Blog Playbooks · The Benchmark Playbooks · Part 5
Reading cost per lead without fooling yourself
Cost per lead is the most useful number in a lead-gen account and the easiest one to misread. Here is how to compare it fairly, and how to tell a real difference from a lucky week.
Cost per lead is simple arithmetic: what you spent, divided by the leads you got. That simplicity is the problem. A number that easy to calculate is easy to calculate wrong, compare to the wrong thing, and trust on too little evidence.
We have run over 18,920 Meta ads and tracked 121,333 leads. When we labeled 2,126 of those ads three ways (how they look, how they open, and how they were produced), the average cost per lead across all of them was $26.62. Some labels came in well under that, some well over.
Source: Creative Benchmarks
Those numbers are useful. They are also easy to misuse. This playbook covers the three ways people fool themselves with cost per lead, and what to do instead.
Mistake one: comparing to the wrong average
There are two averages in play, and they answer different questions.
Your account average is your total spend divided by your total leads over a period. It tells you what a lead costs you, in your market, with your offer, your website and your follow-up. It is the only fair yardstick for your own ads.
A label average, like the $18.73 for before-and-after ads or the $34.35 for headline-only ads in our data, pools ads from many accounts. It tells you which kinds of ads tended to cost less across a lot of businesses. It does not tell you what your before-and-after ad should cost.
So if your account average is well above $26.62, that does not mean your ads are bad. Your market or offer may simply cost more. And if one of your headline-only ads beats your own account average, it is a good ad for you, whatever the label average says.
Use label averages to decide what to try. Use your account average to decide what to keep.
Use label averages to decide what to try. Use your account average to decide what to keep.
Mistake two: averaging the averages
When you want one cost-per-lead figure for a group of ads, add up the group's spend and add up its leads, then divide. Do not take each ad's cost per lead and average those.
Here is why, with made-up numbers. Say Ad A spent $500 and brought 25 leads, so $20 a lead. Ad B spent $30 and brought 1 lead, so $30 a lead. Averaging the two figures gives $25. But the group really spent $530 for 26 leads, which is about $20.38 a lead. The tiny ad pulled the simple average up by almost $5, even though it barely spent anything.
Ads Manager does the right thing when you look at a campaign or ad set total. The trouble starts when people copy ad-level figures into a spreadsheet and average the column.
Mistake three: trusting small numbers
Cost per lead on a few leads is mostly noise. Say an ad spent $90. With 3 leads it costs $30 a lead. With 4 leads it costs $22.50. With 2 leads it costs $45. One lead, which can come down to a single person deciding to fill in a form on a Tuesday, moves that ad from "great" to "terrible".
The same caution applies to our own data. Several of our label groups are small. Before-and-after has 40 ads. Announcement openings have 40. Contrarian openings have 41. Product demos have 69. When we say announcement openings averaged $73.48 a lead, that is a real number from real ads, but 40 ads is not many. A handful of expensive ones can drag a small group's average a long way. Treat small groups as a list of where to test first, never as rules.
Source: Creative Benchmarks, benchmark.sneeze.it
Look at the two ends of that chart. The cheapest and the most expensive openings are both among the smallest groups. That pattern is worth noticing: small groups are more likely to land at the extremes simply because there are fewer ads to even things out. The 611-ad and 626-ad groups sit close to the overall average, which is what you would expect from big groups.
The playbook
- Set your yardstick
Pick a fixed period, such as the last 30 days, and work out your account average: total spend divided by total leads. Write it down with the dates.
- Pick a lead floor before you look
Decide how many leads an ad needs before you will judge its cost per lead, and write that number down first. Choosing the rule after you have seen the results is how people talk themselves into what they wanted.
- Compare each ad only to your own average
For ads above your floor, mark them cheaper or pricier than your account average for the same dates. Ignore other accounts and other industries.
- Pool before you summarize
When you want a figure for a group of ads, such as all your video ads, add up spend and leads and divide. Never average a column of cost-per-lead figures.
- Check that the leads are leads
Before celebrating a cheap ad, look at what came in. Make sure the result column is counting real form fills or bookings, and check a sample of the leads against what your team saw.
- Use label averages to choose tests, not verdicts
If before-and-after or product demo ads are missing from your account, those are reasonable things to try next. Judge them by step 3 when they have run, not by our averages.
Where this leaves you
Cost per lead is still the number to steer by. It just needs three habits: compare to your own average, pool before you summarize, and wait for enough leads. With those habits it tells you the truth. Without them it tells you whatever the last lucky week said.
Cost per lead is also not the only grade an ad deserves. An ad can be cheap and never scale, or expensive and carry your whole budget. Grade every ad twice shows how to read cost per lead next to spend share, and winner rate vs cost per lead covers what to do when the two disagree. The full dataset is in what 18,920 Meta ads say about cost per lead.
