Blog Inside Sneeze It · Inside Studio · Part 2
We test every new ad against your own results before you see it
Before a new ad reaches your inbox, Studio checks it against what has actually brought you leads at a good cost. If the read it is checking against is more than a week old, it refuses.
The usual order of operations in an agency goes like this. The creative team makes an ad. The client approves it. It runs for a few weeks. Then someone looks at the numbers and finds out whether it was a good idea.
That order puts the most expensive step, running the ad, in front of the cheapest one, which is asking whether the ad looks like the ones that already work for you. We flipped it. Since October 9, 2026, every new ad our team builds in Studio can be tested against your own results with one button, before it goes into your review queue.
To explain how the test works, we have to start with what it tests against.
What Studio already knows about your accounts
The test stands on a feature we call Creative Direction, live since September 23, 2026. It reads every ad your Meta accounts ran in the last 120 days, read-only, with spend, leads and the dates each one ran.
Then it sorts every ad into one of four groups, using plain arithmetic:
- Winner: the ad spent at least 10 times what the account's typical ad spent, and at least $500.
- Middle: it ran 28 days or more without becoming a winner.
- Loser: it stopped before 28 days.
- Too new to judge: it has not had time yet.
Next, Studio labels each ad by what it shows and how it opens: offer-led, member story, class in progress, logo only, text only and so on, chosen from a fixed list so the labels stay comparable. It looks at the picture, and for video it also listens to the sound, so a voiceover ad that happens to show only text on screen is not filed as "text only."
Finally, it compares your mix with our industry benchmark of 578,750 Meta ads from 6,015 advertisers (2026), by vertical, and writes findings with the evidence attached: make more of this, stop this, test this. As of October 10, 2026, Studio's public homepage counts 97 creative direction reports written from real results.
One design rule runs through all of it. The AI labels the ads. It never decides who won. Winning and losing is arithmetic, and arithmetic can be pinned down by a test in our code.
The AI labels the ads. It never decides who won.
Why "the platform spent the most on it" is not the answer
Meta decides where your money goes. An ad the algorithm favors gets more spend, and in most reports, more spend reads as success. Our own two-year read of 18,920 Meta ads told a different story. Only 6.2% of ads became winners by the spend rule, and of every 100 of those winners, 47 beat their own account's average cost per lead. The rest got the budget without being the best ads in the account. The full analysis is in What 18,920 Meta ads say about cost per lead.
That is why Direction asks a second question the spend rule does not: did this ad actually produce leads at a good cost? And it is why the test leans on your cost per lead, not on what the platform chose to spend.
Source: Creative Benchmarks, Sneeze It two-year read of 18,920 Meta ads
One button, before your inbox
When our team finishes a new ad, they press Test in the editor, or upload a finished file on the Direction page. Here is what happens next.
- 1LabelThe new ad is labelled the same way as your past ads
- 2Your readDoes your latest Direction say to make more of this kind?
- 3Your costIs this kind cheaper or dearer per lead than your average?
- 4LessonsDoes it break anything we have learned about your brand?
- 5VerdictOn direction, mixed, or off direction, with reasons
The scoring is deliberately simple. The ad gains or loses points for matching or contradicting your latest Direction read, and gains or loses points if that kind of ad has cost you meaningfully less or more per lead than your account average. The industry benchmark only counts when you have fewer than three ads of that kind, because your own history beats an industry average whenever you have enough of it.
Then Studio checks the ad against everything it has learned about your brand from past corrections, for example: the price goes in the first line, this location never shows the pool, and so on. A broken lesson downgrades an "on direction" verdict to "mixed," however good the numbers look.
The result shows up next to the ad itself, drawn inside the real platform frame, on one laptop screen. Our team reads it, fixes what needs fixing, and only then sends it to you.
A stale read is a wrong read
The test refuses to run against an old read. If your newest Direction run is more than 7 days old, Studio says so and asks for a fresh one before it will give a verdict. Accounts move. An offer that was cheap a month ago may have worn out, and a test against last month's picture would approve this month's mistake.
The usual order
Make the ad. Send it for approval. Run it for weeks. Find out from the numbers whether it matched what works.
Our order
Make the ad. Test it against your last 120 days, your own cost per lead and your brand's lessons. Fix it. Then send it to you.
Source: Studio code; public homepage count as of Oct 10, 2026
What you see, and what you do not
A test result is an internal tool first. When it helps a conversation, our team can share it with you as a link, copy it into Slack, email it, or export a slide-sized image, and can stop sharing at any time.
Inside One, our client portal, you see Direction reads that our team has reviewed with you and published, not a raw grade. That is a rule, not an accident: a read is more useful when someone who knows your locations has looked at it first.
The test does not replace your judgment, and it does not approve anything. You still approve every ad yourself, in the real frame, as described in Approve exactly what will run. The difference is that by the time an ad reaches you, it has already been held up against what works for you.
