Google Ads cost per customer: formula, data and a worked example

Divide ad spend by new paying customers.

For a defined campaign and acquisition period, divide its ad spend by the number of distinct new paying customers reliably linked to that campaign’s leads. Use a consistent customer definition and allow time for those leads to buy.

This is the advertising component of customer acquisition cost. Full CAC also includes relevant sales and other marketing costs. A low ad cost per customer does not, by itself, establish profitability.

It also differs from Google Ads cost per conversion. A conversion can be a purchase, a call or another action you have chosen to measure. Check what your conversion actions actually mean before treating the reported count as customers. Google’s definition of conversion actions.

What the figures tell you—and what they leave open.

Campaign A wins on cost per lead. Campaign B wins on observed ad cost per customer. Across both campaigns, the blended figure is £6,000 ÷ 15 = £400 per customer. Averaging the two campaign costs (£600 and £300) would give the wrong blended result.

If every customer contributes £800 after direct delivery costs but before advertising, A leaves £1,000 and B leaves £5,000 after ad spend. These figures still exclude fixed overheads and other acquisition costs. Use your actual margins, refunds and customer values before drawing a commercial conclusion.

The result does not prove that shifting budget to B will reproduce its average cost. Extra spend may reach more expensive prospects; sales capacity and chance may also affect the outcome. Five versus ten customers is a reason to investigate and test, not a guarantee of future performance.

Bring three things together.

1. Campaign spend and dates.Export campaign identifiers, campaign names, dates and spend from Google Ads. Keep currency, time zone and date boundaries consistent. Search-term data helps explain spend but is not a substitute for campaign totals.
2. Deduplicated sales outcomes.Use a CRM or sales export with a stable lead or customer ID, enquiry date, outcome date and sales status. Agree whether “customer” means first payment, a won deal or another milestone. Keep existing customers separate when measuring new acquisition.
3. A reliable campaign connection.Use recorded campaign IDs, captured UTM values or an existing, documented mapping. Check renamed campaigns and duplicate records. Similar names or nearby timestamps alone are not sufficient evidence of a match.

Start with minimum necessary fields and pseudonymous IDs. Customer names, email addresses and free-text sales notes are usually unnecessary for the calculation. Revenue and margin fields are optional for customer-cost analysis, but necessary for credible contribution analysis.

Four mistakes that change the answer.

Mixing this month’s spend with this month’s sales

A sale completed this month may come from a lead acquired months ago. Group leads by acquisition period, then measure outcomes through a stated cut-off. Compare cohorts with similar follow-up time. Google also notes that delayed conversions can temporarily inflate recent CPA. Google’s guidance on conversion lag.

Treating unknown outcomes as lost sales

If only 80 of 100 leads can be reconciled, show the other 20 as unknown. Separate unmatched campaign links, missing statuses and open opportunities. A cost calculated from observed customers alone can be misleading when coverage differs between campaigns. State those limits beside the result.

Calling a campaign with zero sales “free”

When the customer count is zero, cost per customer is undefined, not £0. Report the spend and zero observed customers, then check follow-up time, tracking and relevance before deciding whether to pause anything.

Claiming a link proves causation

A recorded campaign association describes the evidence available. It does not prove the customer would never have bought without the ad, nor establish keyword-level attribution from campaign-level records. Incremental impact needs an appropriate experiment.

Turn the calculation into a decision.

First fix the decision’s weakest input: an unreliable conversion action, missing sales status or broken campaign link. Then inspect the search terms and landing-page promise behind the weaker cohort. Agree a small, measurable test with an owner, an outcome and a review window long enough for sales to mature.

Google supports importing eligible offline outcomes into Ads, but that requires suitable captured identifiers and setup. An export review cannot reconstruct missing historical links. Google’s overview of offline conversion imports.

If you need ongoing data collection, start with a tracking product. If you already have usable records and need help deciding what to change, Wrangg’s analyst-delivered review may fit. Our five product comparisons explain both options.