How Pizza Fusion dropped CPO 38% in 14 days.
A 4-location independent pizza brand in Brooklyn swapped the click destination on its Meta campaigns — same creative, same audience, same budget — and taught the algorithm a new event. Here is what changed, in order.
A click destination is a training signal.
Pizza Fusion did not have an ad problem. It had a measurement problem. Meta optimizes toward whatever event you feed it, and the event the campaign had been trained on for nine months was a Link Click on a website that could not see the order complete on DoorDash. The algorithm was learning what a clicker looked like. Not what a customer looked like.
The team at Pizza Fusion had three delivery platforms, one direct-order page, and a Shopify-hosted website that fired a standard Meta Pixel. When a customer clicked an ad and landed on the site, the Pixel saw them. When that same customer then clicked through to DoorDash to actually order — it did not. For every 100 clicks the campaigns paid for, roughly 72 ended as purchases on third-party platforms where the Pixel was blind.
Meta's algorithm, doing its job, kept pulling audiences that liked to click. Not audiences that liked to buy pizza. The team had tried creative refreshes, audience rotations, and bid cap experiments. Nothing moved cost per order meaningfully below $21. That's where DineRoute entered the flow.
“The moment match quality jumped from 3.1 to 8.4 we stopped running variants and just let Meta cook. It knew what a real Brooklyn pizza customer looked like for the first time since we started running ads.”
The swap took twelve minutes. Paste four platform URLs. Paste the Meta Pixel ID and CAPI access token. Auto-fetch the Pizza Fusion logo. Pick the "slice + route" template. Update the destination URL in Ads Manager. Done. The next 14 days were the test.
- 01
Plant the event
Swapped the Meta campaign destination from pizzafusion.com to pizzafusion.dineroute.menu. Same creative set. Same audience. New click destination, new platform_click event fired server-side with fbc/fbp match data.
- 02
Let Meta relearn
Left the campaign in learning phase for 7 days. No manual bids. No new creative. Let Meta rebuild its audience model against the new event signal instead of the old Link Click signal.
- 03
Open the comparison
Pulled Events Manager match quality from 3.1/10 (Link Click) to 8.4/10 (platform_click) inside the first 72 hours. Meta Ads Manager CPA started tracking real redemption delta against a holdout audience.
- 04
Hold the budget
Did not scale budget during the test window. Pizza Fusion wanted to see the signal isolate without spend being a confounding variable. CPO dropped from $21.40 to $13.20. Same dollars. Better customers.
What the dashboard actually showed.
Cost per order traced downward as soon as the new event signal started training. ROAS followed with an eight-day lag as Meta reallocated impressions toward high-intent lookalikes. No budget increase.
Every restaurant owner who runs Meta ads is already paying for DineRoute. They just have not installed it yet — they are paying Meta to learn from a signal that cannot see the actual order.
Your campaigns are ready for the same swap.
Same creative. Same audience. Same budget. Different click destination. If DineRoute does not move CPO in the first 30 days, the month is on us.