Auditing a growing brand for the money it was leaving on the floor

The theme's checkout events were not firing correctly, so parts of the funnel were reporting against incomplete data. I went into the theme and repaired the tracking before drawing a single conclusion from it. That is unglamorous work and it came first, because a recommendation measured on a broken funnel cannot be defended later.

Everything downstream inherited that correction, and every estimate in the report was delivered as a labelled range rather than a single confident number.

Most audits are called in when something is failing. This one was not. Direct orders up 23%, search up 41%, Instagram up 167%. The honest framing in the report's opening note: this is not a diagnosis of what is wrong, it is a look at where a business this healthy is being less efficient than it could be.

That reframing set the whole method. Instead of proposing a redesign, I ranked six changes by effort against return, and told the client plainly which ones took fifteen minutes and which took a quarter.

Paid TikTok traffic grew 127% year over year while orders from it grew 12%.

The cause: 2026 campaigns were pointed at two collection pages converting at 0.007 to 0.02%. In 2024 the same spend landed on the homepage and top product pages, converting 10 to 25 times better.

The fix was a redirect. Fifteen minutes of work, sized at $19K to $38K a year recovered, with a purpose-built TikTok landing page as the second phase.

The strongest number in the audit had nothing to do with acquisition. The top ten chains and bracelets were being returned at 30 to 54%, against a fine jewelry benchmark of 8 to 15%. Total returns ran $320K, 24% of gross.

The cause was diagnostic, not aesthetic: no size guide on chain and bracelet product pages, and no photography showing the same piece on different wrist and neck sizes. Buyers were guessing, and a third to a half of them guessed wrong.

Recommended: size guides with wrist and neck circumference on every chain and bracelet page, "how this fits" comparison photography, and a product-specific FAQ on the ten worst offenders. Halving the return rate on those ten products is $76K to $114K recovered.

96% of Swedish traffic is mobile, and Swedish desktop converts at roughly twice the rate of Swedish mobile on the same products and the same checkout.

The gap was upstream: add to cart at 5.1% mobile against 9.2% desktop, checkout start at 2.6% against 5.6%. So the friction lives in the product page and the cart handoff, not in the payment step.

The PDP recommendations followed from watching that behaviour: users were scrolling up and down repeatedly to reconcile a variant selection with the image above it. Reduce and standardise image height so more decision-critical content sits above the fold, add click-to-expand to compensate, sticky add to cart on mobile, touch-optimised variant selection, remove the redundant mobile back button. And test Swish, Sweden's default mobile payment, at exactly the step where the device gap is widest.

Kejsar's custom work is the reason artists wear the brand. It sat on a single page unreachable from the homepage, the main navigation, or any product page. Anyone who did not already know to look for it never found it.

Recommended: custom enters the primary nav, a homepage section with portfolio pieces, an updated client roster, and a custom block on every product page. I wrote that block in Swedish, in the brand's voice, ready to paste rather than to brief.

Email converted at 1.26%, 27% better than search, 37% better than direct, and roughly 8 times better than social. It was 0.1% of total traffic.

Black Friday data made the case sharper: email converted at 6.41% during the promotional window, five times its annual rate. The recommendation was a full Klaviyo flow build, sized at $71K to $124K depending on how far the list scales.

That seasonal read also drove the sequencing. A 30% lift in May is worth 30% of May. The same lift in November is worth two to three times that. So the roadmap was ordered to land before Q4, not simply by size of prize.

Every figure in the report was delivered as a range with its confidence stated, and the stacked total was discounted for overlap between initiatives rather than summed. The projection is 21 to 43% growth, not a single number, because a single number would have been more persuasive and less true.

The engagement was priced and reported in Swedish kronor. Figures are converted here at 10.5 SEK to the dollar and rounded to the nearest thousand. The ranges are the report's; the rounding is mine.