Quick definition: Customer acquisition cost (CAC) is the cost incurred to acquire a new defined customer or account during a stated period, divided by the number of those customers acquired.
What is customer acquisition cost?
CAC connects commercial spending to new customer acquisition. It can include paid media, sales compensation, agency fees, campaign production, onboarding incentives, and the portion of staff or tools directly attributable to acquisition. The right scope depends on the decision. A channel CAC may include only channel spend and attributable new customers; a fully loaded CAC may also include sales and marketing payroll. Both can be useful, but neither should be presented without the cost boundary.
CAC is not cost per lead, cost per click, cost per signup, or cost per activated user. Those metrics describe earlier funnel outcomes and are often much lower because many leads do not become customers. CAC is also not customer lifetime value. CAC is an incurred cost to acquire; CLV estimates the future value of the acquired relationship. A healthy CLV:CAC relationship requires matched definitions, margins, time horizons, and customer units.
Define “acquired customer” carefully. A new paid account, first invoice paid, first delivered order, or activated paid workspace can each be valid depending on when value and cost become real. Trial starts and returning customers should not silently count as new paid acquisitions. The same account cannot be counted twice because it reactivates or purchases another product unless the metric explicitly measures re-acquisition.
CAC formula and denominators
The basic formula is:
CAC = acquisition costs during period / newly acquired customers during period
Use costs and customer outcomes that reasonably belong to the same acquisition cohort. A simple calendar-month numerator and denominator can be misleading when campaigns generate leads in one month and paid customers months later. For long sales cycles, cohort costs by lead or campaign start and observe conversion over an agreed attribution window. For self-serve products, a short window may be appropriate, but refunds and payment failures still need treatment.
| CAC type | Costs included | Customer denominator |
|---|---|---|
| Paid-channel CAC | Media and channel-specific fees | New customers attributed to channel |
| Blended CAC | All selected acquisition spend | All new customers |
| Fully loaded CAC | Spend, sales labor, tools, production | New customers under stated rule |
| Incremental CAC | Incremental cost of a change | Incremental customers caused by it |
Attribution is not causality. A last-click model can credit an ad that merely reached customers who would have purchased anyway. Incremental CAC seeks the cost per additional customer caused by activity and normally needs a credible experiment, holdout, or quasi-experiment. Read our experiment-hypothesis guide when designing that causal evaluation.
CAC in A/B testing
CAC is usually a business KPI or secondary outcome, not a fast primary product-test metric. For an acquisition landing-page test with fixed traffic spend, conversion rate and qualified-customer rate can be measured sooner; CAC follows from cost allocation and incremental acquisition. For marketing holdouts, the primary question is often incremental customers or profit per targeted unit, not merely attributed conversions.
Suppose a treatment changes an offer. It might lower immediate CAC by increasing signups, yet raise customer-service costs, refunds, fraud, or early churn. Measure paid acquisition, activation, net revenue, margin, cancellation, and retention across all assigned eligible units. Do not calculate treatment CAC only among people who clicked the offer: clicking is post-treatment behavior. Guidance on combining a primary metric with protection metrics is in this primary-versus-guardrail article.
Predefine spend allocation. If campaign spend is shared by variants, split it using a planned rule such as equal assigned traffic, actual eligible impressions, or controlled budget allocation. Never allocate more cost to the losing arm after seeing conversion results. State whether creative production, discounts, affiliate fees, and sales labor are included.
Worked CAC calculation
A company spends $48,000 on paid search and associated agency fees in April. It attributes 160 new paid accounts to the campaign under a pre-agreed 30-day first-payment rule.
CAC = $48,000 / 160 = $300 per new paid account
It then runs a randomized landing-page experiment under the same channel and bid strategy. Control receives $24,000 of planned spend and produces 72 new paid accounts; treatment receives $24,000 and produces 96. The descriptive values are $333 and $250 per new account. Treatment appears to lower CAC by $83, but the team checks whether traffic quality, refunds, account eligibility, and follow-up time are balanced. It reports the conversion difference and uncertainty rather than implying a fixed $83 saving from one test.
CAC data-quality caveats
Costs commonly live in ad platforms, invoices, finance systems, CRM records, and payroll allocations. Reconcile currency, tax, agency fees, credits, and date rules. Customer counts need stable identifiers to deduplicate leads that convert through multiple channels and accounts created by the same organization. Document the attribution model, window, and how organic, partner, and sales-assisted customers are handled.
Platform conversion reporting can over-credit the platform because it observes clicks but not the counterfactual. Privacy restrictions, cross-device behavior, offline sales, and delayed payments make attribution incomplete. Holdouts, geo tests, and randomized campaign exposure can estimate incrementality, but they also need sufficient sample size, stable delivery, and protection against contamination.
Do not compare CAC across periods with a changing definition. A new sales-team allocation, changed discount policy, or switch from first purchase to activated customer can create a false improvement. Version the metric and retain source-level detail.
Common CAC mistakes
- Dividing by leads instead of customers: cost per lead is not CAC.
- Leaving costs out selectively: a channel number may look cheap only because labor or fees were excluded.
- Relying on attribution as causal proof: credited customers may not be incremental.
- Mixing accounts and people: the cost and customer unit must match.
- Ignoring refund and cancellation windows: early customers can reverse.
- Comparing mismatched cohorts: spend and customer acquisition often occur in different periods.
Frequently asked questions
What costs should CAC include?
Include the costs relevant to the decision and disclose them. Channel CAC and fully loaded CAC are both valid only when labeled clearly.
Is CAC the same as cost per acquisition?
Sometimes, but cost per acquisition may refer to any action, including a lead or app install. CAC specifically refers to customers under a defined rule.
How does CAC relate to LTV?
CAC is acquisition cost; LTV is estimated relationship value. Compare margin-based, unit-matched measures and consider payback timing.
Can CAC be negative?
Not ordinarily, although net promotional credits or accounting reversals require explanation. Negative values usually signal a data or allocation issue.
Why is incremental CAC hard to measure?
It requires knowing which customers would have arrived without the activity, which attribution reports alone cannot establish.
Summary
CAC is acquisition spending divided by new customers under an explicit definition. It is trustworthy only when the cost boundary, customer event, unit, cohort timing, attribution rule, and refunds are clear. For experiments, favor incremental customers and durable value over cheap attributed conversions alone.
Sources
- U.S. FTC: Internet advertising and marketing rules
- Recurly: SaaS metrics guide
- NIST: Confidence intervals