Blog Playbooks · The Benchmark Playbooks · Part 6
Winner rate vs cost per lead: which number to follow, and when
Some kinds of ads bring cheap leads but rarely scale. Others scale often but cost more per lead. Here is how to read both numbers and pick the one that fits the job in front of you.
We have run over 18,920 Meta ads, and when we sorted 2,126 of them by how they look, how they open and how they were produced, we tracked two numbers for every group. The first was cost per lead. The second was winner rate: the share of ads in the group that became winners, meaning they spent at least 10 times their own account's median ad and at least $500.
Most of the time the two numbers point roughly the same way. Sometimes they do not, and those are the cases worth understanding, because they change what you should make next.
Two numbers, two questions
Cost per lead answers: when this kind of ad runs, what does a lead cost? It is a measure of efficiency. Across the 2,126 labeled ads, the average was $26.62.
Winner rate answers: how often does Meta pick this kind of ad and put real money behind it? It is a measure of scale. Across all 18,920 ads, 6.2% became winners. The labeled ads below are a subset of those, 2,126 of them, so treat 6.2% as a rough yardstick for them rather than an exact baseline.
These are different questions, so they can have different answers. An ad can bring cheap leads on a small budget and never be given a big one. Another ad can be given a big budget and bring leads at an ordinary price. Neither is wrong. They are good at different things.
An ad can bring cheap leads on a small budget and never be given a big one.
Where they disagree
Here are the winner rates for a selection of labels, next to their cost per lead.
Source: Creative Benchmarks, benchmark.sneeze.it
Three patterns stand out.
Photo collage: scales often, costs about average. Photo collages had the highest winner rate of any look we labeled, 17.9%, but their cost per lead, $27.75, was a little above the $26.62 average. Meta funded them often. That does not mean each lead was a bargain.
Contrarian openings and before-and-after: cheap, but rarely scaled. Contrarian openings had the lowest cost per lead of any opening, $14.30, yet only 5.9% became winners. Before-and-after had the lowest cost per lead of any look, $18.73, with 6.9% winners. These are the ads that did their job well when they ran, but Meta did not often push them to big budgets.
Curiosity and bold claim openings: good on both. Curiosity openings came in at $23.43 with 16.0% winners. Bold claim openings came in at $20.16 with 14.9% winners. Both were cheaper than average and well above the overall winner rate. Phone-shot UGC sits in the same corner at $22.21 and 15.5%.
And one label was weak on both: the feature and benefit list, at $32.22 a lead and 4.4% winners.
A word on sample sizes before anyone builds a plan on this. Contrarian openings are 41 ads, before-and-after is 40, curiosity is 57, phone-shot UGC is 64 and product demo is 69. Those are small groups, and a few ads either way could move their numbers. Read them as where to test first, not as rules. The photo collage group is 106 ads and bold claim is 99, which is a bit firmer, but still not large.
Which number to follow, and when
There is no single right answer. It depends on the job you need the next round of ads to do.
Follow winner rate when you need to spend more. If you are opening a new location, filling a slow season, or you have budget you cannot get Meta to spend well, you need ads Meta is willing to scale. A kind of ad with a high winner rate is a better bet for that job, even if its cost per lead is only average.
Follow cost per lead when your budget is fixed. If you spend a set amount each month and want the most leads for it, efficiency matters more than scale. A cheap kind of ad that rarely becomes a winner can still be worth running at a modest budget, as long as it holds its cost.
Prefer kinds of ads that do well on both when you are starting a test. Bold claim and curiosity openings, and phone-shot UGC, were cheaper than average and well above the overall winner rate in our data. If you are choosing what to make next and have no strong reason to pick otherwise, start there.
One caution that ties back to the rest of this series. Winning is Meta's grade, not yours. Only 47 in 100 winners brought leads cheaper than their own account's average. So a high winner rate tells you a kind of ad is likely to get spend. It does not promise that the spend will be efficient in your account. You still have to check, which is what grade every ad twice is for.
- Name the job for this month
Write one line: are you trying to spend more, or trying to get more leads from the same spend? That line decides which number leads.
- Label your own recent ads
Go through last month's ads and tag each one with a look, an opening and a production style, using plain names like the ones in the chart. You only need rough labels.
- Find what you have not tried
Compare your labels to the chart. If you have no bold claim or curiosity openings, no phone-shot UGC, or no product demos, those are gaps worth testing.
- Build two or three new ads for the gap
Make each one different in a single way from an ad you already run, such as the same offer with a bold claim opening instead of an announcement, so you can tell what changed.
- Judge them on the number that matches the job
If the job is scale, watch which new ads Meta gives a growing share of spend. If the job is efficiency, watch cost per lead against your own account average, and only once each ad has enough leads to mean something.
- Check the winners against your average
Any new ad that takes a big share of spend gets its second grade: is it bringing leads cheaper or pricier than your account average? Keep the ones that pass both.
If you want the full method for reading cost per lead fairly, including why small groups swing so much, see reading cost per lead without fooling yourself. The full labeled dataset is in what 18,920 Meta ads say about cost per lead.
