Two premium wellness brands, run under one retainer. Both looked healthy. Click-through rates were strong, add-to-carts were climbing, clicks cost under a dollar. Underneath, about a quarter of the monthly budget was going to a placement that had never produced a single sale.
Both brands sell premium wellness products in Canada. One is a spa skincare range, the other an aromatherapy and bath line. We run Meta for both under a single retainer.
Neither account looked broken. Add-to-carts were healthy on both. Click-through rates were strong. Cost per click sat under a dollar. On the metrics most dashboards lead with, everything read fine.
The thing that did not read fine was purchases. Both accounts were producing carts and almost no sales, and nothing in the top-line numbers explained why.
We split the last 30 days by placement rather than by campaign. One line stood out on both accounts.
Audience Network had the best click-through rate on both accounts, by a factor of four. It had the cheapest clicks. On the skincare brand it drove more add-to-carts than Facebook and Instagram combined. It had also produced zero purchases on either account.
On the aromatherapy account the same placement was taking 38% of spend, with the same result.
A placement with a great click-through rate and no purchases is not underperforming. It is buying a different thing than you think it is. Audience Network inventory sits inside third-party apps and games, where taps are cheap, frequent and often accidental. That inflates exactly the metrics most reporting leads with, and it pulls blended cost per click down at the same time, so the account looks more efficient the more you spend there.
We removed Audience Network from all six ad sets across both accounts and left every other targeting parameter identical, so the change could be read cleanly against the previous period.
The same placement and demographic split surfaced a second thing. Both brands sell to a much older customer than their targeting assumed. On the skincare account, 65+ had the best click-through rate and the lowest cost per click. The 25 to 34 bracket had spent close to $200 and produced nothing. Roughly 83% of spend and nearly all purchases were female.
We also flagged the deeper issue. Neither account had the Conversions API live, so Meta was learning from a partial picture of what happened after the click. That is the next fix, and it is the one that makes every number on this page more reliable.
Across August month to date, the skincare brand did $10,359 in total sales, up 71% year on year, on 38 orders, up 111%. Sessions were up 107%, driven by Facebook and Instagram, though last-click attribution credited most of the revenue to email, search and direct.
To be straight about cause: a three-day 25% sale ran inside that window and is the clear peak in the daily curve. The sale drove the revenue lift. The placement fix is a structural change whose value is the roughly $1,050 a month no longer going to inventory that never converted, and a cleaner signal for Meta to learn from.
The aromatherapy brand grew orders 38% over the same period on a softer month, where a lower average order value held revenue back.
Both brands run designed statics built for a 55 to 65+ reader. Larger type, one clear offer, product and ritual rather than ingredient science. Client marks removed and label copy softened below, since both are active clients.
No. It works for some advertisers, particularly app installs and low-consideration products. The point is to judge it on purchases rather than on click-through rate or add-to-carts, because those are the two metrics it reliably inflates. Split by placement and look at what actually converted.
An accidental tap inside a game can still fire a view-content or an add-to-cart event. What it almost never does is come back and buy. That is why the placement looked efficient on every upstream metric and produced nothing at the bottom.
Roughly $1,050 a month across the two accounts, based on the 30 days before the change. That budget moved to Facebook and Instagram, which were the only placements producing purchases on either account.
Neither account has the Conversions API live, and both stores use last-click attribution in Shopify, so Meta sees only part of what happens after the click. We would not present in-platform ROAS as a performance verdict until that is fixed. It is the current priority on both accounts.
We'll split your account by placement and show you what converted and what only looked like it did.
Get a free audit