The scenario
The store carries coffee machines, air fryers, cookware and serveware under one catalogue, with prices ranging from a ten-dinar utensil to a machine well into instalment territory. It built its following on short demo clips — the pour, the crisp, the pour again — and by the time we started, most of its traffic arrived already convinced a specific product would work, only to lose momentum somewhere between the product page and an order.
The store had grown fast enough that the checkout had never been rebuilt for the range of baskets it now had to handle: a single ten-dinar tray, a full majlis serveware set, and a two-hundred-and-fifty-dinar machine, all moving through the same one-size checkout flow built two owners of the theme ago.
- Monthly revenue band
- pending client sign-off
- Average order value
- pending client sign-off
- Fulfilment
- Own warehouse, local courier for small items, scheduled delivery slots for bulky serveware sets
- Team
- Owner, a warehouse hire and one customer-service line on WhatsApp
A checkout built for one basket, carrying three very different ones
The account looked healthy on the surface: strong demo-video traffic, a Google Shopping feed that caught anyone searching an appliance by name, and a product page that held attention past the norm for the category. Nobody had split conversion by product price band, and once we did, the picture stopped being one number and became three problems.
The small-item baskets converted close to expectation. The higher-ticket machines dropped hard at the final checkout step, because the Tabby instalment figure only appeared on the payment page, four steps after the customer had already done the mental maths on the sticker price and half-decided against it. The full serveware sets, the highest-margin basket in the catalogue, dropped even earlier — recordings showed shoppers checking the product page, the reviews, and then leaving to search whether the set would actually arrive before the gathering it was bought for, a question the page never answered.
None of this was a traffic problem or a pricing problem. The store had one checkout trying to serve a ten-dinar tray, a full hosting set and a quarter-of-a-thousand-dinar machine identically, and each of the three needed a different piece of reassurance at a different moment.
What we did — the pages and offers we tested
The losers are here on purpose. A test with only winners was never a test.
Tabby instalment moved from the payment step to the price block
Winner"From KD 20.83/month with Tabby" shown next to the price on the machine's product page
- Format:
- Product page price block, single-variable test against the control
The instalment figure already existed; it just lived on the step furthest from where the price anxiety actually starts. Moving it under the two-hundred-and-fifty-dinar price tag, where the shopper first does the mental maths, cut the drop between product page and cart on every machine in the range.
A delivery-date promise on the serveware and majlis-set pages
Winner"Order by Thursday, delivered before your gathering."
- Format:
- Sticky banner on serveware-set product pages, date calculated from the cart's current contents
The recordings had shown shoppers leaving the page to search for a delivery answer nobody gave them. Stating the date on the page itself, rather than leaving it to a WhatsApp question after the fact, removed the single biggest reason a full-cart session for a set ended without an order.
Local warranty and spare-parts badge on every appliance page
Winner"Two-year local warranty. Spare parts and service in Kuwait."
- Format:
- Icon row under the add-to-cart button, appliances only
The warranty terms had always lived in a footer page nobody opened. Surfacing it under the button, where the buy decision actually gets made, cut abandonment among first-time visitors comparing this store's machines against the appliance aisle of a large retailer.
A countdown-timer urgency banner on the flagship machine
Lost"Offer ends in 04:12:33."
- Format:
- Product page banner, resets on each visit
It lost trust rather than urgency. A shopper researching a quarter-of-a-thousand-dinar appliance is deciding on warranty and reliability, not chasing a clock, and a timer that visibly reset on a repeat visit read as manipulation and measurably hurt checkout completion on that page specifically.
Raising the free-delivery threshold to lift basket size
Lost"Free delivery over KD 40" replacing the previous KD 25 threshold
- Format:
- Cart page banner and checkout summary, single-variable test
A reasonable idea that misread the basket. Small-item shoppers buying one tray or one utensil rarely added a second item to clear the higher bar; they mostly just paid the now-visible delivery fee or left, and it dragged down completion on the exact segment that had been converting best.
What we did — the optimizations, in order
Read the funnel by product price band, not one blended number
We split every funnel step by whether the basket held a small item, a full serveware set or a machine, instead of reading one conversion rate for the whole catalogue.
Why: A blended rate had been hiding three separate stores' worth of behaviour inside one number, and it told the owner nothing specific was wrong. Splitting the read turned a vague sense of leakage into three named, fixable moments.
Sit through recordings on the highest-ticket pages
We watched real sessions on the flagship machine pages and the serveware-set pages specifically, rather than a random sample across the whole store.
Why: The delivery-date search and the four-steps-away Tabby figure do not show up as a single dashboard number — they show up as a pattern of hesitation only a recorded session reveals. This is where the two real, named problems came from, not a guess.
Fix the floor: speed, warranty visibility, delivery promise
Compressed the demo-video thumbnails and product images, put the two-year warranty and spare-parts badge under every appliance's buy button, and added a calculated delivery date to every serveware-set page.
Why: These touch every visitor regardless of which later test wins, which is why they came first. A slow page with a hidden warranty was suppressing conversion on every price band at once, no matter how good the traffic was.
Run one price-block or delivery test at a time, per template
The Tabby placement, the delivery-date banner and the free-delivery threshold were each tested alone against the existing page, with the audience, the price and the ad held identical across the test.
Why: Two live tests on the same template at once is how a store ends up with a result nobody can explain. Testing the threshold and the Tabby placement separately is the only reason we can say with confidence which one helped and which one hurt.
Rebuild checkout so each basket type sees its own reassurance
The checkout now surfaces the warranty badge and Tabby summary for appliance baskets, the delivery date for serveware-set baskets, and a lighter single-page flow for small-item baskets rather than one form for all three.
Why: A ten-dinar tray and a two-hundred-and-fifty-dinar machine are not the same purchase decision, and asking both customers to click through the same steps means one of them is always seeing reassurance that does not apply to them and missing the one that does.
Time the review request after the item has actually been used
For appliances, the review request moved from delivery day to roughly ten days later, timed to land after the customer has cooked or brewed with the product rather than only unboxed it.
Why: A review asked for on delivery day is about the packaging and the courier, not the machine, and it reads as thin to the next shopper deciding whether the price is worth it. Reviews collected later were longer and more specific about the actual result.
Keep a weekly test log by basket type
Every test, its basket type, its result and the decision taken went into one shared log the owner reviews weekly, including the two that lost.
Why: Without a written log split by basket type, a fix that helped machines and hurt small items looks like a wash on the blended number, and the store re-tries the same threshold change again next season.
What changed
The table above carries the numbers once the client signs them off, and the shape worth flagging now is that the gain did not come from one dramatic fix. It came from splitting one checkout's problem into three, because the small-item, serveware-set and machine baskets each needed a different piece of reassurance surfaced at a different point in the page.
The delivery-date banner mattered more than expected, because it was the one fix aimed at the highest-margin basket in the catalogue. Session recordings before the change showed shoppers leaving mid-decision to search for an answer the page should have given them; after it, that specific exit pattern largely disappeared.
The two losing tests earned their place in the log. The countdown timer confirmed that urgency framing damages trust on a high-consideration purchase, and the raised free-delivery threshold confirmed that this store's small-item shoppers are not there to build a bigger basket — they came for one thing, and asking them to add a second item just made the delivery fee visible and resented.
What we would do next
Take the same split-by-basket-type logic into the WhatsApp order flow the customer-service line still runs by hand for shoppers who never finish checkout, since the same three reassurances apply there too.
Second, get the delivery-date banner live on every gifting and hosting page a month before Ramadan and Eid rather than mid-season, since a bulky-item promise made after the rush has already started is a promise nobody trusts.