Every InsurTech, and every insurer adding capability it doesn’t yet have, hits the same fork: build it, partner for it, or acquire it. The fork used to apply mainly to technology. It doesn’t anymore. AI, distribution, underwriting, regulatory capability, data infrastructure and claims operations each get asked the same question now, separately and on their own timelines.
Most executive teams resolve this with conviction rather than analysis. The CTO wants to build, because building is the muscle they have. Corporate development wants to acquire, because capital deployment is the mandate they’re measured against. The CRO wants to partner, because partnerships close in a quarter instead of eighteen months. Everyone is answering a different question with a different set of facts and calling it the same decision.
Whoever argues hardest in that room usually wins the meeting. The market rewards whoever got the sequencing right. Build, partner or acquire is a capability design question before it is an M&A question, and what follows is a way to make it explicit. Read backward, it also explains why several recent deals took the shape they did.
Where the instinct goes wrong
Speed bias. Teams default to building because it is the option that needs nobody else’s sign-off. That works when the capability is codifiable, like a pricing model or a workflow engine. It fails slowly when the capability is a set of institutional relationships a competitor spent a decade earning.
Capital-deployment bias. Corporate development teams sitting on committed capital have an incentive to close something. That pressure quietly converts “should we acquire this?” into “how do we justify acquiring this?”, usually after the target has already been picked.
No shared criteria across functions. The CFO is scoring capital intensity. The CTO is scoring integration complexity. The Chief Underwriting Officer is scoring control. The CRO is scoring time-to-market. Four legitimate lenses and no common denominator, so the board gets consensus-by-exhaustion instead of an actual comparison.
The eight variables that actually decide it
-
Capital intensity
How much capital the path requires relative to your cost of capital and, for regulated entities, how that capital is treated. Acquisitions carry goodwill and, for insurers, a Solvency II hit; builds and partnerships usually don’t.
-
Time-to-parity
How long until the capability is functionally competitive: months for licensing, quarters for a partnership to bed in, often years for an organic build, especially in underwriting, where the model needs seasoned loss experience before anyone will trust it.
-
Regulatory dependency
How dependent the capability is on local licensing, compliance infrastructure and institutional relationships that don’t transfer across jurisdictions. High dependency means it has to be accessed or owned locally; it can’t be built remotely or licensed in from elsewhere.
-
Tacitness / key-person dependency
How much of the capability’s value lives in codified systems and IP versus in specific people and relationships. High tacitness makes acquisition risky unless retention is structurally secured rather than assumed.
-
Repricing / withdrawal risk
How much a partnership arrangement for this capability could be repriced, narrowed or withdrawn by the counterparty: channel deals and capacity agreements where someone else holds the pen.
-
Integration complexity
Tech stack compatibility, data architecture and cultural fit. The variable most often underweighted at signing and most often blamed at year two.
-
Reversibility
Whether the commitment can be staged (minority stake, earn-in, joint venture) or is binary and irreversible. Staged structures let you buy information before you buy full exposure. They can also buy price: Munich Re agreed to acquire At-Bay at $575m against a $1.35bn valuation in 2021, and took full ownership of NEXT Insurance at $2.6bn against a $4bn valuation in 2021. A seat inside the relationship is an option on the down-round.
-
Control requirement
How much operational control the initiative actually needs. Underwriting discipline and pricing usually require full control. Distribution reach usually doesn’t.
The decision matrix
Read at the high end, this is what each variable favours:
| Variable (at the high end) | Build | Partner | Acquire |
|---|---|---|---|
| Capital intensity | counts against | strongly favours | counts against |
| Time-to-parity | counts strongly against | favours | strongly favours |
| Regulatory dependency | counts strongly against | favours | strongly favours |
| Tacitness | counts against | counts against | strongly favours |
| Repricing / withdrawal risk | favours | counts strongly against | strongly favours |
| Integration complexity | strongly favours | favours | counts against |
| Reversibility | favours | strongly favours | counts against |
| Control requirement | strongly favours | counts against | strongly favours |
Acquire wins where the capability is slow to build, jurisdiction-locked, tacit or risky to rent. Partner wins on capital efficiency and optionality. Build wins when integration and control matter and the capability is realistically buildable.
Scoring it: a worked example
Take a mid-market European insurer that needs embedded bancassurance distribution in DACH. Score each variable 1 (low) to 5 (high) and read it against the matrix above:
| Variable | Score | Read |
|---|---|---|
| Capital intensity | 2 | Low. The target is a specialist platform, not a capital-heavy asset, so this doesn’t argue for Partner. |
| Time-to-parity | 5 | Building this from scratch means years of earning institutional bank trust. |
| Regulatory dependency | 5 | Deeply jurisdiction-locked. DACH banking compliance doesn’t transfer in from elsewhere. |
| Tacitness | 4 | High. The value sits in bank relationships and compliance know-how, not code. |
| Repricing / withdrawal risk | 4 | High. A bank distribution partnership here can be repriced or narrowed at renewal. |
| Integration complexity | 3 | Moderate. API-based embedding, not a core-system replacement. |
| Reversibility | 2 | Low. The buyer needs control, not an exit ramp. |
| Control requirement | 4 | High. Regulatory liability sits with the acquirer, not the partner. |
Four of the highest-scoring variables (time-to-parity, regulatory dependency, tacitness, control) land in the Acquire column above. Capital intensity and reversibility don’t argue for Partner here. Integration complexity is a wash. Control also credits Build, but the two variables scored 5 bury it. Map each score of 1 to 5 onto −2 to +2, multiply through the matrix and sum the columns: Build −6, Partner −4, Acquire +16. The weights are equal by default; a board can change them. Either way it is a tally rather than a hunch.
What six transactions show
Run the matrix against six recent InsurTech and insurance-platform transactions and the logic reads backward cleanly. Four of them sit on a control spectrum (partnership, minority stake, staged majority, full ownership) and were placed there deliberately rather than by default:
| Capability type | Matrix answer | Evidence |
|---|---|---|
| Local regulatory and banking relationships, high tacitness | Acquire outright | Cover Genius’s acquisition of Friendsurance for DACH bancassurance capability: regulatory expertise and bank relationships built over a decade don’t transfer any other way. |
| High-tacitness pricing IP where founder/team retention matters | Staged acquire | AXA’s purchase of 51% of Prima for €500m, with call/put options on the remaining 49% exercisable in 2029 and 2030: a structure that buys control now and likely keeps Prima’s team incentivised through the option period. |
| Capability already de-risked through years of partnership | Convert to full ownership: three conversions in 17 months |
|
| Scale and cost synergies in an already-owned category | A different question entirely | Ageas’s £1.295bn acquisition of esure from Bain Capital reads as consolidation for scale in a mature UK market; the criteria that matter are cost synergy and market share rather than tacitness or regulatory dependency. |
Not every acquisition is a capability acquisition
The esure row is the odd one out, deliberately. Ageas didn’t have Prima’s build-from-scratch problem; it already runs UK personal lines. What it bought was density: more premium, a bigger cost base to consolidate and a stronger position on the price comparison sites it already used. Score that on tacitness or regulatory dependency and you are measuring the wrong thing. The right criteria are cost synergy, market share and capital return. Treat every large transaction as a capability play and the matrix starts explaining noise instead of signal.
The point of the exercise
None of this removes judgment from the decision, and it isn’t an M&A framework; most of what it evaluates never reaches a term sheet. But whoever runs the exercise needs to sit close enough to revenue to know which answer the business can absorb, whether that is a licensed regulatory shell nobody can sell through, an underwriting engine the sales team doesn’t trust, or a partnership whose exclusivity terms block the next channel deal. In practice that means the scoring is driven by whoever owns the number the capability is supposed to move, because they inherit the commercial consequences when it doesn’t work. That makes it commercial architecture before it is corporate development.
Score the capability, not the org chart, and the fork in the road gets less political and more likely to survive contact with year two.