€48 242 in. About half a million out.

Meta was a quarter of the store. That is the whole point.

Over seventeen months Meta reported €140 840 of purchase value on €48 242 of spend. That is 2.92x, and on its own it is an ordinary number. The interesting part is what the other three quarters of the store were doing.

Klaviyo › Business performance

Klaviyo, 1 November to 31 December 2024. Two months out of the eighteen the store traded, and the read behind route A.

Two routes, two currencies, deliberately not added together. Klaviyo reported this store in dollars; Meta and Shopify reported it in euro. Either route puts the store around half a million against €48 242 of advertising.

One platform, one placement band, one age band.

Instagram took 97 percent of the money and returned 3.86x. Facebook took the rest and returned 2.05x. Inside Instagram the budget split almost evenly across three placements, and the return barely moved between them.

Meta Ads export, 21 September 2024 to 12 March 2026. Two charts rather than two axes, so neither scale borrows credibility from the other.

The line that got clicked was not the line that sold.

Every ad in the account, exported at ad level and sorted by spend. Discount hooks won the click. Product names won the sale.

The strongest click through rate in the account, 8.55 percent, belonged to a forty percent discount with an urgency threat, and it returned 1.92x. The strongest return, 16.56x, belonged to a sentence that just named the watch.

So we stopped optimising on clicks.

Across the 22 headlines carrying both a return and a click through rate, on €14 339.79 of spend, the correlation between the two is 0.08. That is no relationship at all, and it is the reason this account was never read on clicks.

Case study › What the copy did

The dashed line is the fitted trend, and it is nearly flat on purpose. This is what no relationship looks like when it is drawn instead of asserted.

The sharpest pair in the account ran on almost identical money. Free shipping returned 7.01 on €370.64. Time is running out returned 0.77 on €387.88. On a €230 watch that is not a copy preference. It is the last objection before checkout getting answered instead of rushed.

Grouped rather than paired, discount and urgency lines returned 5.09 on €2 435.95 against 3.59 on the €9 213.58 behind brand benefit lines. That gap is real, and it is also the gap a Black Friday window exaggerates, which is why we report it without recommending a permanent discount.

Forty one posts bought the audience the ads then rented.

This is the part Meta cannot report. Forty one posts over seventeen months built an Instagram account of 124 000 followers, and the paid account was never asked to find a new audience. It bought more of the one already engaging.

The loop that produced them

Case study › What we did

Organic, paid and email, each allowed to do the one thing it was cheapest at. None of them was asked to carry the others.

Each layer was measured on its own terms. Organic on reach and follower growth, never on attributed revenue, because no platform attributes that honestly. Paid on purchase value divided by spend, at segment level, inside the window the segment actually ran in. Email on Shopify ordered product. We never add attributed revenue across channels, because Meta purchase value and Klaviyo attributed revenue contain the same orders.

A list that sold without being sent.

Across November and December 2024 the store took $174 771.71. Email was credited with $55 825.00 of it, and 72.71% of that arrived from always on flows rather than campaigns. Revenue on days nobody pressed send.

Klaviyo › November

1 to 30 November 2024. $95 823.60, nine percent up on the window before it.

Klaviyo › December

1 to 31 December 2024. $78 948.11, and the peak was already behind us.

Case study › What the store did

The same dashboard says the four percent it was down against was the sixty one days before, so September and October ran higher than the Black Friday window that followed.

Klaviyo reports this brand in US dollars. Meta and Shopify report it in euro, so the two currencies are never added together anywhere on this page.

Margin is what made the return affordable.

A 2.92x blended return is only interesting next to the cost sheet. Watches carried 42 to 58 percent. The small accessories carried 67 to 84 percent, and that is what paid for the testing.

Less traffic, more money, then a clean sale.

In the final window sessions fell 59 percent while sales rose 41 percent and orders rose 50 percent. We were spending less and selling more, which is the only version of a wind down worth writing about.

Shopify › Analytics

Shopify, all channels, 1 December 2025 to 22 March 2026.

The same funnel, tighter on every step

The relaunch itself is the honest counterweight. €4 578.52 of spend returned €9 040.46 of reported purchase value across 43 purchases, a blended 1.97 at €106.48 a purchase. On a €230 entry watch that is a loss on paid traffic alone. The audience that had carried three quarters of the revenue was no longer being fed, and rebuilding it would have taken another eighteen months of posting.

So the brand was wound down deliberately rather than left to drift, and the domain, the catalogue, the audience and the automations were sold together on 23 March 2026 through Flippa, with escrow settled two days later.

Where every number on this page comes from.

One row per figure, naming the system that produced it and the window it covers. The rows we worked out by arithmetic are marked as derived, with the arithmetic written out beside them.

Case study › Provenance

Reported means a platform printed it. Derived means we calculated it, and the sum is shown so you can check it.

Three gaps are named rather than filled. There is no store read between January and November 2025, which is why the four month figure stops at four months and the lifetime figure is a ratio rather than a total. No follower or reach export survived the sale of the account, so those two figures are ours. And no cost of goods sits against an actual order, so the margin table is catalogue gross margin and never a net margin per order.

We publish this page for our own brand because it is the page we would write for yours. A number without a source and a window is a claim, not a measurement.

What this changes about how we read your account.

• Your ad account is not your business. We size the gap between what the platform claims and what the store banked before we touch a budget, because every decision after that depends on it.
• Paid reach is an audience purchase. Spend that looks flat on a seven day window is often buying followers who convert later on a channel with no media cost attached.
• Margin sets the ceiling on testing. We read the cost sheet first. It decides how much losing you can afford, which decides how fast you can learn.