Conversion Principles We’ve Learned Through 100+ Audits

1. Filter state resets on mobile grids

Pattern observed in fast fashion layouts similar to Zara

In mobile sessions with filter-heavy product grids, users often apply a size filter, scroll briefly, click into a product, then return to the grid. At that point, the filter state is lost or the scroll position resets.

Zara’s mobile site functions this way. The grid reloads on back tap, with filters sometimes removed and scroll state forgotten.

In audits, this results in users repeating the same actions: reselecting size, scrolling again, revisiting products they’ve already seen. This causes early exits that don’t look like bounces, replays show the intent but analytics often miss it.

This behaviour is common in large catalogues with visual-first browsing. If filters don’t persist, the user assumes the site didn’t register their effort and that  increases abandonment. As a CRO Agency, we audit this early in mobile-first fashion and retail experiences.

2. Checkout flow prompts account creation too early

Pattern observed in flows modelled after Sephora

In beauty and skincare flows, we’ve repeatedly seen login or account creation prompts placed immediately after a user clicks “checkout.” This happens before the user sees delivery costs or fulfilment options.

This sequence mirrors what Sephora uses in some markets. The logic benefits the business, not the customer.

In session data, this often results in users with high intent leaving the flow after cart review. In one of our audits, over 60% of users who reached the account prompt exited without seeing the next step. These were active drop-offs after product commitment.

This behaviour is about the timing. The user doesn’t yet know how the transaction will complete so they hesitate before investing more effort.

CRO Consultants should review login prompt placement relative to shipping transparency. Asking for details too early leads to session loss without error.

3. Size and fit copy doesn’t resolve the real question

Pattern observed in PDPs structured like ASOS

Many fashion PDPs use copy like “true to size” or “model wears size M” without explaining how the garment will feel on the buyer.

ASOS follows this structure. So do many brands we’ve audited. But session data shows replays where users move back and forth between the PDP and size guide repeatedly. 

In one case, over 40% of users opened the size guide more than once, and still left the page. This wasn’t driven by complexity. It was driven by incomplete information.

When we tested revised copy with specific guidance like “fits narrow in the shoulders,” “drops just below waistline” the loop behaviour dropped, and add-to-cart rates increased.

Generic fit copy creates uncertainty. For CRO Experts working in fashion ecommerce, it’s one of the simplest high-impact fixes available.

4. Return sessions show no memory of past behaviour

Pattern observed in multi-session journeys compared with Amazon

In audits of marketplaces and high-SKU catalogues, we often see return sessions with no continuity. Previously viewed items are not shown. Filters must be reapplied. The journey restarts.

Amazon solves this by default. Features like “recently viewed,” “save for later,” and “buy again” keep session memory visible and useful.

When we added persistent modules like “your shortlist” or retained filters in a multi-brand audit, returning users converted faster. The return sessions were shorter but more decisive.

Many ecommerce brands optimise hard for first-session conversion. That ignores how users actually buy. In categories where research is high, memory is more important than urgency.

CRO Agencies working in long-consideration categories should test memory-first design before applying conversion-pressure tactics.

5. Global copy reused across markets lowers relevance

Pattern observed in AU and SEA versions of brands structured like Uniqlo

Uniqlo localises its product copy by market. In Australia, it uses comfort and fit framing. In Japan, it leans technical. In the US, it uses performance and lifestyle language.

In audits for global brands entering APAC, we’ve seen US-origin PDP copy used across AU, SG, and UK markets without adjustment.

In one test, a single product description was localised from “performance cooling mesh with active stretch” to “lightweight, breathable cotton with loose fit.” The change improved add-to-cart rates by 9% for AU mobile users.

The product didn’t change. The copy did.

This is a framing issue. If the decision logic doesn’t reflect the local user’s mental model, conversion slows even when product quality is strong.

Conversion Rate Optimisation Agencies working across markets need to review PDP copy as a regional performance variable, along with CMS asset.

6. Example signals and where they lead

Some of the most consistent conversion issues we see don’t come from major fails they start with subtle behaviours: repeat actions, skipped steps, or short exits after long views. The table below summarises five of the most common signals, what’s likely behind them, and how to respond:

Audit Signals and Opportunities

7. What ecommerce teams measure vs what actually shows friction

Most ecommerce teams measure clickthroughs, bounce rates, and checkout drop-offs. That’s not where friction starts. Friction usually begins one step earlier in the moment a user pauses, loops, or doesn’t trust the page enough to continue.

Monitoring Gaps vs Behavioural Friction

8. Finally,

Most audit issues arent dramatic, they don’t break flows. They lower confidence just enough to stall a session. These mismatches between the business’s logic and the user’s behaviour.

At DIGITXL, we specialise in finding these mismatches through close analysis of how real users act when something feels off.

If you’re trying to scale performance with the same platform and product, this is where you should start. Before you test anything, watch where the user stops.