From project-by-project research to an always-on intelligence function

Client

Date

$3B+ US mutual insurance carrier

08/2026

The challenge

A US mutual insurance carrier asked an innovation team of under ten people to carry the entire forward-looking research agenda for the company, at a moment when new entrants were taking share without competing on price. Every theme was a separate project, costing weeks of the team's own time or a consulting engagement.

The solution

FifthRow was deployed as the innovation team's standing research function: 18 systems running continuously against the carrier's own lines of business, on a schedule nobody has to start, backed by an on-demand library of 347 research templates the team could fire at any new question. A theme goes in. A source-verified, decision-ready brief comes out, on a cadence, without anyone starting a project.

741,419

analyst-hours equivalent, 212 FTEs

139,118

source-verified research questions

Research cadence3,278 studies, twenty months

Time per study26 minutes average, from weeks per project

Standing intelligence18 systems, 873 scheduled runs, 347 templates

Analyst-hours equivalent741,419 hours, 212 FTEs

01 — PRODUCT-PROGRAMME EVIDENCE BASE

The largest single body of work: a purpose-built system covering the external forces acting on one of the carrier's core lines, paired with a second on the volatility and risk conditions in that market. The team described it as the collection of the market – the evidence base under a multi-phase product programme.

Core Activities

Covered that market's external forces in one standing system
Ran a parallel system on volatility and risk
Tracked market structure, growth and competitive position
Extended phase one into adjacent segments and product parameters
Re-ran monthly, so phase two began from current evidence, not a refresh

The Results

A standing evidence base for that line, which the team pulls from rather than commissions. Phase two of the product programme began with the market already mapped.

130,560

analyst-hours equivalent, product programme

02 — STANDING MARKET & COMPETITOR WATCH

The always-on layer. A technology and innovation monitor running to a schedule, plus weekly decision briefs by line of business and a recurring competitor ranking against a fixed peer set.

Core Activities

Ran a scheduled technology and innovation monitor
Produced weekly property and casualty briefs, personal and commercial lines
Ranked a fixed peer set by direct premiums written, with movement explained
Tracked entrants, funding, partnerships and share shifts continuously, not on request
Routed the fact base into the carrier's assistant environment, queryable outside the platform

The Results

Competitive and technology intelligence arrives on a schedule. The team stopped starting a project every time leadership asked what had changed.

121,885

analyst-hours equivalent, standing watch

03 — PEER PRACTICE BENCHMARKING & SEGMENTATION

The most-run system in the account. One combined system covering market-by-market conditions, customer insight and technology adoption across underwriting and claims, alongside persona and segmentation research for the carrier's own personal lines.

Core Activities

Ran that combined state-management, customer-insight and technology-adoption system 51 times
Built personas and segmentation for personal lines and financial services
Converged five demographic and behavioural verticals into one monthly briefing
Encoded the carrier's personas – public perception, internal reality – into the agents
Validated public evidence on how peer carriers redesign operations

The Results

A repeatable read on how peers are redesigning underwriting and claims, and a segmentation base that other systems in the account reuse instead of rebuilding.

46,539

analyst-hours equivalent, underwriting and claims

04 — DEAL SOURCING & WATCHLIST

A rolling 90-day scan of insurance-sector M&A, feeding a monthly executive brief and a running watchlist maintained for the innovation team.

Core Activities

Scanned insurance-sector M&A on a 90-day cycle
Built a monthly system synthesising four domain monitors into one executive brief
Maintained a running watchlist, not one-off reports
Surfaced startups addressing identified pain points as acquisition or partnership candidates

The Results

A standing watchlist and a monthly brief. Deal-adjacent intelligence stopped being an ad hoc request and became a document that updates itself.

90 days

rolling cycle, watchlist never stale

05 — NEW PRODUCT EXPLORATION

Where the intelligence layer turns into product work. Deep dives into the lines the carrier writes very little of, plus concept building for new offers.

Core Activities

Ran deep dives into the lines the carrier barely writes
Built an idea-to-concept system: personas, value propositions, business models, viability assessments
Studied what drives customers to bundle products
Traced an emerging risk driver across several of the carrier's lines
Fed each from the standing systems, not fresh research

The Results

Product-line and bundling questions answered with market evidence in days. One deep dive found a gap where the carrier already held the customer relationship but almost none of the business that goes with it.

63,107

analyst-hours equivalent, new products

What it proved

Over twenty months, close to a year of active days, the team ran 3,278 studies covering 1,315,971 sources and 139,118 source-verified questions. Eighteen standing systems ran 873 of those on a schedule, with nobody in the room. Five systems built around the carrier's own lines replaced a queue of separate research projects. Average run time: 26 minutes.

The team holds a function rather than a set of deliverables. Monthly volume went from 40 runs to 491: the constraint moved from what the team could research to what it wanted to know. Across 3,278 studies the platform records 741,419 analyst-hours saved, the equivalent of 212 full-time analysts. Work once commissioned externally now happens inside the team, on a cadence, at a cost it controls.

Results

Metric	Benchmark	Result	Improvement
Monthly volume	40 runs, first full month	491 runs a year later	12× more
Standing systems	project-by-project, on request	18 systems running unattended	Continuous, nothing to commission
Verification of findings	analyst judgement per project	139,118 questions checked against source	Auditable by default
Analyst-hours saved	weeks per theme, or consultants	741,419 hours, 3,278 studies	Equivalent of adding 212 full-time analysts

UNRATIONED QUESTIONS

Research no longer starts with a project approval, so the team asks every question worth asking rather than only the ones worth commissioning.

ENCODED SYSTEMS

A question is encoded once as a standing system and re-run on a cadence, so the asset is the application rather than any single report.

HANDS-OFF DELIVERY

Formatted briefs arrive on schedule for colleagues who do not work in the platform, so the output reaches beyond the handful of people who run it.

QUERYABLE FACT BASE

The accumulated evidence is reachable from the assistant tools people already have open, so findings get queried in place instead of re-uploaded by hand.

COMPOUNDING CONTEXT

Every cycle builds on the last, so the fact base carries accumulated context that a fresh research project could not reconstruct.