Most CRO advice starts with A/B tests. Wrong order. Learn the audit-first process: where conversion actually leaks, what to fix outright, and when testing earns its place.
Conversion rate optimisation has a branding problem. The phrase summons button-colour tests and tools that cost more than they return. For most small e-commerce businesses, the first 80% of conversion gains come from an audit and a list of fixes, not from experiments. You don't have the traffic to run statistically honest tests anyway, and that's fine.
Figure 1 · Where the funnel narrows (illustrative example, not a benchmark)
Illustrative horizontal bars shrinking at each stage: visit, product view, add to cart, checkout, purchase. Widths are illustrative, not real benchmarks. The point is that each step loses some buyers, and the audit finds which step loses the most. Visit 100 Product view 75 Add to cart 45 Checkout 30 Purchase 15 Illustrative widths only. Pull your own numbers, then fix the widest drop first.
A/B testing answers "which of these two versions is better". An audit answers the question: "Where is the funnel actually leaking?” If you test before you audit, you optimise the wrong page with confidence.
A worked example. A store converts at 1.2% sitewide and concludes the product page needs work. The audit shows product-to-cart at a healthy rate, but 60% of mobile checkouts are abandoned at the shipping step because the rate only appears after address entry. No test needed. Show shipping earlier, watch the funnel.
Pull these five numbers from your analytics. Segment mobile and desktop separately, because they fail differently.
Stage | Metric | What a leak here usually means |
|---|---|---|
Visit to product view | % of sessions viewing a product | Navigation, search, collection page problems, wrong traffic |
Product view to add to cart | Add-to-cart rate | Price clarity, imagery, missing answers, weak offer |
Add to cart to checkout | Cart-to-checkout rate | Surprise costs, forced account creation, cart friction |
Check out to purchase | Checkout completion | Shipping shock, payment options, form length, errors |
Purchase to second purchase | Repeat rate | A retention problem wearing a CRO costume. See the customer lifecycle guide |
Diagnose top-down. The stage with the largest gap relative to your historical baseline is fixed first.
These are corrections, not experiments. Ship them without ceremony:
Show total cost early. Shipping and tax surprises at the final step are the most consistently reported drivers of abandonment in checkout research, including the Baymard Institute's long-running studies.
Guest checkout on. Account creation is a post-purchase ask.
Speed. Conversion and load time are tightly linked. The site speed guide has the ranked fix list.
Mobile forms. Correct keyboards, address autocomplete, no horizontal scrolling.
Answer the question on the page. Returns, delivery time, sizing. Every unanswered question is a tab the buyer opens and doesn't come back from.
Express payment options. Wallets remove the form entirely for a chunk of buyers.
Rough maths: to detect a 10% relative lift on a 2% conversion rate with normal confidence, you need tens of thousands of sessions per variant. Under roughly 1,000 conversions a month, run the audit-fix loop and sequential before/after measurement instead. Above that, test the big swings: offer framing, page layout, pricing presentation. Never button colours.
Tip: When you do before/after measurement, change one thing per fortnight and annotate the date in your analytics. Without the annotation, you're reading tea leaves in three months.
The honest answer: against your own baseline. Published averages blend industries, price points and traffic mixes until they mean nothing. A $400 average-order-value furniture store converting at 0.8% can outperform a $25 accessories store at 3%. Track your own number monthly, segmented by device and channel, and beat it.
Abandoned cart recovery: winning back the checkouts that stalled
GA4 ecommerce tracking: getting funnel numbers you can trust
There's no universal number worth chasing. Conversion varies with price point, traffic source and category. The useful benchmark is your own trailing average, segmented by device and channel.
Not below roughly 1,000 conversions a month. Audit, fix, measure before and after. Testing tools earn their cost once you have the volume to reach significance in under a month.
Showing full costs before the final checkout step and turning on guest checkout. Both address the most common reasons for abandonment identified in checkout usability research.
Qwrki is the operating layer that runs ecommerce growth work end-to-end, so the CRO audit, the fix list, and the before/after measurement live in one place rather than scattered across tools and spreadsheets. We first diagnose the funnel against your own baseline, ship the corrections, and then test only the swings worth testing. Book a call and we'll start with a read of where your funnel actually leaks.
How to do keyword research for an online store: map intent to page types, find buyer keywords, and stop chasing volume that never converts.
A content strategy for online stores that earns revenue, not just traffic: comparison pieces, buying guides and the publishing order that compounds.
Paid traffic deserves a dedicated landing page. Here is the message-match principle, the anatomy of a page that converts, and when the homepage is actually fine.
