Zero purchases in 34 days became 68 in the next 17
A US direct-to-consumer kitchenware brand on Shopify
The Brief
A new store, two product lines, and a pixel that had never recorded a sale.
Two things were broken at launch, and either one alone defeats purchase optimization. The pixel was cold, so Meta had no purchase pattern to aim at. And the Shop checkout was misconfigured, so buyers often could not complete an order.
The account was pointed at purchase optimization anyway, and had zero sales to show for it.
The Approach
Treat the first month as a sequenced build, not a scaling exercise.
Optimize for the deepest event the account could actually produce — add to cart — and use the cold-start weeks to build a conversion signal cheaply while the store was repaired and the creative and audiences were learned.
Hold purchase optimization back until three things were true at once: a checkout that could complete a sale, a pixel warm enough to exit learning, and evidence of which audiences and creative to trust. Then flip to purchase in one deliberate move.
What we Did
Phase one, build the signal. 17 June to 20 July
Add-to-cart-optimized prospecting plus a broad top-of-funnel layer, held to roughly $118 a day. It produced 166 adds to cart and 66 checkouts on $4,030.57, and zero purchases, because the store could not close them. We named the checkout break as the binding constraint and fixed it before spending more.
The enabling event. 21 July
The Shop destination was fixed, so a purchase could complete for the first time.
Phase two, convert. 21 July to 6 August
With the store working and the pixel warm, we launched four purchase-optimized campaigns on the audiences and creative that had won in phase one, and throttled top of funnel to almost nothing to fund them.
Results in Detail
The store fix was the enabling event, and we say so first. Purchases went from 0 across 34 days to 68 across the next 17. The 21 July checkout repair is the single largest reason a sale could complete at all. No media decision alone produced that jump.
What the media built was the readiness: a warm pixel, a proven audience set, and a winning creative, so the account converted on day one of phase two instead of restarting a three-week learning phase.
The full funnel, before and after. Every stage improved, but the base is the story.
The wait was spent building signal, not idling. Cost per add to cart fell 61%, from $24.28 to $9.55. Cost per checkout fell 40%, from $61.07 to $36.40.
These sit underneath the purchase number and are the cleanest evidence of the media improving, because they are barely entangled with the store fix. The same audiences, better taught, got cheaper at every depth.
Cost per event, before and after. Rate metrics, unaffected by the different window lengths.
Budget was sequenced: build the pool early, then move the money. Top of funnel took 40% of spend in phase one to build a retargeting pool at an $8.04 CPM, then dropped to under 4% once the pool existed.
Over-funding a pool you have already built is a common way to waste a third of a budget. The timing of the cut is the lever.
Share of spend by campaign role. Top of funnel fell from 40% to under 4% once the pool was built.
Video beat static decisively, and the gap widened with scale. At scale the winning video reached $3.41 per cart at a 10.13% click-through, while static stayed stuck at $18.11.
Head to head on the same event at comparable budget, video produced adds to cart at $10.70 against static’s $18.21 in phase one. Video got cheaper as it scaled; static did not move. The creative budget consolidated onto video.
Creative at scale, add-to-cart phase. Same product, same offer, same weeks.
Audience testing found the winners, and a product truth. The glass line converts at 2.28x blended return on ad spend, the silicone line at 0.80x, same account and same weeks.
Interest testing in phase one identified which audiences to trust, and the purchase campaigns were built on them rather than on guesses. Product-audience fit now drives budget allocation, not intuition.
The messaging shift aligned the creative with what converted. The creative moved from recipe framing to a materials-safety and clean-living angle, and the ad carrying that message is the same asset that scaled to $3.41 per cart.
We call this strategic alignment, not a measured lift, because it was not run as a controlled test against the old message. The winning creative and the new positioning are simply the same asset, leaned into deliberately.
Purchase optimization, once switched on, worked immediately. Four campaigns ran $3,462.55 for 65 purchases and $5,033.58 in first-attribution value: a 1.45x blended return in 17 days from a standing start, best campaign 2.57x.
CPM nearly doubled to $34.95, and correctly so. Purchase optimization buys scarcer, higher-intent impressions, so price per thousand rises while cost per outcome falls.
The honest attribution
- Six things changed across these windows, entangled by design, because sequencing them together was the strategy.
- We do not claim the purchase switch alone produced 68 sales. The store fix is the larger driver of the raw count.
- The cleanly isolated wins are the single-window ones: video versus static, the cost-per-event cuts, and the glass-versus-silicone split.
- The defensible claim is narrower: the account was sequenced so that the day the store worked, every lever was already in position, and purchase optimization converted on its first day rather than its twentieth.
Tools & technologies
What the build runs on
verified