Cost per acquisition is what you paid, on average, for one customer. It is the number most worth watching, because it is the one that decides whether advertising is a cost or an investment.
How to work out cost per acquisition
Divide what you spent by how many customers it produced. The arithmetic is trivial and the definitions are where it goes wrong.
Decide what counts as an acquisition, and be strict. A form fill is not a customer. A phone call is not a customer. Somebody who paid is a customer, and everything before that is a step towards one with its own separate cost.
Decide also what counts as spend. Media only, or media plus the fee you pay somebody to run it. Both are valid; mixing them between months is not.
Why cost per acquisition beats the metrics above it
Clicks, click through rate and even leads can all improve while the business gets worse.
A campaign can double its enquiries by attracting people who will never buy, which looks excellent on every measure except this one. Consequently cost per acquisition is the first number that cannot be improved by attracting the wrong audience.
Still, each of those earlier measures has a job. They belong to a running order that gives each number its place, which is why any one of them misleads when read alone.
What a good cost per acquisition looks like
There is no industry answer, and anybody offering one without asking about your margins is guessing.
The only honest test is against what a customer is worth to you. That means the profit on a first order at minimum, and ideally the profit over the whole relationship, since a customer who returns can justify a far higher acquisition cost than one who buys once.
A worked example with invented round numbers, not a benchmark. If a customer produces forty in profit on the first order and nothing afterwards, an acquisition cost of thirty leaves ten. If that same customer usually buys three times, they produce a hundred and twenty, and thirty starts to look cheap. Same campaign, opposite conclusion, decided by a number that lives in your accounts rather than in an advertising platform.
The chain that produces your cost per acquisition
It is the product of everything before it, which is why it is diagnostic as well as decisive.
What a click costs, set mostly by competition and relevance.
How many clicks become enquiries, set by the landing page.
The share of those enquiries that are genuine, set by targeting.
Your closing rate on the genuine ones, set by sales and by fit.
A rise in the final number always comes from one of those four. Working through them in order is faster than adjusting bids and hoping.
Changes that actually lower cost per acquisition
Landing pages usually move it most, because a page converting more visitors reduces the cost of every step after it.
Targeting comes next, since removing audiences who never buy raises the proportion of enquiries worth having. Offer and messaging follow, because what you say changes who responds. Bidding matters least, though it is what most people adjust first.
Meanwhile improving how quickly enquiries are followed up often does more than any advertising change, and it costs nothing. A lead answered in ten minutes and a lead answered the next morning are not the same lead, and the advertising did identical work in both cases.
Lead quality, and the feedback loop most accounts lack
An advertising platform knows what it delivered. It does not know which of those leads was any good unless somebody tells it.
So build the loop. Somebody in the business marks which enquiries were real and which became customers, and that information reaches whoever manages the campaigns. Without it, the account optimises towards whatever is easiest to produce, and cost per acquisition quietly rises while cost per lead falls.
Seasonality and patience
Most businesses have months where the same advert produces twice the result and months where it produces half.
Compare against the same period last year rather than against last month, and judge over a window long enough to contain enough customers to be meaningful. A month with four customers cannot tell you much, however precise the arithmetic looks.
When cost per acquisition is the wrong measure
It struggles where the sale takes a very long time, where several people are involved in the decision, or where the first purchase is deliberately unprofitable.
In those cases measure the step you can measure honestly, such as qualified opportunities, and hold the longer number as an annual review rather than a monthly dial. Precision that is invented is worse than an honest approximation.
What to do this quarter
Define an acquisition strictly, agree what spend includes, and build the loop that tells you which leads were real.
Then improve the landing page for your largest campaign, remove the audience producing enquiries that never close, and leave the bidding alone. That sequence lowers cost per acquisition more reliably than anything available inside the platform.
Review the definition itself once a year. Businesses quietly change what they sell, and a measure counting the wrong event is worse than no measure, because everybody trusts it.
Frequently asked questions
How do I calculate cost per acquisition?
Divide what you spent by the number of customers it produced. The arithmetic is simple and the definitions matter: a form fill is not a customer, and you should decide whether spend includes management fees before you start.
What is a good cost per acquisition?
There is no industry answer. The only honest test is against what a customer is worth to you, ideally over the whole relationship rather than the first order. Anybody quoting a benchmark without asking about your margins is guessing.
Why not just watch cost per lead?
Because a campaign can halve its cost per lead by attracting people who never buy. Cost per acquisition is the first number in the chain that cannot be improved by bringing in the wrong audience.
What lowers it most?
Landing pages usually, because a page converting more visitors reduces the cost of everything after it. Then targeting, then offer and messaging. Bidding matters least, although it is what most people adjust first.
Why does my platform not optimise for this already?
Because it knows what it delivered, not which leads were any good. Somebody in the business has to mark which enquiries were real and feed that back. Without the loop, the account optimises towards whatever is easiest to produce.
How long a period should I judge over?
Long enough to contain enough customers to mean something, and compared against the same period last year rather than last month. A month with four customers cannot tell you much, however precise the arithmetic looks.
Does customer lifetime value change the answer?
Completely. A customer who buys three times can justify an acquisition cost that would be ruinous for a single purchase. That number lives in your own accounts rather than in an advertising platform, which is why the platform cannot judge for you.
When is cost per acquisition the wrong measure?
Where the sale takes a very long time, involves several decision makers, or begins with a deliberately unprofitable first purchase. Measure a step you can measure honestly instead, and review the longer number annually.