What it means
Growth costs money, and CPA is the price. Every advertisement, campaign, salesperson and promotion is spent to bring in customers, and dividing the spend by the customers it brought gives the cost of each.
The figure is simple to calculate for a single digital campaign, where the platform records spend and conversions; harder for a business with many channels and a sales force, where attribution of a customer to the activities that won it is a judgement; and hardest at the level of the whole business, where the question is what the total sales and marketing machine costs per customer it produces. The measure is meaningful only against value.
A CPA of $50 is excellent if the customer's first order contributes $120 and terrible if it contributes $30 and the customer never returns. Businesses compare CPA with first-order contribution (does the acquisition pay for itself at once), with the contribution over a payback period (how many months until the customer has covered the cost of winning them), and with customer lifetime value (the total contribution expected over the relationship).
The ratio of lifetime value to CPA is the standard test of a growth model: below 1, every customer loses money; around 1 to 2, the business is buying revenue at a loss once overheads are counted; 3 or more is generally considered healthy, with the payback period as a second check because a 3:1 ratio that takes five years to realise strains cash. CPA differs sharply by channel.
Referrals and organic search are cheapest; email to existing contacts is cheap; search advertising is moderate and scales with bid competition; social and display are variable and often expensive for direct response; outbound sales is expensive per customer but appropriate for high-value accounts; events, sponsorships and brand advertising have CPAs that are hard to measure and long to realise. A marketing budget is allocated by comparing CPA and value across channels and moving spend to the margin where the next dollar buys the most value, subject to each channel's capacity (organic search cannot be bought; the best keywords run out; referrals depend on the customer base).
CPA also rises with scale. The first customers from a channel are the cheapest (the most interested audience, the best keywords, the warmest leads); each additional cohort costs more as the campaign reaches less interested people.
Marginal CPA, the cost of the next customer, is what matters for the decision to spend more, and it is usually above the average CPA that reports show. A channel whose average CPA is comfortably below value may have a marginal CPA above it, and the spend should stop before the average suggests.
Definitions must be fixed and honest. What counts as an acquisition (a lead, a trial, a first order, a customer who has paid)?
What counts as cost (media only, or agency fees, tools, staff, discounts and free trials too)? Over what period (the month's spend against the month's acquisitions, when the campaign's effect lags)?
A CPA that excludes half the costs or counts leads as customers flatters the channel and misdirects the budget.
In practice
Real-world examples.
Example
A software company's fully loaded customer acquisition cost is $9,000 against an annual contribution of $6,000 per customer and a five-year life, giving a payback of 18 months and an LTV to CAC ratio of 3.3.
Example
An online retailer finds that its CPA for customers acquired through discount sites is $12 but their lifetime value is $18, while search-acquired customers cost $40 and are worth $160.
Example
A bank's cost per new current account through branch referrals is $45 and through paid digital channels $210, and it redirects budget to branch incentives.
Think of it
“CPA shows your all-in cost to win each new customer-the price of customer acquisition.
Formula
Calculation
Cost per Acquisition = Acquisition spend / Number of acquisitions
Customer Acquisition Cost (fully loaded) = (Advertising + Sales salaries and commissions + Marketing staff + Tools and agencies + Promotional discounts and free trials) / New customers acquired
Payback Period (months) = CPA / Monthly contribution per customer
LTV to CPA Ratio = Customer lifetime value / CPA
Marginal CPA = Additional spend / Additional acquisitions from that spend
Worked example. A subscription meal-kit company acquires customers through four channels in a quarter. Average first-order contribution $22; monthly contribution per retained customer $35; average customer life 11 months; lifetime value $385.
- Search advertising: spend $180,000; new customers 3,600; CPA $50. Payback 1.4 months; LTV to CPA 7.7
- Social advertising: spend $240,000; new customers 3,000; CPA $80. Payback 2.3 months; LTV to CPA 4.8
- Referral programme: $30 credit to referrer and $30 to referred, 2,500 customers: spend $150,000; CPA $60. Payback 1.7 months; LTV to CPA 6.4
- Television: spend $400,000; attributed new customers 2,200 (measured by uplift over baseline in the campaign period): CPA $182. Payback 5.2 months; LTV to CPA 2.1
Fully loaded: add marketing team $120,000, agency fees $60,000, tools $20,000, and the first-box discount (average $15 on 11,300 customers, $169,500): total spend $1,339,500; total new customers 11,300; blended CAC $118.50. Payback 3.4 months; LTV to CAC 3.2.
Marginal analysis: the search channel's spend was increased from $120,000 to $180,000 during the quarter; the additional $60,000 produced 900 additional customers: marginal CPA $67, still well below value. Social's increase from $160,000 to $240,000 produced 700 additional customers: marginal CPA $114, above the average and approaching the point where payback exceeds three months. Television's marginal CPA cannot be measured within the quarter.
Reallocation for the next quarter: search up to $240,000 (expected marginal CPA rising to about $80, still attractive); social held at $200,000 (trimming the least efficient campaigns); referral credit raised to $35 each (expected to lift referrals 20% at a CPA of $70); television cut to $250,000 and measured against a two-quarter attribution window rather than one, since its effect on brand search and referrals appears later. Expected blended CAC about $105 with the same customer volume.
Definition check: the company's marketing team had been reporting CPA on media spend only ($970,000 / 11,300 = $86) and counting a customer at first order. The finance team's fully loaded figure ($118.50) and its insistence on counting only customers who paid for a second box (which cut the count to 9,800 and raised CAC to $137) changed the picture: the LTV to CAC ratio fell from 4.5 to 2.8, and the board's growth plan, which assumed 4.5, was revised.Case study
Seen in the real world.
A direct-to-consumer mattress company grew from nothing to $60,000,000 of revenue in three years on a marketing spend that rose to $28,000,000, with its board tracking CPA on media spend ($210 per order against an average order of $900 at a 45% gross margin, $405 of contribution) and concluding that acquisition was highly profitable. A new chief financial officer built the fully loaded figure: media $22,000,000; agency and production $2,500,000; marketing staff $1,800,000; free trial returns and refurbishment (a 100-night trial with a 14% return rate) $3,900,000; promotional discounts $4,200,000; and affiliate commissions $1,600,000. Total $36,000,000 against 66,000 orders: $545 per order, above the $405 of contribution.
Every order lost $140 before overheads, and the company was selling mattresses that customers rarely bought twice. The board had believed the business was scaling into profit; it was scaling into loss, and the difference was the $14,000,000 of acquisition cost that had never been counted as acquisition cost.
The company cut its media spend by 40%, focused on the channels with the lowest fully loaded CPA (search and referrals), reduced the trial period and tightened return handling, and grew revenue only 5% the following year while reaching a positive contribution per order for the first time. The chief financial officer's board paper was titled "What a customer actually costs."
Watch out
Common mistakes.
- Calculating CPA on media spend alone, excluding agency fees, staff, tools, discounts, free trials and returns, which understates the true cost by a third or more.
- Comparing CPA with revenue rather than with contribution, and with first-order value when the business depends on repeat purchases that may not come.
- Scaling a channel on its average CPA when its marginal CPA has already risen above the value of the customers it brings.
Questions
People also ask.
What is a good cost per acquisition?
One that is comfortably below the value the acquisition brings: for first-order businesses, below first-order contribution; for repeat businesses, a lifetime value at least three times CPA and a payback period the business can fund, commonly under twelve months.
What is the difference between CPA and customer acquisition cost?
CPA usually refers to a specific campaign or channel's cost per conversion. Customer acquisition cost is the fully loaded cost of all sales and marketing per new customer. Both are used, and the definition should be stated.
How should CPA be attributed across channels?
Imperfectly. Last-click attribution over-credits the final step; multi-touch models spread credit; incrementality tests (holding a channel out and measuring the difference) are the most reliable and the most expensive. The business should choose a method, apply it consistently and remember its limits.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%