Super Bowl: What Didn’t Work — Dino Albizzati
← Back to all case studies
Meta Ads · Email · Local/In-person · Postmortem
The one to show when someone asks about a campaign that failed.

Super Bowl: What Didn’t Work, and What I Changed

Clean execution, wrong audience. This shows the read — catching a mismatch the ad metrics didn’t hand over for free.

Same month, side by side — purchases on spend
Super Bowl
1
PURCHASE ON ~$52 SPEND
Fat Tuesday
8
PURCHASES ON $141 SPEND
Purim
11
PURCHASES ON $237 SPEND
The ad itself
2.01%
CTR · $0.72 CPC — PEOPLE CLICKED, THEY JUST DIDN’T BUY
Context

February 2026, alongside Fat Tuesday and Purim. Super Bowl felt like an obvious seasonal moment to capture — big cultural event, high attention, easy creative angle.

Problem — found only after running it

The campaign underperformed badly relative to the same month’s other campaigns, despite clean execution. The ad worked as an ad; the offer didn’t fit the audience. Email told the same story: the Super Bowl sends had the highest unsubscribe count of the month (68 combined across two sends), alongside below-average click rates.

What I concluded & did

This wasn’t an execution problem, it was an offer-audience mismatch — Super Bowl isn’t culturally relevant to Zeitlin’s identity-driven Jewish-deli audience, the way Fat Tuesday and Purim are. “Sports-themed campaigns are not recommended for this audience” went directly into the following month’s strategy doc as a standing rule, not just a note.

I used it to push a process change with ownership: marketing should act as a first filter on new catering offers — validating audience fit before an offer is finalized and handed over for execution, not after. I named the Super Bowl campaign explicitly as the reference case when making that recommendation to the owners.

Campaigns built around what the audience actually identifies with (Purim, Fat Tuesday) consistently outperform campaigns built around assumed demand (Super Bowl) — a principle that shaped how I scoped every campaign afterward, including Passover.

“Anyone can show a win. This shows the read — catching a mismatch the data didn’t hand me for free, tracing it to the real cause, and turning it into a process change instead of a shrug.”
The hard data FEBRUARY 2026 · META + BEEHIIV
{{ r.k }}
{{ r.v }}

No Super Bowl email screenshot is shown — the source deck reused the wrong asset, so nothing real backs a Super Bowl-specific image.

Want someone who reads the data honestly?