again into each carrier portal during quoting.
  • Quote details from a carrier response are transcribed back into the management system for comparison.
  • Bound policy data is keyed into the system of record from a declarations page.
  • Certificate requests received by email are entered again into a certificate tool.
  • Endorsement details are entered into both the carrier portal and the internal record.
  • Commission statement lines are matched manually against expected amounts.
  • Renewal exposure data is collected again from the client because last year's answers are sitting in an attachment that nobody can query.

Brokerages that perform this exercise typically find between 15 and 30 touches on a mid-market commercial account. The count can be surprising. Personal lines generally have fewer touches per policy, but the volume across the book can make the overall burden significant.

The total count matters, but where those touches occur matters more. Most of them tend to collect around three or four transitions. Those transitions are where the biggest opportunities for improvement usually sit.

The broader industry picture makes this worth addressing. Analyst research on insurer technology points to rising IT spending as a share of premium, while data problems are emerging as a bigger constraint on returns than tooling gaps. If a brokerage keeps adding systems while leaving the handoffs between them manual, the expected return from that technology will be harder to achieve.

That is why the touch count should come before the software evaluation. A baseline turns a general discussion about efficiency into something measurable. It also gives the vendor a clear problem to solve.

Where Insurance Broker Software Should Remove Work Entirely

Four areas account for many of the touches that can be eliminated.

Intake to quote. Client and risk information captured once should flow into every carrier submission without being entered again. Where carriers provide application programming interfaces (APIs), that transfer can happen directly. Where they do not, structured pre-fill and form automation can still remove much of the manual keying. A broker entering the same address seven times for seven markets is doing work that technology could have removed long ago.

Quote to bind. Carrier responses should come back into the system of record as structured data rather than documents that someone has to read and transcribe. That allows comparisons to be made using data already held by the system. Once a policy is bound, the existing information can carry forward instead of forcing someone to start again with a declarations page.

Bind to service. Certificates, endorsements, and policy changes should use information already stored in the record instead of relying on a service request email. Certificate generation is a straightforward example. A standard request can use a template and current policy data to produce the document without requiring someone to transfer the information manually.

Cash to reconciliation. Commission statements often arrive in different formats, which makes reconciliation one of the more persistent manual tasks in brokerage operations. Automated matching against expected commissions can handle the routine work and send exceptions to a person for review. That turns a full reconciliation exercise into an exception queue.

Margin pressure makes all four areas more important. Deloitte's outlook expects the combined ratio to worsen through 2026, putting more pressure on the cost of servicing each policy. Work that was easier to absorb in a soft market becomes much more noticeable when carriers tighten, and organic growth slows.

There is an important distinction here. These four areas are boundaries between processes, not individual software functions. Yet software for insurance brokers is usually sold by function. That can leave brokerages with capable modules that still depend on people to connect them.

Elimination and Acceleration Are Different Purchases

Not every productivity improvement removes work. Some simply make existing work faster.

Better search, quicker screens, and mobile access can all help someone complete a task more efficiently. But the underlying task is still there.

That creates a limit. If a broker transcribes information 30% faster, the transcription still happens. The possibility of an error remains, and the reconciliation that follows remains. Those errors may show up later as an incorrect certificate, a coverage gap, or a commission dispute.

Elimination is different. When a data transfer disappears, the transfer error disappears with it. The reconciliation created by that transfer can disappear as well.

This is a useful distinction to make during software demonstrations. When a vendor shows a faster way to complete a task, ask why the task exists in the first place. Sometimes there is a good reason. A person may genuinely need to make a decision. In other cases, the task exists simply because information is not moving between systems.

The Integration Question Behind Every Demo

If the value of insurance broker software lies in removing work at system boundaries, integration capability becomes critical. What matters is not whether a vendor says its platform integrates. What matters is what those integrations do for your brokerage.

Start with five questions.

  1. Ask which carriers are connected today, by name. Do not settle for a number or a list of planned integrations. Look at your top markets and find out which ones actually connect today. That tells you how much intake-to-quote work the platform can address.
  2. Find out what data moves in each direction. Submission data may flow out, but can quote information come back? Does policy data return to the system? Can endorsements be sent out? Does claim status come back? A connection that handles only one part of the process should not be treated as a complete integration.
  3. Confirm download coverage and what it includes. Policy download can handle a meaningful portion of personal lines, but commercial lines coverage is less complete. The gaps are where manual entry continues.
  4. Test commission reconciliation with real statements. Take three carrier statements in their actual formats and ask the vendor to demonstrate how matching works. More importantly, ask to see what happens when the system cannot match something automatically.
  5. Check whether you can access your data cleanly. Direct query access to your own book determines whether your team can answer its own questions or has to submit requests whenever it needs specific information.

Vague integration claims are one of the easiest ways for a software evaluation to go wrong. A brokerage can avoid that problem by checking its own top carriers before signing a contract. Otherwise, it may discover too late that the carriers the vendor has connected are not the carriers the brokerage needs.

What Insurance Brokerage Software Cannot Fix

Knowing what software cannot solve is just as important as knowing what it can.

Carrier connectivity will always depend on what carriers make available. If a market provides only a portal, no platform can create a direct data connection where none exists. Some manual work will remain, and the amount will depend on how much of the book sits with carriers that have limited connectivity.

Regulatory and licensing requirements create another fixed layer of work. Surplus lines filings, state-specific disclosures, and producer licensing checks involve genuine compliance responsibilities. Insurance broking software should make the evidence collection and filing mechanics easier, but it should not claim to replace a licensed professional's judgment. The system can support the process. The compliance decision still belongs with the person who holds the license.

Data quality is another constraint. Duplicate accounts, inconsistent names, and incomplete records can break automated matching. A new platform does not automatically fix poor data. It inherits the condition of the information that gets migrated into it. Data cleanup therefore needs to be part of the implementation plan from the start.

Process variation inside the brokerage can also limit what automation can achieve. If three teams service accounts in three different ways, the brokerage has a choice. It can standardize the process or configure the software to accommodate each variation. The second option usually comes with more configuration and maintenance. That is an operating decision, not a software problem.

Broker insurance software also cannot fix a compensation model that rewards activity instead of outcomes. If producers are measured by the number of touches they make rather than placed premium and retention, reducing those touches can create resistance. No platform can resolve that incentive problem on its own.

Building the Business Case on Touches

The business case becomes much stronger when it is based on actual touch counts rather than broad claims about efficiency.

Start by multiplying the number of eliminated touches per policy by the number of policies handled each year. That gives you an estimate of the hours returned to the business. Value those hours based on who is doing the work. Producer time should be valued against its revenue opportunity cost, while service staff time should use loaded salary costs.

Then account for the errors that disappear when the transfers disappear. These can include certificate corrections, endorsement rework, and commission disputes.

The error cost is often overlooked. A brokerage issuing several thousand certificates a year can spend substantial time correcting even a relatively small number of errors. Those corrections also affect client relationships, yet that cost rarely appears on a financial statement.

Set a specific target instead of using a broad goal such as "improved efficiency." For example, reducing an account from 24 touches to nine gives the vendor a clear target and gives the brokerage something it can measure after implementation.

Keep measuring after go-live. Touch counts can creep back up when teams create workarounds for gaps in the new process. A quarterly review can identify that drift while it is still manageable.

Implementation sequencing matters too. Trying to automate every boundary at once can stretch the project and delay the first visible results. It is often better to start with the largest cluster of manual work, whether that is intake to quote or commission reconciliation.

The priority should be the size of the opportunity, not how easy an integration is to configure. The simplest integration is not necessarily the one that removes the most work.

Start With the Count

Insurance broker software is often bought based on features, but its value is experienced through the handoffs between systems. Brokerages that get the most from these platforms usually begin the evaluation with a clear understanding of where their teams spend time.

They know because they counted.

Insurance brokerage software earns its cost by removing unnecessary transfers between quoting, binding, servicing, and reconciliation. That improvement can be measured before implementation, during the rollout, and after go-live.

Take one commercial account through its entire lifecycle and mark every point where someone moves data that already exists somewhere else. That count gives you a practical starting point. It also gives you a much better way to evaluate the next vendor demonstration.

Don't Ask Employees to Support Change...

...ask them to help build it. Transformation in insurance requires involving employees early to identify real problems before selecting technology solutions.

Insurance

The insurance industry is changing faster than ever. Artificial intelligence, automation, new CRM platforms, digital claims and advanced analytics are attracting significant investment as insurers seek to become faster, more efficient and easier to do business with.

But even good technology does not guarantee successful transformation.

Sometimes the problem begins before a company selects a technology or writes a single line of code.

It begins when we try to change the way people work without involving those people in defining what actually needs to change.

The Timing Problem: Bringing Employees in Too Late

A typical transformation project follows a familiar pattern.

Management identifies a problem. Leaders review KPIs and discuss possible solutions. A project is launched. IT teams or external vendors design and build a new process or system.

Only then are employees brought into the process.

They are shown the new solution, trained to use it, and told how their work will change.

And when resistance appears, we often reach a familiar conclusion: people don't like change.

But perhaps resistance to change is not always the real problem.

Perhaps we simply invited people into the process after all the important decisions had already been made.

If employees first hear about a transformation when they are invited to training on a new system, there is a good chance we have involved them too late.

Beyond Users: Employees as Problem Solvers

Insurance executives see the business through metrics: premium volumes, loss ratios, costs, service levels, productivity, conversion, and customer retention.

We need those metrics.

But employees who work every day with customers, policies, claims, applications, and internal systems see another part of the business.

They know where customers have to call twice. They know where an operator manually transfers information from one system to another. They know where a manager spends 10 minutes completing a task that should take 10 seconds. They know which procedures create work without creating value.

A dashboard can tell management that a process has a problem.

The people working inside that process can often explain why.

That is why early employee involvement should not be treated simply as a matter of corporate culture or employee engagement.

It is also a way to improve the quality of management decisions.

Problem First, Technology Second

We recently looked at the workload within one of our service processes.

From a management perspective, the problem seemed straightforward: a high volume of requests, significant pressure on operators, and a large number of repetitive tasks.

We could have started by looking for a technology to solve it.

Instead, we first asked the people who worked with the process every day: What takes most of your time? Which requests are repetitive? What could be removed or automated? Which situations genuinely require human involvement?

Their answers helped us define the problem more accurately.

Only then did technology become part of the solution.

Today, AI is able to handle around 30% of requests within our medical assistance contact center. In another part of the process, AI supports initial assessment, helping us significantly reduce the volume of routine calls and allowing employees to focus more of their time on cases where human judgment and communication really matter.

But the most important lesson for me is not that AI can replace a certain amount of manual work.

It is this:

Technology can accelerate change. But it cannot always tell you which change your organization actually needs. Very often, your employees can.

Finding the Balance Between Input and Authority

There is, of course, another extreme.

Involving employees in transformation does not mean every management decision should be made collectively.

Leadership still needs to set strategy, establish priorities, allocate budgets, and take responsibility for results.

Not every employee suggestion will be right. People naturally see a process from the perspective of their own role, while management has to consider the entire organization.

But there is a large space between "employees decide everything" and "employees are informed once the solution is ready."

In my experience, this is where the most effective transformation happens.

Leadership sets the direction.

Employees help define the problem.

Technology helps solve it.

And data tells us whether the solution actually worked.

The Ownership Effect

There is another important benefit of early involvement.

When a new process simply arrives from above, employees naturally see it as something external: another system, another procedure, or another management requirement.

The situation is different when people have helped identify the problem, tested potential solutions, provided feedback, and seen that feedback influence the final result.

Not because management has somehow persuaded them to "like transformation."

But because they have some ownership of the change.

This becomes particularly important in large organizations, where even a good strategic decision can lose much of its impact somewhere between the boardroom and the hundreds of employees expected to implement it.

Three Questions to Begin Every Transformation

Before launching the next major CRM, AI, automation or customer experience project, leaders could start by asking the people who work with the process every day three simple questions:

What takes up most of your time today?

What in this process creates no real value for either the company or the customer?

If you could remove or change one thing tomorrow, what would it be?

Their answers will not replace strategy, business analysis or data.

But they may change the problem the company decides to solve.

And that matters, because one of the most expensive mistakes in transformation is not the poor execution of a good solution.

It is the flawless execution of a solution to the wrong problem.

The True Competitive Advantage

Technology is increasingly available to almost every insurer. AI models, cloud platforms, CRM systems, and automation tools are becoming less likely to provide a sustainable competitive advantage on their own.

But an organization's ability to listen to its own people, identify the real problems quickly, and turn that knowledge into action can still be one.

So perhaps the next transformation should not begin with the question:

"What technology should we implement?"

It should begin with a simpler one:

"What do our people already know about this problem that we don't?"


Mykhailo Hrabovskyi

Profile picture for user MykhailoHrabovskyi

Mykhailo Hrabovskyi

Mykhailo Hrabovskyi is a regional director with 17 years of experience in insurance, specializing in business development, innovation, and organizational leadership across Ukraine.

Insurance Claims Data Needs Better Context

The insurance industry's data problem isn't scarcity — it's the inability to confidently compare claims benchmarks across different sources and populations.

Insurance

Claims organizations have never had more data available to them.

Carriers collect increasingly detailed information throughout the claim lifecycle. Regulators publish market-conduct and financial information. Catastrophe events generate loss estimates, claim counts, and payment updates. Public agencies release datasets. Industry studies examine everything from severity and litigation to cycle time and customer experience.

Yet a claims professional can still run into trouble very quickly trying to answer a seemingly straightforward question:

What is a reasonable benchmark for this claim metric?

The problem is not necessarily finding a number. The problem is determining what that number actually represents — and whether it is comparable to the claims being evaluated.

That distinction matters as claims organizations become more data-driven. A benchmark can influence vendor evaluations, operational targets, staffing decisions, executive reporting, and perceptions of performance. If the reference point itself is poorly understood, greater analytical sophistication does not solve the problem. It can simply give an unreliable comparison more authority.

The next step in claims analytics therefore should not be only about obtaining more data. It should also be about creating better context around the data we already have.

The Label Can Hide the Difference

Consider something as basic as "cycle time."

At first glance, it sounds like a standardized measure. But what starts the clock?

First notice of loss? Assignment to an adjuster? Receipt of documentation?

And what stops it?

First payment? Final payment? Administrative closure? Closure of a claim that may later reopen?

Two sources can report a cycle-time figure using the same label while measuring different intervals. Neither figure necessarily has to be wrong. They may simply answer different questions.

The same issue appears with severity. An average may be based on paid claims, incurred loss, closed claims, claims with payment, or all reported claims. A catastrophe-heavy population can behave very differently from ordinary experience. A book containing complex commercial property claims should not automatically be evaluated against a reference dominated by high-volume personal lines.

Loss ratios create their own challenges depending on the definition, expense treatment, valuation period, and underlying population.

The metric name is therefore only the beginning of the analysis.

Before asking whether a result is above or below a benchmark, claims professionals should ask a more fundamental question:

Are we measuring the same thing?

Public Data Is Valuable — but Fragmented

There is substantial useful public insurance information in the United States, but it was not all created for the same purpose.

State insurance regulators, the National Association of Insurance Commissioners, federal programs, public insurance entities, catastrophe reporting, and other sources can each provide valuable pieces of the claims picture.

The NAIC's Market Conduct Annual Statement illustrates the importance of standardization. MCAS was created to provide regulators with a more uniform system of collecting market information, and participating jurisdictions use published definitions and ratios intended to support more meaningful comparisons. That structure matters because raw numbers become much more useful when the population and definitions surrounding them are known.

But the availability of a standardized dataset in one area does not mean an equivalent national benchmark exists for every line of business or every claims metric.

Specialty claims make the problem particularly visible.

Finding a defensible national reference for homeowners or private passenger auto may be fundamentally different from finding one for earthquake, builders risk, inland marine, or other specialized exposures. Some lines have substantial public information. Others may offer only state-level experience, event-specific reports, insurer disclosures or indirect evidence.

The temptation in those situations is to fill the gap with a generalized "industry average."

That can be worse than acknowledging that the national benchmark does not exist.

Broader Does Not Always Mean Better

Claims professionals understandably gravitate toward national averages. "National" sounds comprehensive and therefore authoritative.

But geographic breadth is only one dimension of benchmark quality.

Imagine there is no reliable national cycle-time benchmark for a specialized line, but a state insurance entity publishes well-defined claims information for a clearly identified population.

That narrower reference may actually be more defensible than a national-looking number whose methodology, population, or origin cannot be established.

California earthquake insurance provides a useful illustration of the type of source that may exist in a specialized market. The California Earthquake Authority publishes annual reports and, following qualifying seismic events, event-specific reports on its program operations and claims experience. Those figures should not simply be relabeled as national earthquake benchmarks. They represent a particular geography, insurance structure and population.

But when those boundaries are clearly disclosed, they can still provide meaningful reference information.

The correct response to imperfect coverage is not to pretend a narrow reference is universal.

It is to label the reference correctly.

State-level data should look like state-level data. Event-specific experience should look event-specific. An estimate should look like an estimate.

Transparency about scope makes information more useful, not less.

False Precision Is a Claims-Management Risk

Claims organizations are accustomed to working with uncertainty. Reserving itself requires professionals to evaluate incomplete information and revise conclusions as facts develop.

External benchmarking deserves similar discipline.

A number shown as 47 days can feel far more authoritative than a range of approximately 30 to 60 days, even when the underlying evidence supports the range more strongly than the single point estimate.

The appearance of precision should not be confused with the quality of evidence.

This is particularly important when external benchmarks reach executive dashboards or vendor scorecards. Once a benchmark becomes a red, yellow, or green threshold, the assumptions behind it can disappear.

A TPA may appear to underperform an external reference even though its claims involve different limits, jurisdictions, complexity, catastrophe exposure, or closure definitions. Conversely, a favorable comparison may conceal weaknesses if the external population is substantially harder to resolve.

Benchmarking is useful because it gives performance context.

Benchmarking becomes dangerous when the context is removed from the benchmark.

A Better Reference Layer

The insurance industry has invested heavily in systems designed to collect, organize, and analyze claims information. The next opportunity is to become equally disciplined about the reference information against which that data is interpreted.

A useful claims reference should answer more than "What is the number?"

It should answer:

Definition: What exactly is being measured?

Population: Which claims are included and excluded?

Geography: Is the reference national, multi-state, state-specific, or event-specific?

Period: When was the underlying experience measured?

Source: Can the original source be identified and reviewed?

Method: Is the result directly observed, calculated from source data, estimated, or generalized from industry experience?

Comparability: How similar is the reference population to the claims being evaluated?

Limitations: What conclusions should not be drawn from the number?

Keeping these questions attached to a benchmark creates something more valuable than a database of statistics. It creates a reference layer.

That reference layer becomes increasingly important as automation and artificial intelligence move deeper into claims operations. Analytical tools can process enormous volumes of information, identify patterns and generate comparisons at unprecedented speed.

But analytical speed does not make two unlike populations comparable.

Better technology increases the importance of knowing exactly what sits behind the data.

Sometimes the Best Answer Is That There Is No Benchmark

There is another habit claims organizations should become more comfortable with: saying that a reliable external benchmark has not been identified.

That is not a research failure.

It is information.

The absence of a defensible benchmark tells decision-makers something about the maturity of the available data and the level of confidence that should accompany any external comparison.

An organization can still use internal trends, peer information, state references, selected studies, or an explicitly identified industry-informed range. But those alternatives should not be presented as something they are not.

A transparent limitation is preferable to a fabricated certainty.

From More Data to Better Context

Claims data will continue to expand. More granular information, improved reporting, connected systems, and increasingly sophisticated analytics will give insurers capabilities that were difficult to imagine only a few years ago.

But volume alone does not create comparability.

The more data claims organizations consume, the more important definitions, source lineage, scope and methodology become.

The central question should no longer be simply:

"What is the industry benchmark?"

It should be:

"What is the best available reference for this decision, and what does the evidence actually allow us to conclude?"

That change sounds small, but it produces a very different approach to claims benchmarking.

The goal is not to eliminate uncertainty or force every claim population into a universal standard.

It is to make the uncertainty visible, preserve the context behind the number, and use external data at the level of confidence it actually deserves.

Claims do not need another number detached from its source.

It needs better reference information around the numbers it already has.


Jacob Kalman

Profile picture for user JacobKalman

Jacob Kalman

Jacob Kalman is a senior TPA oversight specialist at Palomar.

He holds the CPCU designation and has experience spanning E&S property, builder’s risk, catastrophe, wind, flood, earthquake, inland marine, and motor truck cargo.

Making Innovation Stick in Insurance Organizations

Sustainable innovation in insurance requires organizations to master change adoption through culture building and strategic training design.

Insurance

Insurance organizations are getting very good at introducing things. Systems, workflows, automation, AI capabilities. New ways of capturing, analyzing, and moving data. But introducing them is one thing; the harder part is getting those changes to stick.

Technology can work exactly as designed and still fail to become part of the way an organization actually operates. Employees find workarounds. Producers return to the process they already know. Managers use the new system differently from one team to the next. Six months later, everyone is technically using the same tool, but not necessarily in the same way or for the same reason.

Innovation means change with effect. Making innovation stick has far less to do with what the change is than who is taking on the change. Success in innovation hinges on the ability of the organization as an entity to make changes. Thankfully, that ability is not formed on luck, but is a literal set of skills that individuals and teams can master.

The Mindset

The concept of growth mindset isn't a new one. There's endless literature for individuals about cultivating a growth mindset and using it to achieve personal goals. A group has a mindset, too – it's just called "culture," and it exists whether it was intentionally formed or not.

The first applicable step in culture building to prepare for innovation is to unwind anti-growth mantras. Leaders, whether it be CEOs, department heads, sales managers, training leaders, or anyone in a position to influence a team, need to scan their speech for ways they might be blocking their own goals by verbally placing limitations on how their team can learn. People do it all the time and don't even notice, but the growth of a business can't outpace the growth of its people for long, no matter what tier of people you start with.

There must be constant and consistent effort applied to the task of building belief, real belief that work and change is desirable, possible, and worth it.

The Dance

Getting a whole organization, particularly a large one, to move together with agility is a bit like ballroom dancing. The best dancers aren't just the experts in a set of pre-choreographed movements, but trained in a system of perceiving and responding to signals that communicate a goal or status. A pair of dancers is called a "Lead" and a "Follow" – the Lead sends signals for objectives, and watches the Follow for signals of status.

An organization is a little different from a ballroom dance pair. Most obviously, there are more people in the mix. More people to think, more people to act, more places where single variables can spin energy into unknown directions. This fact about organizations cannot be changed with any amount of tighter regulation or greater discipline, but it does have its upsides (more on that later). While movement through change is a little more complex than a single Lead and a single Follow, in the most effective organizations the direction and type of signals sent are the same.

Lead From the Top

Successful adoption starts well before the implementation. People need to understand what the organization is trying to accomplish and how the change will support that goal. A new system can't be dropped in with the assumption that everyone will automatically see the connection.

The role of the CEO or leader is to give people something to align around. A clear vision in its absolute simplest, most condensed form is the most desirable output. The vision should not be overly detailed and prescriptive at the operational level (as hard as that might be to resist), but it should be unquestionably clear. It's best stated as a simple goal with context, and perhaps a corresponding anti-goal. Because of [context], we will [goal]. We must avoid [anti-goal] because of [additional context]. This is the "signal for objective."

Clear leadership has to create the reference point. If the organization can define what success looks like, teams can make better decisions about how a new process fits into the bigger picture. This is the aforementioned "upside" to the chaos of having so many moving parts in an organization; as the signal for objective moves down the chain of command closer to the actual work, the goal can remain as simple and as clear as ever, but the context increases through every descending layer until it is quite apparent what action needs to be taken.

Once the action needed is clear, then the organization is ready for invention, implementation, and training.

Teach to the Top

Much can be said about ways to handle the invention and implementation pieces of innovation. Training deserves a little extra love in this article, as it's the matching piece to leadership and is critical to adoption whenever a human touches a product or process.

What does the best training look like? Is it adequate to simply get everyone together on a webinar once or twice to go through the new workflow? There's a school of thought, a movement even, within the education profession that recommends this method for bringing a whole group of diverse students to their best performance: Teach to the top, and scaffold down.

The truth is that within our organizations, there's a diverse spread of performance ability. There is strong temptation to design training to target the least able so that no one is left behind, but the outcome of doing that is usually the inverse of what is desired. If the 20% of high performers that create 80% of the productive output (pareto distribution) are supported in using the new innovative process or product to 100% of its capacity, the majority of the change that was needed has already happened.

When "teach to the top" is applied properly, high performers soar, but the performance floor also rises because of the high target and scaffolding provided. What does that scaffolding look like? A few points to ponder:

  • Direct, rigorous instruction
  • Acceleration through depth, not speed
  • Optimal concept sequencing
  • Strongly positive relationships
  • Excellence exhibitions

Teaching to the top doesn't mean skipping half the content because it's presumed to be understood. Teaching to the top means digging in a little deeper at every level.

Humans at the Center

Culture building to support agility and change-aptitude can't happen just once. It has to continue as the organization learns. Successful innovation happens when products and trainings are designed around the idea that expanding human minds leads to business growth.

The technology, process, and data all matter. But in any insurance process that still has a human in it, the human matters too. That is often the difference between introducing something new and creating a change that actually sticks.


Abrianne Harmon

Profile picture for user AbrianneHarmon

Abrianne Harmon

Abrianne Harmon is the director of innovation at ALKEME Insurance, a full-service insurance agency providing commercial and personal insurance, employee and executive benefits, retirement, and wealth management services.