Blog Playbooks · The Benchmark Playbooks · Part 2
Most Meta ads barely run. How to judge one in its first week
Two of every three Meta ads we graded spent under $100 in their whole life. Here is how to read a new ad early without cutting a winner before it starts.
Here is the number that surprises owners most when we show them Creative Benchmarks: most Meta ads hardly spend anything. We have run over 18,920 Meta ads, and two of every three spent under $100 across their entire life. A quarter spent under $10.
That changes how you should judge a new ad. If you expect every ad to get a fair, even share of your budget, you will misread almost all of them. This playbook explains what the spread looks like, why it happens, and a first-week routine that protects your budget without killing a winner early.
What the benchmark shows
We grouped every graded ad by how much it spent over its whole life.
Source: Creative Benchmarks, benchmark.sneeze.it (18,920 Meta ads)
Add the first two bars and you get 65%: about two ads in three never reached $100. About one in twenty ever spent more than $1,000.
Meanwhile 6.2% of all ads became winners. In Creative Benchmarks a winner is an ad that spent at least 10 times its own account's median ad, and at least $500. So the picture is a long tail of ads Meta barely showed, a middle group that got a real look, and a small group that took most of the money.
An ad Meta barely spent on has not failed. It has not been tested.
Why most ads barely run
Meta does not split your budget evenly. When an ad set holds several ads, delivery leans toward the ones it expects to get results and starves the rest, often within the first days. That is the system working as designed. It also means an ad that spent $8 has told you almost nothing about whether people want it. It may be a weak ad. It may be a good ad that lost an early race to a sibling.
The opposite mistake matters more. When an ad starts taking a large share of spend in its first week, owners often get nervous about the money and pause it, or they judge it on two or three days of cost per lead. That is how winners get cut before they become winners.
So a first-week review has two jobs: do not read meaning into ads that barely spent, and do not panic about the ones that are spending.
The playbook: a first-week routine
A hypothetical to set the scene. Say you spend $1,500 a month and have just launched three new ads into one ad set. After seven days, one has spent $140, one has spent $22 and one has spent $6. Here is how to handle it.
- Wait out the first few days
Do not judge anything on day one or two. Delivery is still sorting the ads, and cost per lead from a handful of results swings wildly. Set a calendar reminder for day seven and leave the ads alone until then.
- Sort by spend, not by cost per lead
On day seven, sort the new ads by amount spent. Spend is Meta's vote. In the example, the $140 ad is the one Meta is backing; the $6 ad has had no real test.
- Leave the leader running
The ad taking the most spend gets more time, even if its cost per lead looks a little high after a week. Judge it against your account's average cost per lead after it has gathered a meaningful number of leads, not a few.
- Do not grade the starved ads
An ad that has spent only a few dollars has no verdict yet. Do not call it a loser in your notes. If you believe in it, give it a fair test: move it into its own ad set, or launch it again later with a different set of siblings.
- Act on real problems only
Pause early only for clear faults: a broken link, a wrong offer or price, an ad that is spending and getting clicks but zero leads while the others convert. Those are mistakes, not test results.
- Track the leader toward the winner line
Check the leading ad each week against your winner line (10 times your median ad, at least $500). In the example, if the winner line is $500, the $140 ad would pass it in about four weeks if it kept that pace.
- Then grade it twice
Once an ad crosses the winner line, check its cost per lead against your account average. Meta's pick and your pick are not always the same ad, as grade every ad twice explains.
Two mistakes this prevents
Pausing the ad Meta likes. The ad taking most of the spend in week one is the one Meta is backing, and every winner starts as an ad Meta backed. Cutting it to "save money" usually just means starting over.
Writing off ads that never ran. If two ads in three barely spend, your list of "ads that did not work" is mostly ads that were never really tried. Before you give up on an idea, check whether it ever got spend.
Both mistakes come from treating every ad as if it had the same chance. It did not, and the spread above shows it. If you want more ads to get a fair shot, the answer is usually cadence and structure, not overriding Meta: see how many new ads a week for setting a launch pace your budget can carry, and what 18,920 Meta ads say about cost per lead for which kinds of ad to try.