How to Set an Ad Cost Ceiling Before You Spend
How to set an ad cost ceiling: start from what a customer is worth after margin, then work back to a most per lead and per click. Write them down before launch.
To set an ad cost ceiling, take what a customer pays you over a period you can count on, multiply by your gross margin, decide how long you can wait to be paid back, and work backwards through your own conversion rates to a most you can pay per lead and per click. Write the three ceilings down before you spend anything.
Without them, the platform tells you what you are paying and almost any figure looks reasonable. With them, you have a number to hold the platform against, and a threshold that ends a test without an argument.
How to set an ad cost ceiling: start from what a customer is worth
Every ad decision reduces to one comparison: what a customer costs, against what a customer is worth. Do the arithmetic in advance, from your own numbers, and away from the ad manager. The tool is designed to make spending easy, and it is a poor place to be sensible.
Work through the figures in this order.
- What a customer pays you over a period you can honestly count on. Not the best case. The period you would stake the plan on.
- Your gross margin. Multiply by it, so you are working with what you keep rather than what you invoice.
- How long you can wait. Decide how long you can afford to be paid back, because ads are rented and the rent never stops.
The result is the most you can afford to acquire one customer. It is your first ceiling, and everything else follows from it.
Work backwards through your own funnel
You do not buy customers directly. You buy clicks, which become leads, which become conversations, which become closed deals. Each step has a rate, and you already know roughly what yours are from your own sales. Use them to translate the ceiling per customer into a ceiling per lead and per click.
| Ceiling | How you get it | What it governs |
|---|---|---|
| Most per customer | Customer value over your chosen period, after margin, within the payback you can afford | Whether the whole campaign is acceptable |
| Most per lead | The per-customer ceiling, worked back through leads to conversations and conversations to closed deals | The number the test is judged on |
| Most per click | The per-lead ceiling, worked back through the share of clicks that become leads | A quick warning that something has gone wrong |
If your conversion rates are weak, the ceilings come out low. That is not a flaw in the method. It is the method telling you something true about what you can afford.
Why the ceilings must exist before the spend
Ad platforms report generously on their own performance. They show a cost per result, and nearly any cost per result looks reasonable when it is presented without context. A figure only means something against a threshold.
Writing the three ceilings down first gives you that threshold while you are still calm. Later, mid-campaign, there will be a promising trend and an emotional attachment to the creative, and it will be hard to judge a number fairly. A figure you chose in advance is your defence against the platform's persuasive habit of making any cost look fine.
These ceilings become the kill thresholds in how to run a small ad test, where they sit beside a fixed budget and duration.
Pick the outcome the ceiling is measured on
A ceiling per lead only makes sense if you know what a lead is. The platform optimises towards whatever event you report back, so choose that event deliberately and make it as close to revenue as your volume allows.
| Outcome | When it suits | Caution |
|---|---|---|
| Clicks or page views | Almost never | Cheap, plentiful and only loosely related to buying |
| A completed lead form | The usual starting point for a business that sells through conversations | Counts people who fill in forms and never buy |
| A qualified lead or a purchase | When you have enough of them for the platform to learn from | May be too rare to read in a small test |
If you use a proxy because the real outcome is too rare, say so plainly, and write down that your ceiling is based on a proxy. How to send that outcome back properly is covered in how to send conversions back to an ad platform.
A worked shape, using only your own figures
You do not need invented numbers to see how this fits together. Here is the shape of the working, with the blanks you fill from your own records.
- Write what a customer pays you over your chosen period.
- Multiply it by your gross margin and write down what you keep.
- State the payback you can accept. That is your ceiling per customer.
- Find the share of leads that reach a conversation, and the share of conversations that close.
- Divide the per-customer ceiling back through those shares. That is your ceiling per lead.
- Do the same once more through the share of clicks that become leads. That is your ceiling per click.
If you do not know a rate in the middle, that is a finding in itself. It means you cannot yet say what you can afford, which is one of the conditions in when not to run paid ads.
Research spend and acquisition spend need different ceilings
Before you finish the arithmetic, settle which kind of spend this is. Early on, a small test can tell you which description of your offer makes strangers stop. The aim of that spend is a finding, not customers, so the useful limit is how much you are willing to pay for the finding.
Acquisition spend is different. It exists to produce customers, and the three ceilings above are exactly what it must stay under. The danger is the quiet drift from the first kind to the second, when a research budget becomes an acquisition budget and nobody decided that it should. Write one line stating which kind this is, and set the ceiling that matches.
Where ceilings usually go wrong
- Using revenue instead of what you keep. Skipping the margin step makes every ceiling look more generous than it is.
- Borrowing the platform's figure. If the first number you write is the platform's cost per result, you have let it set the standard.
- Leaving out the wait. A ceiling with no payback period quietly assumes you can wait for ever.
- Treating the proxy as the truth. A lead form is a stand in for a customer. Name it as such.
- Not writing it down. A ceiling held in your head moves when the numbers get uncomfortable.
Each of these is a failure of definition, not of skill. The fix is a page with the working on it and the three figures at the bottom, kept where you will see it when the test ends.
What the ceilings are not
A ceiling is the most you can afford, not the target. It is where a test ends, not where it should aim. A campaign that lands exactly on its ceiling has broken even on your own terms at best, and any slip beyond it is a loss you agreed in advance not to accept.
The ceilings are also not permanent. When your margin changes, when your conversion rates improve or when you learn how long customers really stay, rework the arithmetic. Keep the working visible so a colleague can see where each figure came from. An unexplained number is one nobody will defend when the campaign is under pressure.
Check the ceiling against the real pipeline
The ceiling per lead assumed that your rates are right. Test that assumption by following real ad leads through to conversations and closed deals, and compare what happened to what you assumed. A cheap lead that never becomes a conversation is not cheap, so a campaign can sit comfortably under its lead ceiling and still be losing money.
Give the check time. A lead from an ad today may close weeks or months from now, so judge cohorts on revenue only once they have had your typical time from first contact to closed deal. The method is set out in how to judge ads by deals not clicks.
When the platform suggests spending more
At some point the platform will offer a recommendation, usually to raise the budget. It will arrive when a result looks promising, which is exactly when judgement is weakest. The ceiling gives you a calm reply: if the result is under the ceiling for the whole test, you have earned a slightly larger test. If it is not, the suggestion is irrelevant.
That is all a ceiling does. It turns a dozen persuasive figures into one question with a yes or no answer, and it lets you answer it in the same way every time.
Frequently asked questions
Why start from the customer rather than the click?
Because the click is a means. What a customer is worth decides what you can afford, and the click ceiling follows from it. Starting from the click lets the platform's figures set your standard.
Why multiply by gross margin?
So you work with what you keep, not what you invoice. A ceiling based on revenue alone would let you pay for customers that cost you more to serve than they bring in.
What if I do not know my conversion rates?
Then finding them is your first job. Use your own sales records from leads to conversations and from conversations to closed deals. Until you have them, you cannot say what you can afford.
Can I change the ceiling in the middle of a test?
Not to rescue a result. Change it between tests, when the arithmetic itself has changed, and write down why.
The full method, with the checklists, is in The Paid Ads Playbook, a free 31-page playbook.