Blog Inside Sneeze It · Inside Sneeze It · Part 2
We have run over 18,920 Meta ads. Here is how that changes the way we run yours.
Most ads barely run, the platform's favorite is often not your cheapest lead, and the accounts that launch more new ads find more winners. Each of those findings is now a rule in how we work.
Every agency says it knows what works. Ask how it knows, and the answer is usually someone's taste, or the platform's dashboard, or the last campaign someone remembers. We wanted a better answer, so we graded every Meta ad we have run for the multi-location brands we work with.
We have run over 18,920 Meta ads. Together they brought in 121,333 leads. Only 1,175 of them became winners, meaning Meta put at least ten times the account's typical ad spend behind them, and at least $500. That is 6.2%, and those few ads took 49% of all the money.
Source: Creative Benchmarks
We published the findings in What 18,920 Meta ads say about cost per lead. This post is about the other half: what we changed in our own work because of them. Four findings became four rules, and they decide what happens in a client's account every week.
Rule one: plan for most ads to fail
Two out of three ads in our data spent under $100 in their whole life. Meta tried them, lost interest and moved the money elsewhere. The auction works that way in every account we looked at, good work included.
So we treat a new ad as a question for the auction. Most answers will be "no", and that is fine as long as the account keeps asking. The work that matters is making sure the next question is a better one, which is what the other three rules are for.
Most new ads will hear no from the auction. The accounts that keep asking are the ones that find a yes.
Rule two: keep new ads coming
When we lined up accounts by how many new ads they launched each week, the pattern was plain. Across the 97 accounts with at least 50 leads, the median account launching fewer than one new ad a week found 0.37 winners a month. Accounts launching four or more a week found about one winner a month.
Source: Creative Benchmarks, median across 97 accounts with 50+ leads
The typical account in our data launched 1.6 new ads a week, so that is the number we push up: more new ads means more chances at a winner. The data shows more winners found. It does not show that every lead gets cheaper. So we treat volume as the way to find good ads, and cost per lead as the way to judge them.
That is also why the free benchmark report we send owners ends with a launch plan for their budget and four ad briefs ready to make. A plan that does not produce new ads every week does not move this number.
Rule three: make the kind of ad that wins
Volume only helps if the new ads have a fair shot. We labeled 2,126 ads three ways: how the ad looks, how it opens and how it was made. Three results changed what our team makes.
The first seconds matter most. Ads that opened with a bold claim became winners 14.9% of the time, more than twice the 6.2% rate across all our ads, and brought leads about 24% cheaper than average. Ads that opened on a relatable moment cost about 42% more per lead than average.
Some of the most common designs cost more. Headline-only ads, one of the designs we see most (323 of the labeled ads), cost about 29% more per lead than average. Photo collages became winners 17.9% of the time.
High production did not help much. We labeled 784 ads as high production, and 7.0% of them became winners. Only 64 were phone-shot, in the style of a member or staff video, and 15.5% of those became winners. Sixty-four ads is a small group, so we treat phone-shot as the place to test first, and we test it a lot.
What most brands make
Polished, high-production video and ads built around a headline. They look finished and take longer to make. In our data, high production won 7.0% of the time, and headline-only ads cost about 29% more per lead than average.
What we make more of
Openings that make a bold claim, photo collages and phone-shot clips of real people, made quickly and often, so the account always has fresh ads in the test.
These are patterns across all of our accounts, not findings about one industry. Where a label has only a small number of ads behind it, we treat it as a place to test first.
Rule four: judge every ad on what a lead cost
The finding we act on most: of the 1,175 winners, only 47 in 100 brought leads cheaper than their own account's average. The other 53 took the budget and brought leads at or above average, or none at all, even though Meta kept spending on them.
So every ad we run gets two grades: Meta's, by how much it spent, and ours, by what a lead from it cost. When the two disagree, we side with the cost. We explain how that works inside Studio, our creative tool, in We judge ads by cost per lead, not by what Meta chose to spend. Since October 9, 2026, every new ad is also tested against the client's own results before the client sees it, and that test refuses to run on a read older than seven days.
- 1AskLaunch new ads every week, on purpose
- 2MakeFavor the kinds of ads that win in our data
- 3GradeMeta's grade, then cost per lead
- 4DecideMake more, stop or test, with the evidence beside it
What this means for you
If you run 20 locations, you should be able to ask your agency why it makes the ads it makes, and get an answer with numbers in it. Ours comes from watching 18,920 ads run, writing down what happened and building our process around it. Every number in this post is on our public benchmark page, where you can check it.
