Cross-border storefront rebuild
A regional beauty group was winning the click and losing the sale. We rebuilt the storefront around the one screen that decides everything.
Traffic was fine. The last screen was not.
The group sold across nine Asian markets from a single storefront that had been extended market by market for four years. Paid media was performing. Sessions were up year on year. Revenue was flat.
The analytics pointed at the checkout, but not at a single culprit. Shoppers in three markets were dropping at the address step. Shoppers in two more were dropping the moment shipping cost appeared. On mobile, roughly a third never reached the payment screen at all.
The underlying cause was the same everywhere: a checkout designed for one market, patched eight times for the others.
Four changes that mattered.
There were about forty changes in total. These four account for most of the result.
- 01
Honest cost, earlier
Duty, tax and shipping were calculated and shown on the product page rather than revealed at the final step. Conversion on the product page fell slightly. Completed orders rose sharply, because the people who continued had already accepted the price.
- 02
Addresses that fit the market
One address form had been serving nine countries. We built market-specific formats with the right field order, the right validation and no mandatory fields that do not exist locally.
- 03
Local payment methods
Card-first checkout in markets where cards are not the habit. We added the wallets and bank transfer options each market actually uses, and put them above the fold rather than under a "more options" link.
- 04
A checkout that survives a bad connection
The old flow lost everything on a dropped connection. The rebuild saves state at each step, so a shopper on the MTR who loses signal comes back to where they were instead of an empty basket.
What changed.
Measured over the first full quarter after launch against the same quarter the previous year, with media spend held roughly flat. Checkout abandonment fell 34 per cent. Mobile conversion rate slightly more than doubled. Average order value moved very little, which was expected — this was a completion problem, not a basket-size one.
They rebuilt the checkout in six weeks and stopped arguing with us about whose fault the drop-off was.
What we would do differently.
We spent the first three weeks redesigning the product page because that is where the traffic landed. It was the wrong place to start. The analytics had already told us the checkout was the constraint and we let the visible surface distract us.
We also underestimated the payment integration work by about ten days. That did not move the launch date, but it did cost the client's engineering team a weekend they should not have had to give up, and we should have flagged the risk earlier.
Back to all work