As insurance agencies continue to invest in AI, not all are seeing a return on investment. Some are seeing productivity gains right away as employees get certain work done faster. But often when leadership looks for the corresponding lift in revenue, client satisfaction, or operational transformation, the numbers aren't there yet. A recent study of new AI use cases announced by some of the world's largest insurers in the first quarter of 2026 found that only 18% reported a measurable result, and four out of five of those were productivity gains with no revenue or satisfaction figures attached.
The instinct for many organizations is to throw their hands up and label their AI investments a failure, but the delayed effect AI-driven efficiency has on measurable business outcomes is a predictable stage in a technology adoption cycle. AI is moving through a pattern that starts with broad adoption, shifts to prioritization, and eventually produces measurable results. Most agencies are somewhere between the first and second stage.
Expecting bottom-line transformation during what is still an adoption phase puts organizations at risk of making reactive decisions, such as pulling back on investments or doubling down in the wrong places. Rather than scrambling to justify AI spending or scrapping programs that haven't paid off yet, agencies should use this interim to make three deliberate moves: Focus AI on the priorities that matter most to their business, measure success with the KPIs they already trust, and keep human expertise at the center of how AI work gets reviewed and governed.
This is especially relevant given where the insurance brokerage business is in its own cycle. The industry has been acquisition-driven for the last three to five years, but the transaction pace has slowed. Firms that brought together multiple organizations are now looking inward, trying to create uniformity in process across disparate systems and teams. AI can play a large part in creating internal efficiencies, as long as organizations are deliberate in how they apply it. That standardization is what ultimately drives greater cost-effectiveness, particularly as firms look to shift toward outcome-based approaches.
The sharpest insurance brokerages right now are making those deliberate choices. Many are prioritizing front-of-office work because they want to drive greater client experience, and they know this will have a clear effect on satisfaction, NPS, and retention.
Similarly, measuring the impact of AI should go beyond speed or cost savings. Think about the shift from typewriters to personal computers. The work was done a little differently and certainly got faster, but that didn't change the standards for what qualified something as well-written. The same thing happened with the BlackBerry, and then again with the iPhone. These tools had a massive effect on how effective people were in their daily lives, but the underlying measures of success stayed the same.
We don't need to reinvent the wheel to measure AI's success. If any agency was tracking client satisfaction, retention rates, and growth before the technology, those metrics still apply. A meaningful early indicator might be straightforward: Are KPIs that the organization was previously falling short on starting to get hit? If so, the tools are working, and the agencies don't need to change how they measure success. They need to figure out what's most important to their business, aim AI at it, and measure with the instruments they already trust.
Even as agencies focus AI on the right priorities and track the right metrics, none of it holds together without human oversight. Anyone who has used an agent to summarize a meeting knows the technology doesn't catch everything. When AI agents are handling workflows that touch claims, coverage, or regulatory obligations, the margin for error shrinks considerably.
Agencies that are just bolting third-party AI onto their workflows without understanding what's under the hood are taking a big risk. When there is no ability to show what work has been done, you cede control to an AI agent. The right path is to have comprehensive quality assurance processes and regression testing capabilities around every agent doing day-to-day work, with checkpoints and compliance oversight built into the workflow.
There hasn't yet been a major AI-related claims dispute, but that day is coming. When it does, firms will need to understand exactly what happened and where. They can't just point to the black box of an agent. This is where domain expertise paired with that technology becomes the differentiator. Agencies working with partners who bring decades of insurance operations knowledge alongside the technology — and who maintain human oversight as a fundamental part of the process — can actually govern what their AI is doing. And as AI adoption grows, the demand for people with the expertise to oversee and evaluate what agents are producing will grow with it.
The effect of the decisions that agencies are making now will be clear by the first quarter of 2027. Nearly 60% of carriers and managing general agents expect AI to transform their business models within the next one to three years. That's when we'll be able to look back at the first two quarters of 2026 and see where speed and efficiency gains actually translated into revenue and operational results. That's when those lagging indicators will catch up.
