How We Decide an Ad Is Dead: What Feast's Own Ad Account Taught Us
Feast's own account found a clear early-purchase pattern. Use it to separate a dead ad from broken tracking or a campaign that needs more time.
A July 22, 2026 summary of Feast's own acquisition account lists 11 converting ad sets from an April 25 to July 22 review. Their median cumulative spend through the day of the first reported purchase was about $135, with a range of $17 to $579. These are ad-set totals through a reporting day, not the exact cost or time of an individual ad's first sale.
The summary describes Feast's own customer-acquisition campaigns. It is not a restaurant guest benchmark, and the $579 high end is not a stopping threshold for your ads.
The pattern is strong enough to guide us, not strong enough to become a law
The retained note gives the review dates and 11 converting ad sets, but not the total number of ad sets that ran, the full row-level export or all attribution settings. The script used daily ad-set reporting and Meta purchase events. The note does not show those events reconciled to paid customers.
The $135 median describes only ad sets that converted. Ad sets with no purchase are absent from that median, so it cannot tell you how likely a new test is to work. The historical maximum is an observed value, not evidence that more spending can never produce a first purchase.
Your restaurant needs its own boundary based on what a verified visit is worth and how much uncertainty it can afford.
Do not call the ad dead until you know the result can be recorded
An ad with zero reported sales may be bad. It may also be sending people into a broken link, an untracked reservation flow, an offer that staff cannot redeem, or a checkout that loses campaign identity.
Before judging the creative, complete the guest path yourself. Confirm that the page loads, the response is saved, the visit can be identified, and the completed sale reaches the report. Otherwise, you are measuring the tracking failure along with the ad.
Find the first part of the campaign that stopped working
The useful question is not simply “Did it sell?” It is “Where did the response disappear?”
- If the ad cannot earn attention or clicks from the intended local audience, change the creative or message.
- If people click but do not respond, repair the page or offer.
- If people claim or reserve but do not arrive, look at intent, reminders, availability, and the guest experience.
- If they arrive but no sale is matched, repair identification or POS reconciliation.
- If verified visits arrive but cost too much, the economics—not necessarily the creative—are the problem.
This keeps the team from replacing a strong ad when the actual failure sits after the click.
Set the loss limit from restaurant economics
Start with the contribution available from a first visit after the ordinary variable costs of serving it. Decide how much of that amount you are willing to spend on acquisition, then set a total learning budget the business can afford to lose.
Include media, creative, offer fulfillment, and campaign-specific fees once each. Do not quietly assume years of repeat purchases unless your own retention data supports them.
Feast's $135 median and $579 high end can provide context, but they should not replace this calculation. Restaurants with very different check sizes cannot share one rational cutoff.
Give a promising ad enough time for the restaurant result to mature
Give a promising ad enough time for its intended restaurant result to occur.
Impressions and clicks appear quickly. Reservations need time to occur. Offer claims need time to turn into visits. Settled checks may arrive days after the first exposure.
Decide separately whether to keep buying new responses and whether earlier responses are ready to judge. A reservation for Saturday dinner is pending on Friday. Pausing the ad does not stop that reservation from becoming a completed visit.
Keep a dated log of spend and responses for each run period. Record when the booked visits are due, when claimed offers expire and how long check reconciliation actually takes. Evaluate that period after its visit window and reporting delay have closed. Include all of its campaign costs, even spend that produced no response. Keep pending visits and unresolved check matches visible separately.
Keep buying only while tracking works, responses are credible and new spend stays within the learning budget. At the budget limit, pause new spend and let the existing responses mature. A broken path or an operating problem is a reason to pause sooner. Then compare the mature period's cost per verified completed visit and visit contribution with your written limits before restarting, repairing or ending the test.
An ad is not dead because the team is bored with it. It is dead when the campaign has had a fair, trackable chance and the restaurant outcome still cannot support the spend.
If you want to connect ad responses to visits and POS sales before deciding what to cut, Book a demo of Feast.