The Five People and the One Engine

Part 1 argued that insurers price everybody else's risk and have never priced their own, and that two unmeasured limits, a cost floor and a fixed product mould, now decide how much of the market they can reach. This part is about the five people who have to agree before either limit moves, and the single operating model all five can say yes to. It closes with one page any carrier can lay over its own operation before speaking to anyone.

Seventy per cent

You have heard the number: seventy per cent of transformations fail. It turns up in steering committees, in board papers, and in the careful email explaining why the pilot will not be extended this year.

It has no evidence behind it. The figure comes from a 1993 book on business re-engineering, where the authors offered what they called their "unscientific estimate" that 50 to 70 per cent of re-engineering efforts fell short of the dramatic results intended. Two years later one of them, writing with a co-author, called it a descriptive observation that had been distorted into a rule. In 2011 a peer-reviewed study traced five published versions of the claim back to their sources and found no valid, reliable empirical basis for any of them.¹

Thirty years and thousands of citations later, an unmeasured number is still being repeated as a reason not to act. I cannot tell you how many decisions it has changed, and I will not pretend to. I can tell you I have heard it quoted in the room, more than once, by people with no idea where it came from. An industry that has never measured its own cost of change is being steered by a statistic nobody measured either.

Part 1 ended on three layers (structure, legacy, and the time it takes people to decide) and promised that the third would be the subject of this piece. So, into the room where those decisions are made.

01The five people in the room

Every meaningful decision about how an insurer works passes at least five chairs. There are usually more, a group function, a joint-venture partner, a distribution head with a view, but these five recur, and the same five sit inside every administrator, claims operator and reinsurer in the chain.

I have sat across from all of them. None is a blocker. Each protects something that needs protecting, and each answer is correct when judged by what that person is measured on. The right-hand side of each card is a specification, and the orange line beneath it describes what meets it. An interest to declare: Gangkhar builds an operating layer of this kind, which is why what follows is written as configurations and use cases rather than as a product pitch. If the mechanism is wrong, the vendor does not matter.

02Why five right answers make a wrong one

Put those five in a room and something quietly illogical happens. Each protects their own exposure, five reasonable cautions add up to a veto, and the outcome nobody chose becomes the outcome everybody gets. This is not a failure of character. It is a well-documented feature of how people decide.

The pull of the default

In 1988 two economists showed that people stick disproportionately with whatever is presented as the existing arrangement, and that the advantage of the status quo grows as the number of alternatives rises.⁴ The more options a committee is shown, the likelier it is to choose none of them.

Insurance has seen this at scale. When New Jersey and Pennsylvania offered drivers a choice of motor cover in the early 1990s, each with a different default, only about one New Jersey driver in five and about one Pennsylvania driver in four moved off the default they were given, despite real differences in price and cover.⁴ The industry has relied on that behaviour ever since; it is part of why renewal books are as sticky as they are. We have been earning from our customers' status quo bias for thirty years while being governed by it ourselves.

Everyone thinks the room agrees with them

BCG's research into how change works, drawing on 6,000 executives and employees across fifteen countries, identifies false consensus as one of the behavioural traps behind stalled change: leaders assume their own view is widely shared. In that research 68% of executives, against 45% of employees, said they felt positive about change in their organisation.⁵

Read that beside the five chairs. Everyone leaves the meeting believing the others are broadly on board, and each waits for someone else to move. Nothing is refused and nothing happens, which is the most expensive outcome available and the only one that never appears in the management accounts.

So the useful question is what single piece of evidence would answer all five at once, and what would have to exist for that evidence to be worth having.

03The engine, in configurations

One operating model answers all five chairs, which is why it can get through the room at all. Every "what changes the answer" above is a property of the same engine: beside the core, one workflow at a time, governed and reversible.

It does not involve replacing the core. Core replacement is where the first chair earned their scars, and lowering the floor no longer requires touching it. Three layers of capability now sit above whatever a carrier already runs. Part 1 promised an engine built in modules, where a new product shape is assembled rather than rebuilt, so that customisation stops raising the cost of a policy. The clearest way to describe it is by what can be configured in it.

At Gangkhar these three layers are Sherpa+ (product configuration and rules), Sherpa+Works (the governed agent fleet) and Sherpa+Lens (continuous optimisation). The mechanism matters more than the names, so the rest of this section stays with the mechanism.

What is actually configurable

"Configuration" has been used loosely enough in this industry to mean very little. Concretely, these are the nine surfaces that stop being code and become parameters, each editable, versioned, and checked against the filed product before it goes live. Unlike the use cases further down, this list is meant to be complete: between them the nine cover the life of a policy from definition to reporting. Any step in your operation that falls outside all nine is exactly where a launch will stop.

A product is the combination of the nine, not a block of code. Change one surface and nothing else has to be rebuilt, which is the whole of the modularity claim, stated so it can be tested.

Six use cases, and the Part 1 problem each answers

Configuration matters only for what it lets an insurer sell and run that it could not before. These six were chosen because each closes a gap named in Part 1. The better ones usually come from inside a book someone already knows.

None of this requires the policy administration system to change. It needs a layer that reads from the core, writes back to it and runs the work around it, which is why a first workflow can go live in weeks or a few quarters rather than years (about six weeks of technical build for an embedded programme, longer for an enterprise back-office workflow), and why it can be stopped without a migration.

That last property is the one that makes "yes" available to someone who has said "not yet" for a decade. It is also what Part 1 asked for: a floor that can be lowered and a mould that can be broken in the same programme, because the engine was built in modules.

04The proof, and the arithmetic

None of what follows is a forecast. Every figure was published by the organisation that produced it, and where a number is company-reported rather than independently audited, I say so.

The floor moving, in public

The clearest evidence sits in a listed company's shareholder letter. Between the third quarter of 2022 and the end of 2025, Lemonade, a digital insurer, added 1.2 million customers while its headcount fell 6%. Its customer count grew from about 1.78 million to 2.98 million over that period, in-force premium reached US$1.24 billion, up 31% year on year, and premium per customer rose 7%.⁶

Read together, that is what Part 1 said rarely happens in insurance: far more customers, served by fewer people. Cost per customer did not fall because somebody negotiated harder with suppliers. It fell because the work changed shape.

Source: Lemonade Inc., Q4 2025 shareholder letter. Company-reported.

Speed, at scale, in this region

Ping An Life reports that 93% of its policies are underwritten within seconds, and that its quick-claims service closes an average life and health claim in 7.4 minutes.⁷ These are company-reported figures from a group serving more than 250 million retail customers.

Set that against the benchmark from Part 1: an average of 40.7 days to final payment on US homeowners' property claims in J.D. Power's 2026 study, and that after a year of improvement.⁹ Different lines and different markets, so the comparison is not like for like. It still shows how far apart the two operating models are.

And it is not only the digital natives

Zurich has reported cutting underwriting triage time from around 75 minutes to around 15 in an agentic AI rollout across five countries. At Sabadell Seguros, its Spanish joint venture, a voice-analytics tool is credited with lifting customer retention by close to 20%.⁸ Both figures are company-reported, and they show a direction of travel rather than a guarantee. The retention figure matters most here, because it is growth: the same book, kept.

The honest position is this. The evidence that the floor can move is public. What nobody can yet say is how far it moves in this market, for this product mix, under this regulator, because the starting point has never been measured, by us or by anyone else. That is what the next page is for.

05The Reach Canvas

Everything above remains an argument until someone lays it over their own operation. Here is the instrument in full, with nothing held back for a meeting.

One page, one line of business, one working session. Read it left to right along the top: the floor, then the mould, then where the two together put you. Below sits what is holding it, then what it costs you and what you would do about it. Fill it with all five chairs in the room. The point is less the answer than all five looking at the same page at the same time.

Six blocks, one page. Part 1 closed with three questions; this is what they become when taken seriously. Nothing on it requires a vendor, a budget or a decision.

If the canvas comes back with a low floor, a short time to market, an empty never-shipped list and a small unreached segment, the argument in these two pieces does not apply to you, and I would like to meet you for an entirely different reason. If it does not, you are holding something the industry has rarely had in writing: the shape of your own unpriced risk, on one line of business, on one page.