Why insurance transformations fail: the mindset shift leaders overlook

Most insurance transformations stall not because of technology or budget, but because they start in the wrong place entirely. While carriers pour resources into system replacements and digital roadmaps, the root cause of stalled progress sits upstream of any platform decision: mindset.

This insight comes from a recent episode of The Underwriting Intelligence Podcast, where hyperexponential Co-founder and CEO Amrit Santhirasenan spoke with Ben Gower, Partner at Elixirr, about why change programs stall and what the industry must do differently.

Key takeaways

The conversation surfaced several principles that run counter to how most carriers approach transformation today.

• Insurance transformations fail when they target actions without first addressing the mindsets behind those actions

• Insurance is a “team sport” where cross-functional silos between pricing, underwriting, data, and IT create invisible drag on every initiative

• Starting from the pricing and underwriting core, rather than operational systems, delivers faster wins and stronger momentum

• Tracking leading indicators like experiments run and lessons learned, not just binary success or failure, builds the culture transformation requires

• Command-and-control mandates around AI tool adoption rates miss the point entirely

What are insurance transformations and why do most fail?

Insurance transformations touch digital infrastructure, operating models, and ways of working across carriers, brokers, and reinsurers. The ambition is usually sound, but execution is where things unravel.

The pattern Gower described is one of fundamentally misplaced starting points. The instinct has been to change what people do: adopt new technologies, follow new processes, hit new metrics. That approach produces temporary compliance at best. As Gower put it, the industry has spent years trying to change actions directly while missing that “if you start with the mindset people have, the actions change dramatically and the results change dramatically.”

This is the core thesis: actions are downstream of mindset. Unless the beliefs and mental models of the people doing the work shift first, no amount of tooling or process redesign will stick.

Why insurance requires a different transformation approach

One of the most compelling points from the conversation is that insurance is structurally harder to transform than other financial services. Banking transformations can often optimize individual functions in relative isolation. Insurance cannot.

As Gower explained: “You can’t win in insurance unless you’re operating as a team. Because if you’ve got crazy leakage in claims, doesn’t matter what you selected up front, you’re still going to lose.”

This means transformation programs touching only one function, whether that is claims automation, distribution technology, or underwriting workbenches, will always hit a ceiling imposed by cross-functional dependencies.

The cross-functional disconnect

Gower shared a telling pattern from workshops that bring together pricing, underwriting, data, and IT teams for the first time. When asked how many times these four groups have been in a room together, the answer is almost always: never.

That single data point explains more about why transformations stall than any technology gap. Pricing, underwriting, data, and IT are the four groups whose collaboration determines whether a pricing model reflects reality, whether underwriters have the data they need, and whether IT builds solutions anyone actually uses.

The mindset-first framework for lasting change

The fundamental shift Gower and Santhirasenan discussed is deceptively simple: stop trying to change behaviors directly and start with the beliefs that drive them. In the AI era, this matters even more. Jobs are changing, and daily activities are being reshaped by tools like LLMs and Agentic AI. Without helping people through the mindset change, behavioral shifts won’t follow, and neither will better business results.

The broker CEO who saw a RAG demo

Two and a half years ago, Gower and his colleague Steve were demoing a simple RAG (retrieval-augmented generation) solution at a strategic broker conference that could summarize what a policy covered. One CEO stood up and declared the broker job was gone.

Gower’s response was direct. If your job is summarizing someone’s policy, then yes, your job is gone. But a broker’s job has always been to help clients manage risk and secure the best cover available, work that still requires human judgment and market knowledge, particularly given that most AI tools are built on backward-looking data.

From fear to opportunity

The shift this requires is from reacting to AI as a threat to engaging with it as an opening. The broker CEO who watched a RAG demo and declared his job gone is the fear response in its purest form: collapse the role into the single task the tool happened to demonstrate, and assume the rest follows. The opportunity-side mindset asks a different question. If summarization compresses, what does the broker get to spend more time on? If routine processing speeds up, what does the underwriter focus on instead?

Reaching that mindset requires environments where curiosity is rewarded, experimentation is safe, and people have genuine permission to try things that might not work. As Santhirasenan framed it, the death of command-and-control leadership is a prerequisite. Mandating that 90% of people use a specific tool a specific amount each day is completely the wrong approach. Adoption-rate targets produce compliance, not the experimentation that lets teams find where the tool actually helps and where it gets in the way.

Building from the heart of the business

Gower shared a perspective on where transformations should begin. For years, the default has been to start with the customer experience, the broker interface, or a massive admin system replacement.

His current view flips that. The heart of insurance is how companies price risks and what they believe they need to do to cover those risks. Building outward from pricing and underwriting connects the transformation to the function that most directly drives combined operating ratio (COR) and return on capital.

Connecting the four teams

When pricing teams, underwriters, data teams, and IT are brought together and asked what would make their lives better, practical solutions emerge: improvements that can be rolled out in six months, not aspirational items on a five-year roadmap.

The key insight is that these solutions already exist in the collective knowledge of the four groups. They have simply never had the forum to surface them.

Avoiding the admin system trap

The most common mistake Gower flagged is starting with operational system replacements: massive admin system programs that take ten years, consuming budget and executive attention while the pricing and underwriting core remains unchanged. Starting with the heart of the business builds momentum through tangible wins, and lets operational systems follow once the case for change is established.

Shortening time horizons: a critical success factor

Five-year transformation plans look impressive in board presentations. However they do not inspire confidence among the frontline underwriters and pricing actuaries who have watched previous plans evaporate. Gower noted that the people doing the real work are massively skeptical of long-horizon plans, and rightly so. AI’s rapid evolution makes this even more acute. The tools available six months from now will differ meaningfully from those available today.

The experimental mindset

Santhirasenan offered a sharp reframe. If you treat AI adoption as a skills accumulation problem, you will never catch up. What you can do instead is shift your mindset to be experimental, iterative, and comfortable with much shorter time horizons than insurance has traditionally used.

This means tracking leading indicators alongside lagging ones. If you only track whether something succeeded or failed, the outcome is binary and the learning is thin. If you also track what people tried, what they learned, and what adjacent opportunities emerged, you build a rich base for continuous improvement.

Where transformation programs break down

Three patterns come up repeatedly, and each is avoidable once you know what to look for.

Scaling before validation

Santhirasenan shared a principle he uses internally at hyperexponential: do not pour fuel on a fire that is not burning. When a pilot has not yet proven its value for the first team, expanding it to four teams multiplies the waste, not the value.

Ignoring expert advice

Gower recounted a story about one of Europe’s biggest ERP implementations. Elixirr pitched an alternative approach. The client rejected it, insisting on their preferred methodology. Two years and significant capital later, the client called back asking for help fixing everything that hadn’t worked.

A culture of comfort admitting failure

Many insurance organizations struggle to acknowledge when an initiative is not working. A trial gets presented to the board as a commitment, and when it fails, admitting it becomes politically untenable. Building a culture where experimentation is genuinely safe requires structural support, not just verbal encouragement.

Getting started: practical steps for transformation leaders

If your pricing, underwriting, data, and IT teams have never been in the same room, start there. Then:

• Apply rigorous focus: does this initiative move COR or return on capital? If not, stop doing it.

• Get technical teams into the room early using a forward deployed engineer approach, where technical talent works directly alongside the business. This model is exemplified by firms like Elixirr, which acquired Responsum (AI) and iOLAP (data engineering) to place engineers directly alongside client teams.

• Whether working with LLMs, exploring Agentic AI use cases, or evaluating platform partners, define what predictable and breakthrough success look like before launching, and maintain a human in the loop for accountability and trust.

These three steps do not require a new budget line or a board mandate. They require the willingness to work differently, starting with the people closest to the risk.

Build your transformation from the underwriting core

The conversation between Santhirasenan and Gower surfaces a truth the industry needs to internalize: insurance transformations are fundamentally people problems that technology can help solve, once the humans involved have shifted their mental models.

Start by bringing your pricing, underwriting, data, and IT teams together, shortening your time horizons, and measuring what people are learning rather than only tracking what succeeded or failed. For deeper insights, listen to the full conversation between Amrit Santhirasenan and Ben Gower on The Underwriting Intelligence Podcast.

The hx platform is built from the underwriting core outward: actuaries build and deploy models in the Decision Engine, underwriters act on them in Pricing & Rating, and Portfolio Intelligence gives teams the visibility to track whether decisions are improving COR and return on capital.

Explore how the hx platform supports the kind of transformation Gower and Santhirasenan describe.

Frequently asked questions

What role does AI play in insurance transformation programs today?

AI acts as an accelerant rather than a destination. Tools like RAG (retrieval-augmented generation), LLMs, and Agentic AI are reshaping daily workflows. However, three questions must be answered for any AI initiative: on what data, with what intelligence, and which human is accountable for its accuracy.

Why do consultancies sometimes give counterproductive transformation advice?

Traditional consulting’s time-and-materials billing can misalign recommendations with client outcomes. Risk-reward fee structures better align interests, though procurement teams often resist open-ended upside arrangements.

How do leading and lagging indicators differ in transformation measurement?

Most programs track only whether a project succeeded or failed. Leading indicators track what people tried, what they learned, and what adjacent opportunities emerged. This richer feedback loop avoids the binary thinking that discourages experimentation.

What is the “forward deployed engineer” model in insurance transformation?

Rather than placing technical talent behind a sales layer, this model embeds engineers directly with business teams from day one. Underwriters and pricing actuaries can evaluate whether a partner genuinely understands their systems before any contract is signed.

Why is it so hard to admit failure in insurance transformation programs?

When a trial gets presented to the board as a strategic commitment, failure becomes harder to acknowledge. The initiative stops being an experiment and starts being a promise, and admitting it is not working carries political cost. Building a culture where early failure is acceptable requires more than verbal encouragement. It means tracking what teams tried and learned alongside whether projects succeeded, and creating space at the leadership level to call something off without it reflecting on the people who ran it.