The scenario
The brand makes personalised name necklaces, bracelets and small gift sets, sold direct to consumer on its own Shopify store with KNET, Tabby and cash on delivery at checkout. Every product carries a customisation step — a name, a date, sometimes a short phrase in Arabic script — before it goes into production, which makes it structurally different from a jewellery store that ships from stock.
The founder built the store around Instagram and Snapchat discovery, and most of the traffic arrives already knowing what occasion it is buying for: Eid, a graduation, Mother's Day on the twenty-first of March, an anniversary. The order is rarely impulsive. It is a gift with a deadline attached, and the checkout had never been built around that deadline.
- Monthly revenue band
- pending client sign-off
- Average order value
- pending client sign-off
- Fulfilment
- In-house engraving, gift-box packaging, next-day courier across Kuwait
- Team
- Founder, one engraver, one customer-service hire on WhatsApp
The personalisation was the product. It was also where the customer got stuck
Traffic ahead of Eid and Mother's Day was strong, and the product page held attention well: shoppers were spending real time typing names into the preview tool and watching the necklace update. The number that did not match that interest was how many of those sessions turned into a completed order, and it was worse on mobile and worse again in Arabic, where most of the personalised gift orders originated.
Session recordings on the checkout explained why. The name and date entered on the product page had to be retyped into a separate field at the very last step, and a mistake there meant an engraved piece with the wrong spelling. The engraving lead time — three to five working days before a piece even ships — appeared nowhere until an order confirmation email the customer only saw after paying, by which point a gift needed in two days for Mother's Day was already going to be late. Arabic name previews, when the customer chose Arabic script, rendered the letters disconnected rather than joined, which is not how anyone writes their own name and made the whole tool look broken rather than premium.
None of this showed up as a targeting or a budget problem, because the ads and the product page were both doing their job. The checkout was asking a gift buyer to gamble on a deadline it never told her about, on a spelling it made hard to verify.
What we did — the pages and offers we tested
The losers are here on purpose. A test with only winners was never a test.
The name and date preview locked in on the product page, not re-typed at checkout
WinnerThe exact spelling the customer typed and confirmed carries straight through to the order, with no second entry field
- Format:
- Product page personalisation step, carried through cart and checkout as read-only
Every retyping step is a chance to introduce the exact spelling mistake this product cannot recover from once it is engraved. Locking the confirmed name in from the product page removed the single largest source of checkout abandonment and cut spelling-related engraving errors, which had been costing the brand a remake on its own margin every time.
A delivery-by-date calculator shown against the engraving lead time
Winner"Order by Wednesday to receive it before Eid" calculated live from the current date
- Format:
- Banner on the product page and cart, recalculated daily against the occasion calendar
This answered the one question a gift buyer actually has before she will commit: will it get here in time. Showing the real cut-off date rather than a generic shipping estimate moved the decision earlier in the session and reduced the volume of "will this arrive by Thursday" messages the customer-service line was fielding by hand.
Arabic name preview rebuilt with connected calligraphy
WinnerThe Arabic preview font switched from a disconnected default to a joined-letter engraving typeface
- Format:
- Font and rendering fix inside the existing personalisation tool, no layout change
A name that renders with disconnected letters does not read as a preview problem to a customer, it reads as a sign the brand does not know how to write Arabic properly, which is a trust break in a product whose entire pitch is personal. Fixing the typeface lifted completion specifically among customers who chose Arabic script, the majority of the gift-order segment.
The cash-on-delivery fee and gift-note option shown before the final step, not after
WinnerFee and gift-note checkbox moved from a final-step surprise to the cart summary
- Format:
- Cart page summary, single-variable test
A customer who has already decided this is a gift wants to add a note and see the true total before she reaches for her card, not after. Surfacing both earlier reduced the number of carts abandoned at the final step over a fee or an option the customer had not planned for.
A storewide urgency countdown timer on every product page
Lost"Only 2 hours left at this price" shown on every page regardless of any actual offer
- Format:
- Sitewide countdown banner, storewide app installation
This is the loser worth keeping in the record. A generic scarcity timer on a considered, sentimental gift purchase read as manipulative rather than motivating, and because it reset itself on every visit, returning customers noticed it never actually expired. It measurably increased bounce on the product page without moving completion, and it was pulled within the first test week.
What we did — the optimizations, in order
Read the funnel by device, language and occasion
We split product view, personalisation completion, cart and checkout by device and locale, and separately tagged sessions that landed from an Eid or Mother's Day campaign against evergreen traffic.
Why: A blended conversion rate was hiding that occasion-driven Arabic mobile sessions, the highest-intent traffic the brand had, were the ones falling away hardest. Splitting the read turned a vague sense that Eid traffic under-performed into three specific, fixable moments in the funnel.
Sit through session recordings on the personalisation flow, on a phone
We watched real recorded sessions from the name-entry step through to order confirmation, in both languages, on the same phone models the customer base actually uses.
Why: The retyped-name field, the missing lead time and the broken Arabic font do not appear in an analytics dashboard as a single number — they appear as hesitation, backspacing and abandonment that only a recorded session makes visible. This is where the three real problems were found, not guessed at.
Fix the floor: carry the confirmed name through, show the lead time everywhere
Locked the confirmed personalisation through cart and checkout as read-only, put the engraving lead time and a live delivery-by-date calculation on the product page and the cart, and fixed the Arabic preview typeface.
Why: These touch every personalised order regardless of which later test wins, which is why they came before any page test rather than after. A checkout that hid its own deadline was suppressing completion on every single campaign the paid team ran into it, however good the ad.
Run one page or offer test at a time, per template
The delivery-date calculator, the cart-level fee and gift-note visibility, and the urgency timer were each tested one at a time against the existing page, with the audience, the ad and the price held identical.
Why: Two live tests on the same template at once produces a result nobody can trust. Testing the timer separately from the fee-visibility change is the only reason the team can say with confidence which one helped and which one actively hurt.
Rebuild checkout around the gift buyer's actual worry
Phone number as the first field, the confirmed personalisation shown as a locked summary card, the delivery-by-date promise repeated at the final step, and a WhatsApp link for anyone who stalls partway through.
Why: A gift buyer who can see exactly what she confirmed and exactly when it arrives has nothing left to gamble on, which is what actually earns the final click on a non-refundable, engraved order. Repeating the promise rather than stating it once is what made it register under time pressure.
Time the review request after the gift has been given, not shipped
The review request moved from the shipping date to a date timed against the occasion itself — the day after Eid or Mother's Day — rather than a fixed number of days after the order.
Why: A review asked for before the gift has even been given is a review about packaging, not about whether the recipient liked it, and it reads thin to the next gift buyer deciding whether to trust the brand with something this personal.
Keep a weekly test log through the occasion calendar
Every test, its hypothesis, its result and the decision went into one shared log reviewed weekly, including the countdown timer that lost, tagged against which occasion window it ran in.
Why: A small team without a written log re-runs the same urgency-timer idea again before next Eid because nobody remembers it hurt completion the first time, and pays the same lesson twice during the busiest weeks of the year.
What changed
The table above carries the figures once the client signs them off, and the shape worth noting now is that the lift did not come from a discount or a new campaign — it came from removing three specific sources of doubt from a checkout the traffic was already reaching. Checkout completion moved because the gift buyer could finally see the deadline and the spelling she had already confirmed.
The Arabic-typeface fix mattered more than its size suggested. Because it touched every Arabic-script order rather than one test variant, it lifted the baseline every later test was measured against, and it disproportionately helped the segment carrying the most gifting traffic around Eid and Mother's Day.
The countdown timer was worth running even though it lost. It confirmed that this category's buyer responds to certainty about a real deadline, not manufactured urgency about a fake one, which changed what the team tests next: more specific delivery promises, not more pressure.
What we would do next
Bring the same locked-name, visible-lead-time logic to the WhatsApp order flow the customer-service line still handles by hand for shoppers who message directly instead of finishing checkout online — right now that channel repeats the exact retyping risk the website just removed.
Second, put the delivery-by-date calculator in front of the paid team a month before each occasion rather than after the campaign launches, so the cut-off dates shown in ads and on the site never contradict each other during the highest-traffic week of the season.