Removed forced account creation
Hypothesis: New customers abandon rather than create an account before a first purchase.
+11.4%Kept
Checkout completions, measured over three weeks. Kept and rolled out.
Engineering & delivery
Your checkout is where the money leaks out
Data-led analysis, A/B testing and targeted redesigns that cut abandonment and get more baskets over the line.
Including the one that failed
Four tests, and we only kept two
Every agency shows you its winners. That tells you nothing, because you are only seeing the ones that worked. Here is a representative run of tests with the loss and the flat result left in.
Added address autocomplete
Hypothesis: Typing a full address on mobile is where the form loses people.
+6.2%KeptLargest effect on mobile, negligible on desktop. Kept.
Moved to a single-page checkout
Hypothesis: Fewer steps should mean fewer places to drop out.
-3.1%RevertedThe longer single page read as more work, not less. Reverted after two weeks. We reported it as a loss.
Added a trust badge row at payment
Hypothesis: Reassurance at the payment step reduces last-moment hesitation.
No changeRemovedWithin the margin of error. Removed, because unmeasured clutter is still clutter.
Why we show you the loss
Because the alternative is a page of wins that proves nothing. Half of well-run tests do not beat the original, and an agency claiming otherwise is either calling tests early or only telling you about the good ones. What you are buying is a method that can tell the difference, and the discipline to revert when we are wrong.
Figures are representative of engagements of this kind rather than a single client, shown to illustrate how results get reported.
What we do
Checkout optimisation capabilities
Checkout audit
A step-by-step read of where people leave, and what it is costing you per year.
A/B testing
Properly powered tests, run long enough to mean something, reported honestly either way.
Abandonment reduction
Guest checkout, address lookup, saved baskets. The unglamorous changes that move the number.
Mobile checkout
Where most abandonment happens and where most redesigns still start from a desktop mock.
Speed
Every second of checkout load costs conversions. We treat performance as a revenue task.
Trust signals
Delivery cost, returns and security shown where people actually hesitate, not buried in a footer.
How we work
The method is the product
Anyone can suggest checkout changes. The value is in knowing which ones actually worked, which needs tests run properly and reported honestly.
See the full funnel breakdown- 01
Measure what you have
A full funnel read before touching anything, so there is a baseline to argue against later.
- 02
Size the prize
Each fix gets an estimated annual value. That decides the order, not which is most interesting to build.
- 03
Change one thing
Tests that move several things at once cannot tell you which one worked.
- 04
Run it properly
Long enough to reach significance. Calling a test early is how agencies manufacture wins.
- 05
Keep or revert, and say which
Losses get reported as losses. You are paying for the method, not for good news.