Ecommerce conversion rate optimization
Traffic to your category pages is already paid for or already earned. We find where it leaks between the category and the order, using GA4 and session recordings, and every issue comes with its evidence: page, element, recording and exact timestamp.
Where we start
No single source is enough. We cross-check the GA4 funnel, session recordings and heatmaps (Microsoft Clarity, Hotjar or Zoho PageSense), Search Console, Merchant Center, Google Ads and the store’s real orders.
Comparisons are made on the same device: mobile from ads against mobile from other channels, not mobile against desktop. Before looking, we clean the data: bots excluded, team sessions excluded, prices visible in recordings and form fields masked.
- Every issue with page, element, recording and exact timestamp
- Severity on three levels: blocking, serious, minor
- Conclusions that don’t hold up are withdrawn in writing
What we check, page by page
These are the points where orders are most often lost. Each has a precise criterion.
Mobile first screen
The buy button belongs on the first screen: compact header, image within half the height, sticky bar with price, size and button.
Category pages
Sold-out items on top, images that don’t load, missing filters, brand and name swapped on product cards.
Product page
Buttons that look disabled, wrong savings math, a single photo, specs inside an image, related products of the wrong gender.
Site search
Zero results for top brands, exact match only. It needs multi-word queries, synonyms, typo tolerance and suggestions.
Cart and checkout
Wrong default country, fees that contradict the promise, a prominent coupon field, total and order button off-screen.
Trust
Reviews near the price and in checkout, clear returns and shipping before the button, the same promises across site, feed and ads.
In what order things get fixed
First whatever stops a loss, redesign last.
1. Daily losses
Ads landing on the wrong pages, feed errors: fixable without touching the site, and they stop a daily loss.
2. Consent, prices, reviews
Requirements for the cookie banner, price display and reviews, to be checked with your legal advisor.
3. Bugs that break the page
JavaScript errors, elements that don’t load, pages that shift while loading.
4. Redesign
Only then: without consent, prices and tracking in order, you couldn’t even measure whether it worked.
What you get
A single numbered list of changes for whoever builds your site, each with severity, effort (low, medium, high) and a link to where it’s explained. When needed, an interactive prototype as the spec.
Product sorting on category pages
The “most relevant” order decides what visitors see first. We propose a score where availability outweighs everything, followed by recent sales, new arrivals and active discounts. Margin only enters once costs are complete.
Before proposing it we backtest it on historical data and present it as an order of magnitude, not a guarantee.
A/B tests: when they make sense
A test only measures something with enough sessions. To detect a 20% lift at a 1% conversion rate you need about 39,000 sessions per variant. With few orders a month, a test measures nothing.
Below that threshold we rely on solid external evidence, such as published usability research, and measure before and after on equal-length periods. Where possible we try one or two sample categories before rolling out.
The measurement plan
Before optimizing, the numbers have to be right. These are the requirements we verify.
GA4 ecommerce events
view_item, add_to_cart, begin_checkout and purchase, with correct values and items.
Consent Mode v2
Consent denied by default, updated only after the choice; “Accept” and “Reject” with equal weight.
One conversion per purchase
A single primary purchase action in Google Ads: no orders counted twice.
Reconciliation
GA4, Google Ads and real orders compared, with the gap stated.
Duplicate tags and errors
Tools installed twice, JavaScript errors, wrong landing pages.
Recording samples
Recordings cover only part of the sessions: a “zero” in the sample isn’t proof.
Our process, month by month
One point of contact, a monthly call, a report with evidence and verified numbers.
Step 1Access and clean data
GA4, Clarity or Hotjar, Search Console, Merchant Center, Google Ads; bot and team filters, consent check.
- Step 2
Evidence-based audit
Segments by channel and device, recordings, issues with evidence and severity.
- Step 3
Change list
A single numbered list with priority and effort; a prototype when needed.
Step 4Verification
Same segment, equal-length periods, before and after every change; then next steps and tuning.
Ecommerce CRO FAQs
The questions we hear before we start.
How much traffic do I need for CRO?
For the audit, the recordings and funnel you already have are enough: issues show up even with a few hundred sessions. A/B tests, on the other hand, need tens of thousands of sessions per variant.
Do you run A/B tests?
When the numbers allow it. Below the threshold we rely on solid evidence and measure before and after, without calling something a “test” when it isn’t.
Which tools do you use?
GA4, Microsoft Clarity, Hotjar or Zoho PageSense, Search Console, Merchant Center and Google Tag Manager, server-side too.
Do I need a site redesign?
Almost never as step one. First we stop the losses, fix consent and tracking and correct bugs; redesign comes later, when it can be measured.
What do I get from the CRO audit?
A PDF with the issues and their evidence, plus a single numbered list of changes with severity and effort, walked through on a call.
How much does it cost?
Each activity has its own monthly price, which you can calculate on our Pricing page by choosing items and quantities. CRO and measurement cost 600€ a month. Alternatively there’s the tailor-made Enterprise package, with a reduced fee and a performance bonus.
Where are orders leaking on your site?
With access to GA4 and your recordings, we’ll show you the first issues, with their evidence.