Kite & Anchor: insurance onboarding people actually finish

Kite & Anchor sells landlord and small-business insurance in four states — the kind of cover people buy quickly, reluctantly, and usually on a phone between other tasks. Their product was good. Their quote flow was a filing cabinet wearing a web form: forty-one fields, one long page, and a completion rate of 19%.

When they engaged us, their growth team had already tried the obvious. The button was rounder. The hero had a photo of a happy couple with a dog. Completion was still 19%. The problem wasn't persuasion — it was arithmetic. Nobody finishes a form that asks for everything before it has given anything.

The challenge

We mapped the original flow and found the usual insurance pathology, in concentrated form:

• Front-loaded everything. Fourteen fields before the visitor had seen a single dollar figure. The flow demanded commitment before it demonstrated relevance.
• The fatal question. Step nine asked for an excess amount with no explanation. Session recordings showed people stalling there — forty, sixty, ninety seconds — then leaving. It wasn't a question, it was an exam.
• Compliance copy that protected nobody. The product disclosure statement was linked as a PDF. The on-page summary was written to indemnify, not to inform. Customers were clicking "I understand" on text nobody understood, which is the worst of both worlds: legally exposed and confusing.
• No way back. Abandon the flow and your answers died with the session. Returning customers started again — and most didn't.

Kite & Anchor's head of product, Sarah Chen, framed the brief beautifully: "We're asking strangers to do paperwork. Make the paperwork feel like a conversation with someone competent."

The approach

Progressive disclosure with a spine. We rebuilt the quote flow around the discipline we describe in progressive disclosure for genuinely complex tools: reveal complexity at the moment of decision, never before. The flow now asks five questions to produce an indicative price — state, property type, building value band, construction decade, occupancy. Everything else unfolds after the number exists, each question arriving with its reason attached: "Your insurer uses this to price storm risk," not "Field required". Forty-one fields became forty-one moments, sequenced by how much the answer moves the premium.

The excess explainer, in dollars. The fatal question became the flow's best moment: a slider with real numbers. Drag the excess from $500 to $2,000 and watch the annual premium move — the actual trade-off, computed on their real quote, updating live. Nobody leaves a slider confused. Almost nobody left this one.

Honest defaults, always labelled. We pre-fill what public data allows: construction era from the address record, estimate bands from council data. Every pre-filled answer wears a small "estimated — tap to correct" tag. This was the hard-fought design principle: a default you can see and edit builds trust; a default you can't see is a dark pattern with a suit on. Pre-filled fields had a 91% acceptance rate, and corrections clustered where the public data was genuinely stale, which told us the defaults were honest rather than coercive.

Plain-English summaries with a paper trail. Each cover tier gets a 90-second plain-words summary — what's covered, what isn't, the three exclusions most likely to surprise a landlord. Every sentence hyperlinks to the exact clause in the PDS it summarises, and the whole mapping is versioned, so compliance can diff a wording change against the source in one glance. This was the compliance team's idea, not ours, and it changed the engagement: the lawyers stopped being a gate and became reviewers of an artefact they could actually check. Sign-off time on wording changes dropped from weeks to days.

Resume without ransom. Leaving mid-flow saves your progress; returning arrives via a magic link straight to where you stopped. No forced account creation, no email hostage exchange. This one feature recovered 14% of abandoned quotes within 72 hours. It's the same philosophy as our writing on forms people actually finish: a form is a promise, and breaking the visitor's work breaks the promise.

Comparison without theatre. The three tiers display side by side, differences highlighted rather than similarities padded. We banned the strikethrough-price anchoring the old site used; it tested well and explained badly, which is a trade we don't make.

The outcome

The rebuilt flow launched in March 2025 across all four states. Illustrative figures from the first two quarters, compared with the same months a year prior:

• Quote completion: 19% → 41%. The biggest single lift came from the indicative-price-first restructure, not from any copy change.
• Median time to quote: 8 minutes 40 seconds down to 3 minutes 50 seconds — and the flow now feels shorter than that, because every question arrives with its reason.
• Excess-related drop-off: effectively eliminated. The slider question now has the highest engagement of any step.
• Quote-to-bind rate: up 47%, which the team attributes to customers arriving at purchase already understanding what they bought.
• Support volume: "what does my policy cover" calls to the service centre fell 33% — the plain-English summaries did their job on both sides of the phone.

"Every agency told us to make the form shorter," Sarah says. "Brassfern made it answerable. Those are different projects, and only one of them was the one we needed."

What we'd tell other regulated-product teams

Your compliance team is a design partner if you give them an artefact they can check — and a bottleneck if you hand them prose. Show every default, attach a reason to every question, and put a real number on the screen before you ask for the tenth field. The flows that finish are the ones that feel like they're being filled in with you.

We do this work across regulated spaces — see how we gave a community bank a human voice, browse our fintech practice, or brief us on your flow.