90-Day CRO Sprint — Case Study
Pretty Little Fit had traffic.
The site was losing it quietly.
84% of their visitors arrived on mobile. Less than 1% of them bought anything.
This is not a traffic problem. It is a structural conversion problem.
90-day results
Measured against an identical prior baseline.
The situation
Strong ads. Weak site. The gap in between.
Pretty Little Fit had healthy ad click-through and consistent social traffic. The problem was downstream. Visitors arrived and left without buying. The instinct was to spend more on ads. The right answer was to diagnose first.
Overall conversion rate at baseline. Industry benchmark for this category sits at 2.5%. Every point below that is compounding revenue leakage.
Cart abandonment rate. Three in four shoppers who added to cart did not complete the purchase. That is a checkout path problem, not a traffic problem.
Mobile conversion rate. 84% of traffic arrived on mobile. The site was not built for how its actual audience was shopping.
Average items per order. Coordinated sets existed in the catalogue. Customers were never discovering them.
The diagnosis
Seven friction points. Each one leaking revenue.
Heatmaps, session replays, and exit surveys mapped where visitor intent was collapsing. None of these were design opinions. All of them were observable and testable.
-
01
Social proof below the decision point
HIGHReviews existed and were positive, but not visible where the purchase decision was forming. Visitors looked for validation above the fold, did not find it, and left.
-
02
Size guide buried three taps deep
HIGHExit surveys ranked sizing uncertainty as the second-most cited reason for not purchasing. For a category where fit is the primary barrier, this is a structural trust gap.
-
03
Forced account creation at checkout
HIGHFunnel analysis showed 61% drop-off at account registration, before shipping details were entered. Asking a first-time buyer to commit to an account is friction with no commercial justification.
-
04
Product visualization gap
MEDIUMFlat-lay photography dominated every product page. Customer interviews confirmed uncertainty about fit was a consistent hesitation point before adding to cart.
-
05
No signal to decide now
MEDIUMSessions regularly ended with cart adds followed by exits to think about it. With no low-stock or time signals present, the site was training visitors to delay.
-
06
Add to Cart below fold on mobile
HIGHOn 375px screens, the primary CTA required scrolling to reach. Many sessions ended before a visitor ever saw the button the page was built around.
-
07
No cross-sell path for coordinated sets
MEDIUMAverage order was 1.1 items. The range had coordinated sets designed to be bought together, but no recommendation structure surfaced them at the point of decision.
The approach
Diagnose first. Test everything. Deploy what wins.
Each friction point became a testable hypothesis. Nothing was deployed based on assumption. Every change was validated before rollout.
-
Week 1–2
Diagnostic Setup
Tracking infrastructure deployed. Baseline metrics locked. Exit surveys and user interviews completed before any changes were made.
-
Week 3–4
Hypothesis Development
Each finding converted into a testable hypothesis. Success criteria defined before any variant went live.
-
Week 5–11
Controlled A/B Testing
Tests run to 95% statistical confidence, minimum two-week cycles. Any test showing negative impact was paused immediately.
-
Week 12–13
Implementation
Winning variants deployed to 100% of traffic. Post-rollout regression monitoring active. Next cycle scoped.
The results
90 days of measurement against an identical baseline.
All metrics compared against the 90-day period immediately prior to the engagement. Same traffic sources, same seasonality window, same attribution method.
Session duration increased. Visitors were reading, not bouncing.
Of product page visitors clicked the size guide after inline placement. Previously, most never found it.
Cart recovery email open rate, against a 28% industry average for the category.
The heatmaps made the problems obvious.
People were looking for information we had buried.
What I valued most was the discipline around testing. Nothing changed without evidence first.
We could see exactly what was working and why.
Key learnings
What this engagement confirmed.
Mobile-first is not optional
84% of traffic was mobile. Mobile conversion was 33% lower than desktop. Optimizing for the dominant device had outsized impact on the overall result.
Friction compounds across the funnel
No single barrier was catastrophic on its own. Seven friction points in sequence produced a conversion rate roughly half of benchmark.
Proof placement beats proof volume
Hundreds of genuine reviews existed but were not visible at the decision point. Moving them above the fold produced measurable lift without adding a single new review.
Not every hypothesis wins
Several expected improvements had neutral or negative results in testing. That is exactly why you test before deploying. Assumptions are not a strategy.
Start with a diagnosis
If your traffic is not converting, the answer is in the data.
The Free Scanner surfaces structural signals in your digital presence in under 60 seconds.
No forms. No sales call. Just a clear read on where your funnel is leaking.
Takes under 60 seconds. No credit card required.