The scenario
The store carries two lines on the same shelf: licensed niche fragrances imported under agreement, and its own in-house line developed to sit near well-known profiles at a fraction of the price. Both sell to the same Kuwaiti audience, on the same Shopify theme, at price points from a modest discovery set up to a ninety-dinar bottle, with KNET, Tabby and cash on delivery at checkout. A five-person team splits sourcing, content and customer service, and most of the traffic since the brand's early days has come from Snapchat, with Google picking up anyone searching a specific note or name directly.
The founder had always assumed the two lines needed the same page template, since both are perfume. What that assumption missed was that a shopper landing on the in-house line arrives with a different question in her head than one landing on the licensed import, and the store's product pages answered neither question until she had already scrolled well past the point most mobile sessions give up.
- Monthly revenue band
- pending client sign-off
- Average order value
- pending client sign-off
- Fulfilment
- Own stock, local courier, next-day across Kuwait
- Team
- Five people across sourcing, content and customer service
A store that could sell to a buyer, and quietly lost every gift-giver
The blended conversion rate looked survivable, which is exactly why nobody had gone looking further. Split by scent line, the picture changed: sessions that landed on an in-house product page converted at roughly half the rate of sessions landing on a licensed import, and the gap widened sharply in the two weeks before Eid and National Day, the exact weeks carrying the most traffic all year.
Session recordings on the highest-traffic product pages in Arabic showed the same behaviour on both lines but for opposite reasons. On the in-house pages, shoppers scrolled up and down between the photo and the price several times, apparently trying to work out whether they were looking at the original or the store's own version — a disclosure that existed, but three screens down, in a formal Arabic sentence nobody was reading that far to reach. On both lines, customers who did reach checkout during gifting weeks routinely abandoned the form, then reopened WhatsApp to ask whether the order could ship to a different address with a card attached, because the checkout itself had never asked.
Two entirely different products were being sold through one page built for neither. The store had a checkout for a woman buying a bottle for herself, and most of its traffic, especially in the two seasons that mattered most, was someone buying for somebody else.
What we did — the pages and offers we tested
The losers are here on purpose. A test with only winners was never a test.
A plain-language line label at the top of every product page
Winner"Licensed import" or "Our own line, inspired by" shown as a badge above the price, not in the description
- Format:
- Product page badge, single-variable test against the buried disclosure
The disclosure had always existed; it just sat below the point most mobile sessions gave up. Moving it into plain Kuwaiti Arabic above the fold turned a source of suspicion into a stated fact the shopper could decide about immediately, and it lifted conversion on the in-house line specifically, without touching the licensed pages at all.
A gift mode at checkout: different address, a note, wrap
Winner"Sending this as a gift?" toggle that opens a recipient name, address and card message field
- Format:
- Checkout step addition, tested against the single-address form
Most of the WhatsApp messages arriving after checkout were some version of this request, meaning the store's own customer-service line had been quietly running a manual gift service nobody had designed. Building it into the form removed the manual step, cut the number of orders that stalled between payment and confirmation, and made the cash-on-delivery fee visible to the recipient, not the sender, which is the person who should be deciding whether to pay it.
A locked discovery-set price held for the full bottle, offered right after unboxing
Winner"Loved one from your set? Get the full bottle at the set price for the next two weeks."
- Format:
- Post-purchase page and follow-up message, discovery-set buyers only
The store had never connected its own discovery set to a second order; it was sold as a standalone product and left there. Timing the full-bottle offer to arrive once the customer had actually worn a scent for a few days, rather than the day the set shipped, caught the decision at the moment she had an answer instead of a guess, and turned the set from a low-margin trial item into the front door of the funnel it was always meant to be.
A forced bundle: discovery set plus a full bottle in one checkout, at a discount
Lost"Try and commit together, save 15%"
- Format:
- Product page bundle offer replacing the standalone discovery set for two weeks
A reasonable idea that read the buyer backwards. The entire point of a discovery set is that the customer has not decided yet, and asking her to commit to a full bottle in the same checkout put the decision back before the reason it existed — actually smelling the thing — had happened. It suppressed discovery-set sales without lifting full-bottle sales enough to justify the discount, and it was the clearest instructive loss of the round.
A full-screen occasion picker on landing: Eid, National Day, wedding or 'just for me'
Lost"Who is this for?" shown before any product on the homepage
- Format:
- Full-screen modal on homepage entry, mobile
The idea behind it was right — most visits are occasion-driven — but asking the question before showing a single product added a step nobody wanted on a phone with limited patience, and it measurably raised the bounce rate on paid traffic that had already told the ad which occasion it was responding to. The same segmentation worked far better placed inside the cart and checkout, where the customer had already committed to browsing.
What we did — the optimizations, in order
Read the funnel by scent line and traffic source
We split every funnel step by whether the landing page was a licensed import or the in-house line, and again by whether the session arrived from Snapchat, Instagram or a branded Google search.
Why: A blended rate had been hiding a healthy licensed-import funnel sitting on top of an in-house line converting at half the rate, and averaging the two told the founder the store was merely average rather than pointing at exactly which pages needed the work.
Watch Arabic mobile sessions on both lines
We watched recorded sessions on the highest-traffic product pages for each line, in the Arabic locale on the same phone models the customer base uses, rather than testing the theme on a desktop browser.
Why: The scrolling back and forth between photo and price, and the WhatsApp messages arriving after an abandoned checkout, only became visible as a pattern once we watched real sessions rather than reading a bounce-rate number in a dashboard.
Fix the floor: speed, right-to-left, payment and authenticity visibility
Compressed product images shot for print rather than web, corrected right-to-left spacing on the scent-note pills, and moved the KNET, Tabby and cash-on-delivery badges plus the line label above the fold on every product page.
Why: These fixes touch every visitor before any test runs, which is why they come first. A slow page with a buried authenticity label was suppressing conversion on both lines regardless of which offer test ran on top of it.
Run one offer or page test at a time, per line
The line label, the forced bundle and the occasion modal were each tested individually against the existing page, holding the audience, the price and the ad creative identical for the duration.
Why: Testing the bundle and the label change at the same time would have made it impossible to say which one caused the drop when the bundle failed, and separating them is the only reason the failure could be attributed correctly rather than blamed on the wrong change.
Rebuild checkout around the gift-giver, not just the buyer
A gift toggle that opens a separate recipient name, address and card-message field, cash-on-delivery fee logic that follows whoever receives the parcel, and a WhatsApp fallback for anyone who stalls mid-form during gifting weeks.
Why: The store had been running a manual gift service through customer service without ever admitting it in the design. Building the actual workflow into checkout removed a step that was quietly costing an order every time a customer decided messaging was too much effort and left instead.
Time the full-bottle offer to when the scent has actually been worn, and keep a weekly test log
The full-bottle offer for discovery-set buyers moved to a few days after delivery instead of the day it shipped, and every test — including the two that lost — went into one shared log the founder reviews weekly.
Why: An offer that arrives before the customer has an opinion is a guess she has to make twice, and a written log is the only reason the team will not re-test the same forced bundle again next Eid because nobody remembered it already failed.
What changed
The table above carries the numbers once the client signs them off, and the shape worth noting now is that the improvement was concentrated exactly where the split had pointed: the in-house line and the gifting weeks. Checkout completion moved because the form finally admitted that most of the traffic in those weeks was not the person paying, and the line label moved conversion because it answered the authenticity question before it turned into a doubt.
The discovery-to-full-bottle path mattered more than a single test result suggests. Once it existed, it gave the store a second, cheaper conversion event to optimise — a set buyer choosing to commit — rather than treating every visitor as a cold decision on a ninety-dinar bottle, and it is the piece the founder now considers permanent rather than seasonal.
The two losing tests earned their place in the log. The forced bundle confirmed that a discovery set only works if the customer is still allowed to be undecided, and the occasion modal confirmed that segmentation belongs inside the funnel a shopper has already entered, not in front of the door.
What we would do next
Take the gift toggle into the WhatsApp order flow the customer-service line still handles by hand for shoppers who never finish the online checkout, since that is the same fix currently paying for itself only on one channel.
Second, get the gift-mode checkout and the line labelling fully stress-tested a month before Eid and National Day, when gifting volume multiplies and a checkout that still leaks a little at normal volume leaks a lot at peak volume.