Home Insurance Quote Funnel: Build One That Produces Sellable Leads

If your homeowners form converts at a rate you are proud of and your buyers keep rejecting the leads, the funnel is doing exactly what you built it to do. A home insurance quote funnel is the sequence of questions a visitor answers between clicking an ad and becoming a record you can sell or route to a producer. Optimise it for form completions and you will get form completions: half-typed addresses, renters filling in a homeowners form, and people in ZIP codes where nothing on your panel is writing new business. Optimise it for accepted, quotable, bindable leads and your completion rate will fall while revenue per session rises. Take that trade.

This post is the implementation companion to the two economics posts in this cluster. If you are still deciding whether to own the funnel at all, read the affiliate versus lead generation comparison first, and check what homeowners leads actually cost in 2026 so you know what an accepted lead is worth before you start designing for one.

Why conversion rate is the wrong target

An accepted lead is a record the buyer's system takes, pays for, and does not claw back. Between a submitted form and an accepted lead sit four gates, and only the first one is about your form. The record has to pass the buyer's API validation, it has to be quotable against a carrier that has appetite for that property, it has to be contactable by a human, and eventually it has to bind. Every field you remove to lift completion rate removes information from gates two through four.

Run the arithmetic on a thousand sessions instead of on a percentage and the picture changes. A four-step funnel that completes at 12 percent produces 120 leads; if 70 percent are accepted, that is 84 paid records. Strip it down to a two-field form, lift completion to 18 percent, and you get 180 leads; if acceptance falls to 40 percent because half of them are renters, duplicates, or missing the roof data the buyer prices on, that is 72 paid records. You improved the metric on the dashboard and lost revenue. These numbers are illustrative, not benchmarks. The point is the shape of the curve, and you can only find your own inflection point by measuring acceptance by funnel variant.

Quotability is the gate most publishers underweight, and it is mostly a property-data problem rather than a consumer-intent problem. We covered the mechanics separately in why quotability is the real constraint on home leads. The funnel's job is to collect, in the smallest number of questions possible, the specific facts that decide whether a carrier will look at this house at all.

Address first, contact last: the order that works

Almost every homeowners funnel we are asked to fix opens with name, email and phone. The logic is that you capture contact data before the visitor can escape. The logic is wrong, and it is wrong in a way that costs money twice.

First, contact-first gives you a phone number attached to nothing. You cannot ping it, you cannot price it, and you cannot route it to a producer who has anything to say on the call. Second, it forces you to show the consent disclosure before you know whether you will ever have a sellable record, so your certificate volume balloons with junk and your buyers' compliance desks start asking why your consent-to-lead ratio looks like that.

Address-first inverts both problems. The moment you have a validated street address you can enrich the property, check whether any carrier on your panel writes that ZIP, and kill the session early if the answer is no. You spend the visitor's patience on questions that are still live.

Four steps is the right number for homeowners. One long page loses badly on a phone because the visitor can see the whole ask at once and prices it. Nine micro-steps with one question each look great in a demo and train fatigue by step five; every extra screen is another render, another chance for a slow network to lose someone. Group fields that are trivially easy, and give a hard question its own screen.

The obvious objection to contact-last is remarketing. If you never get an email, you cannot chase abandoners. The honest answer is that partial capture at step four, fired on blur of the email field, recovers most of what you would have had, and an abandoner captured at step one with no property attached is worth close to nothing anyway. You also cannot legally call them, because you never took consent. Contact-first buys you a list you are not allowed to use.

Which property fields to ask for and which to enrich

The rule we use: ask for a field only if it changes routing or price right now, and a normal homeowner can answer it in under four seconds without leaving the page. Everything else gets enriched from property data, or asked by the producer on the call.

Two fields sit on the fence. Mortgage escrow status is worth one tile question only if your specific buyers pay more for it, so test it against acceptance rate rather than adding it on principle. Bundling intent, meaning whether they would also quote auto, is usually worth its screen real estate because a bundle record prices higher with most buyers; the economics are laid out in what an auto and home bundle lead is actually worth.

Where the consent language goes without tanking completion

We are not going to re-explain TCPA mechanics here. The current state of play for lead buyers and sellers, including what changed after the Eleventh Circuit vacated the FCC's one-to-one consent rule in early 2025 and why most buyer contracts kept the named-seller discipline anyway, is in our TCPA consent guide for lead buyers. The certificate side, meaning what TrustedForm and Jornaya actually capture and what a buyer's QA team looks for in a replay, is covered in the consent certificate breakdown. Treat both as a starting point and get your own counsel; state mini-TCPA statutes in Florida, Oklahoma and Maryland did not move when the federal rule did.

What matters for funnel design is placement, and the answer is narrower than most people expect.

Beyond placement, four things are worth being strict about. Set the disclosure at your body font size, not at ten-pixel grey; contrast and size are the first things a regulator or a plaintiff's firm screenshots. Do not pre-check the box if you use one, and if you are using clickwrap instead of a checkbox, make the button label describe the action. If your partner list is long enough to need a link, make the link target a stable page and store a version hash of the list at submission time, so you can prove what it said on the day. And log a consent snapshot alongside the lead: the full disclosure text as rendered, the URL, timestamp, IP, user agent, and the certificate ID.

On completion rate: a clear disclosure above the button costs less than people fear. What actually costs completion is the surprise, and putting it on step four after three steps of investment converts better than putting it on step one before any. The reason to put it late is compliance-sound and conversion-sound at the same time, which is rare enough to be worth noticing.

Mobile form behaviour and the input types that matter

Most homeowners quote traffic is on a phone, frequently on a mediocre connection, frequently one-handed. The following are not polish items. Each one of these has, on its own, moved step completion by a visible amount on funnels we have rebuilt.

• Never use type="number" for phone or ZIP. Use type="tel" with autocomplete="tel" for phone and inputmode="numeric" with autocomplete="postal-code" for ZIP. Number inputs add spinners, strip leading zeros, and change value on an accidental scroll.
• Set input font size to at least 16px. Below that, iOS Safari zooms the viewport on focus and your layout jumps sideways. People lose their place and leave.
• Feed autofill properly. autocomplete="street-address", "address-line1", "email", "given-name", "family-name". A visitor who autofills four fields in one tap is a visitor who reaches step four.
• Drop the fifty-state select. Infer state from the validated address or the ZIP. Long native selects on mobile are a genuine drop-off point.
• Avoid aggressive input masking. Masks that reformat as the user types fight password managers and autofill, and they produce phone values your validation then rejects.
• Validate on blur, never on keystroke. And never clear a field on a validation failure. Telling someone their phone number is wrong while they are halfway through typing it is the most reliable way to lose them.
• Keep the primary button in the viewport. A sticky footer button on small screens, with the consent text above it and both visible together on the final step.
• Test with autofill switched on. On a real iPhone, in Safari, with a saved address and a saved card. Simulators do not reproduce the failures that matter.

What to do when nobody on your panel quotes that state

Carrier appetite in homeowners has narrowed sharply in coastal and wildfire-exposed markets, and whole ZIP ranges now have no bid at any price. We wrote about how that reshapes bidding in panel shrinkage and what it does to home lead bids. The funnel question is what you do with a visitor you cannot monetise the normal way.

The cheapest fix sits upstream of the funnel entirely: exclude the ZIP ranges at the media buying layer so you never pay for the click. Do that first. The branch at step one is for organic traffic, referral traffic, and the inevitable gaps in geo targeting. Be honest with yourself that the branch monetises at a small fraction of an accepted lead. Build it anyway, because the alternative is zero revenue plus a visitor who tells people your site wasted their time.

Speed to contact, and tracing a bound policy back to a step

Speed to contact is usually discussed as a call floor problem. It is mostly a funnel design decision. If your form writes to a database and a job picks it up on a schedule, you have built a delay into the product and no amount of dialer discipline will remove it. Post on submit, synchronously, and treat a failed post as an incident rather than a retry that happens quietly at 3am.

Three other design choices move speed more than anything the call floor can do. Put a click-to-call option on the confirmation screen, because a visitor who rings you within thirty seconds of submitting is the best home lead you will handle all week. Do not gate the post behind an email verification loop; verify asynchronously and flag afterwards. And if you are ping-posting, understand that your step three answers are what determine whether the ping returns a real bid at all, which is the practical argument for asking roof age and claims before contact details rather than after. The mechanics are in how ping post lead distribution works.

The instrumentation question is simpler than people make it. Generate one lead ID when step one renders. Carry it through every event, write it into the buyer post, store whatever ID the buyer returns, and keep both. That single decision is what makes the rest possible.

Two things make this fall apart in practice. Buyers return rejection reasons as free text, and every buyer uses different words for the same thing, so build a mapping table into your own taxonomy on day one: duplicate, invalid contact, out of area, appetite, missing consent, other. Second, your accepted count and the buyer's accepted count will not match at month end. Reconcile anyway. The gap is real money, and the act of asking about it changes how carefully a buyer handles your volume.

Report it as revenue per thousand sessions, split by traffic source and by state, cohorted by the week the lead was generated rather than the week the money arrived. Blended numbers hide everything worth knowing. Once you can look at acceptance rate broken out by a step three answer, you will find that one or two answers account for most of your rejections, and at that point you can decide whether to stop buying that traffic or route those records to a different buyer.

Common questions

What is a good conversion rate for a home insurance quote funnel? There is no useful industry benchmark, because the number depends entirely on traffic source, state mix and how many qualifying questions you ask. A funnel taking branded search traffic and a funnel taking social display traffic will differ by a wide margin. Measure accepted leads per thousand sessions instead, and compare your own variants against each other rather than against a figure from a vendor's blog post.

Should the property address come before the phone number? Yes. Address-first lets you enrich year built, square footage and construction type from property data, and check whether any carrier on your panel writes that ZIP, before you spend the visitor's patience or collect contact details you may never be able to monetise. Contact-first produces phone numbers attached to no quotable property, which is the most common reason a homeowners lead gets rejected.

How many steps should a homeowners insurance quote funnel have? Four. Property address and ownership, confirmation of enriched property data, the four appetite questions, then contact details with consent. One long page loses on mobile because the visitor sees the full ask at once. Nine single-question micro-steps look good in a demo and create fatigue plus extra render points where a slow connection loses people.

Where should TCPA consent language go on a quote form? Immediately above the submit button on the final step, in your normal body font size, on the same rendered page where the phone number is entered and the consent certificate script is running. Not in the footer, not behind a modal, not bundled into a generic terms acceptance line. A certificate replay that does not show the disclosure on screen will be rejected by a buyer's compliance team.

Which property details should I ask for versus look up? Ask for roof age, prior claims in the last five years, current carrier and expiry, and occupancy, because those drive carrier appetite and a homeowner can answer each in seconds. Enrich year built, square footage and construction type from property data and show them for confirmation. Never ask for the dwelling coverage limit; consumers do not know it.

How do I handle a visitor in a state where no carrier will quote? Branch at step one on the ZIP code, before any other question. Show a state-specific page explaining the residual market option such as a FAIR plan, a route to a local independent agent, and an alternative such as an auto quote. Never run them through the full funnel to show a dead end, and never sell the resulting record into a buyer with no appetite.