·The Market Challenge
Most of the market is underinsured, and nobody finds out until a claim. Around 76% of UK commercial buildings carry sums insured below the cost of rebuilding them, down from a peak near 83%. The cause is rarely dishonesty: it is declared values carried forward unchanged through the construction cost inflation of 2021–23, assessments left to age past the three-year point, and rebuild figures that omit demolition, professional fees, Building Regulations compliance and irrecoverable VAT — additions that on a benchmarked portfolio average 37% on top of net rebuilding cost.
The penalty falls on partial losses, not just total ones. The condition of average reduces a settlement in the same proportion as the underinsurance. A £400,000 rebuild insured for £300,000 settles an £80,000 claim at £60,000. A business does not need to burn down to be exposed, and most claims are partial — so this is a live cost, not a tail risk. Beyond average, the Insurance Act 2015 permits a proportionate remedy or avoidance where under-declaration breaches the duty of fair presentation, and Consumer Duty makes defensible sums insured part of evidencing a good outcome.
Four errors produce most of it. Comparing a rebuild cost to whole-portfolio TIV without stripping contents, plant, stock and business interruption. Using market value instead of reinstatement cost. Failing to index-link between assessments. Omitting the additions the RICS standard requires. Each is easy to make and invisible on a schedule.
And a full assessment cannot be commissioned for everything. A RICS reinstatement cost assessment is a chargeable professional instruction. Across a book of hundreds of SME risks, the commercial question is not whether values are wrong — they are — but which ones warrant the spend, and how to evidence the decision afterwards.
The Solution
RESTITUTIO, a single-file, fully offline application that tests declared sums insured against independently derived RICS reinstatement costs, triages what it finds, and carries the corrections through to an endorsed policy. It runs on 500 synthetic SME risks across 75 classes of business, with a complete distribution graph of eight brokers, three MGAs, three binders and four carriers behind them. Every figure on every screen clicks through to the evidence underneath it. On that portfolio it returns 79.8% adequacy, with 71.8% of assets below the 90% watch threshold and £110m of reinstatement cost undeclared against a £510m book.
Benchmark. A £/m² rate by archetype, construction, age, listing and region, applied to gross internal area, then demolition, fees, Building Regulations uplift, inflation across the lead-in plus half the build period, and irrecoverable VAT where the trade makes exempt supplies. Declared TIV is decomposed so only the buildings element is compared.
Evidence. Six to eight source documents per asset — policy schedule, statement of fact, EPC, OS MasterMap, title register, prior assessment — each rendered as a facsimile with captured fields in gold, compared in blue, conflicting in red. Where four parties measured the same building and got four answers, the tool raises the exception.
Triage and remediation. A dot plot places every asset by adequacy so the shape of the book is visible at a glance. Flagged assets become cases with a full lifecycle: instruction, assessment, client letter, recorded response, revised declaration, endorsement. Once a surveyor has measured a building their figure supersedes the model, and the asset re-scores live.
Defensibility. Every assumption change records what moved and how many assets changed triage band. Documents opened, letters sent and responses recorded are timestamped and exportable.
Four Lenses: The challenge each actor faces, and what it delivers
| PERSONA | THEIR CHALLENGE | WHAT RESTITUTIO GIVES THEM | OUTCOME IT DRIVES |
| ARBITER™ the analyst one who goes to see, and decides | Hundreds of risks, a finite assessment budget, and no defensible way to choose. Ranking by adequacy percentage inverts priority — a small unit at 60% matters less than a large one at 82%. | Ranking by recoverable exposure rather than percentage. A distribution view of the whole book, drill-through to every figure, and the engine marked against what surveyors actually measured. | Assessment spend concentrated on roughly a fifth of the book. Triage thresholds tuned to this portfolio rather than a market average, with the confirmation rate published. |
| INTERPRES™ the broker the go-between; literally, a broker | Values carried forward year on year. Renewal is the only moment a declared value moves, and it arrives without warning. No file evidence of the steps taken if a claim later exposes a gap. | A scored client book, a twelve-month renewal calendar, and a client letter that performs the average calculation on the client's own figures. Every response recorded, either way. | A decline documented in writing is the strongest Insurance Act and Consumer Duty position short of a correction. The hygiene scorecard names which of four behaviours to change. |
| MANDATUM™ the MGA business done on another's authority | Accountable to a carrier for a book you did not underwrite, assembled by brokers of very different quality. A bordereau carries declared values only — nothing in it evidences how they were reached. | Every reported line reconciled against benchmark with exceptions surfaced, and a broker league table built on four observable behaviours rather than opinion. | Underinsurance stops being a portfolio statistic and becomes a distribution problem with names against it: 92.6% down to 65.8% adequacy across eight brokers on the same book. |
| FIDEIUSSOR™ the carrier the surety who stands behind | Binder-level underinsurance is invisible until a loss. Premium is rated on declared values while exposure sits against real reinstatement cost, and the two are not the same number. | Aggregate exposure against declared, delegated measured against direct, binder-by-binder performance, and the claims where average was applied. | Delegated business running at 76.4% adequacy against 81.5% written direct — the figure that goes into the binder review. Over-declaration offsets part of the correction, so the net premium effect is smaller than it appears. |
Choosing a persona re-scope every shared view — the distribution plot, the triage queue, the three-year cycle and the headline figures all narrow to that actor's rows. Seven reconciliation invariants run on every rebuild so the four lenses cannot drift apart and contradict one another.
One case, four lenses: Why it is one system and not three
A listed public house flags at 64.4% adequacy. The analyst raises the case; it appears at once on the broker's client book, inside binder CB-4471 for the MGA, and in the carrier's delegated exposure. It routes to a site survey rather than a desktop run, because heritage fabric cannot be modelled from imagery and a rate table. The surveyor returns £8.24m against the engine's £7.28m — a 13.2% variance, squarely in the range the routing rules anticipate for listed property, which is the evidence the triage call was right. A client letter performs the average calculation on the client's own figures; the acceptance is recorded, the declaration revised, and the asset re-scores from 64.4% to 100.0% with the measured figure superseding the model. Every step is timestamped. The same event moves the number on all four dashboards, because there is only one record underneath them.
| 7 / 7 reconciliation invariants passing | 500 · 75 risks · classes of business | 8 · 3 · 4 brokers · binders · carriers | 6–8 source documents per asset | 0 external dependencies |
What is Not Built
Multi-user assignment with real identity, a server-side append-only audit log, and a live cost-data feed. The rate table is synthetic and stays that way until you load your own: BCIS data is licensed, and embedding a plausible-looking placeholder as though it were real would undermine the one thing this tool is for. No competitive benchmarking against commercial reinstatement platforms has been carried out, because that would require access to those products rather than assumptions about them.