Most teams respond to a flat conversion rate by redesigning something. A new homepage, a new popup, a new checkout. Weeks later the number is unchanged, because the redesign guessed at the problem instead of locating it. A conversion audit replaces that guess with a short, repeatable investigation: measure every step, find the step that loses the most people, then look at why.
It is one pass through your funnel, from first landing page to confirmation screen, with a number attached to each step and a piece of evidence attached to each number. Done well it produces a ranked list of fixes instead of a mood board. For the wider context, see our guide to increasing website conversion rates.
What is a conversion audit, and what is it not?
A conversion audit is a structured review of every step between a visitor arriving and completing the action you want: purchase, signup, demo request or email capture. For each step it records traffic, completion rate and the drop-off between it and the next step, then combines those numbers with qualitative evidence from recordings, surveys and support tickets.
It is not a heuristic checklist ("does your button have contrast?"), though that comes later, and it is not a list of best practices copied from another site. Both assume the problem is already known. The audit's job is to prove where the problem is before anyone touches a pixel.
How do you map the funnel before you measure it?
Write the path down in the order a real visitor takes it, not the order your sitemap lists pages. A typical ecommerce path runs: ad or search result, collection page, product page, cart, checkout, confirmation. A SaaS path runs: landing page, pricing, signup form, activation step, first real use. Add every popup, chat prompt and form that interrupts the path, because they change behaviour at the step where they appear.
Name the single conversion you are auditing: "checkout completion" or "trial signup", not "more revenue". One audit, one conversion, otherwise the numbers argue with each other.
Which metrics belong in the audit?
For each step record five things: sessions that reached it, sessions that continued, the step conversion rate, the median time spent, and the top exit page for people who left. Split each row by device, traffic source and new versus returning visitors.
| Step | Metric to record | What it usually tells you |
|---|---|---|
| Landing to collection | Scroll depth, click-through | Offer or headline mismatch |
| Product page | Add-to-cart rate, time on page | Price, proof, or unclear variant choice |
| Cart to checkout | Cart abandonment rate | Shipping shock, forced account creation |
| Checkout | Field-level completion | Form friction, missing payment options |
Add tracking for the prompt itself. Our breakdown of popup analytics metrics covers what a capture prompt should report so it fits into the same table.
How do you find the biggest leak first?
Convert every step into the same unit: visitors lost per thousand sessions. A step that converts at 60 percent loses 400 people per thousand. A step at 95 percent loses 50. The largest absolute loss is almost always the best place to start, even when another step has a worse-looking percentage.
Ranking by percentage alone sends teams to fix a rarely visited page while the checkout leaks hundreds of buyers a week. Multiply by traffic, then by the average order value of the people lost, and the list often reorders itself.
Why do teams mistake traffic problems for page problems?
A conversion rate can fall because the page got worse or because the traffic changed. A campaign that brings colder visitors lowers conversion without anything on the page breaking. Before attributing a drop to design, check the source mix: did paid share grow, did a new geography appear, did mobile share climb?
Segment the rate by source and device for the same period. If every segment converts at roughly the same rate as before and only the mix shifted, the page is fine and the targeting needs work.
What qualitative evidence should you collect?
Numbers say where people leave. Recordings, surveys and search queries say why. Watch ten session recordings from the worst step and note what visitors hesitate on, what they click twice, and where they scroll back. Read the last twenty support tickets that mention pricing, shipping or returns, and run a single-question exit survey on the leaking page. Our guide to heatmaps and session replays covers how to read those recordings without over-interpreting three angry clicks.
Site search queries and chat objections are unprompted, so they are more honest than a survey you design. Keep the raw quotes next to the metrics: the quote convinces a stakeholder, the metric justifies the work.
How do you turn findings into a test list?
Write each finding as a hypothesis with a mechanism: "shipping cost appears only at checkout, so showing it on the product page will reduce cart abandonment." Add the expected effect, the effort and your confidence. Outright bugs get fixed immediately, without a test.
Everything else becomes a queue you work through with controlled experiments, and each experiment feeds the next audit. If testing is new to your team, read common A/B testing mistakes before shipping the first variant: a badly framed test produces a confident answer to the wrong question.
Keep the definitions: same metric, same segment, same date range, next quarter. Reporting one number two ways is how funnels get declared "fixed" without improving.
FAQ
Q: How long does a conversion audit take? A: One focused day for a small site, two to three for a larger funnel, assuming your analytics already track the main steps. Most of that time goes into recordings and support tickets.
Q: Do I need a dedicated analytics tool? A: No. Event tracking in your existing analytics is enough to build the funnel table. Session recordings and a one-question survey add the "why" cheaply, and both are easier to justify than a new platform.
Q: How often should we repeat the audit? A: Quarterly is a sensible rhythm, plus once after any change to pricing, checkout or traffic mix. Repeating it with the same definitions is what makes the comparison meaningful.
Q: What if our conversion rate is already high? A: A high rate still hides a large absolute loss, because traffic is high too. Audit the step with the biggest loss per thousand sessions, not the worst percentage, and check whether email capture is leaving a second conversion unclaimed.
If email capture is one of the steps in your funnel, you need to know which prompt, on which page, produced which signup. That is what heycustomer.byako.dev reports, so your next audit starts from real numbers rather than a guess.