How we set a realistic cost-per-acquisition target before spending a dollar

Most cost-per-acquisition targets we inherit from a new client were picked because they sounded reasonable, not because anyone checked them against the actual unit economics of the business. That's usually the first conversation before any ad account gets touched, because a wrong target makes every later optimization decision wrong too, no matter how well the campaigns themselves are run.

A sustainable CPA target is a function of customer lifetime value and how much margin you're willing to spend acquiring that value up front. For a subscription business with a strong retention curve, spending close to or even above first-purchase margin to acquire a customer can be entirely rational. For a low-repeat, low-margin product, the same CPA would be a fast way to lose money at scale.

We ask for actual retention and repeat-purchase numbers before setting a target, not because we distrust the client's instinct, but because the instinct is usually anchored to a competitor's public numbers rather than this specific business's economics, which are almost never identical to a competitor's even in the same category.

Once lifetime value sets the ceiling, we work backward through the actual conversion funnel: site conversion rate, close rate if there's a sales step, average order value. That produces a target cost per click and cost per lead that the campaign has to hit, which is a very different exercise than picking a CPA that looks good on a dashboard without any connection to what the business can actually afford.

New accounts almost always start above the eventual target while the algorithm and creative both find their footing. We set an explicit ramp: an initial acceptable range for the first four to six weeks, tightening toward the real target as data accumulates. Clients who expect the final number on day one usually panic and pull budget right when the account is starting to learn, which resets the learning process and makes the actual target harder to reach, not easier.

Sometimes the honest answer, after a fair testing period, is that the target CPA isn't achievable at meaningful volume in this channel for this product. We'd rather say that directly and discuss either adjusting the target, the offer, or the channel mix, than keep spending against a number that was never realistic to begin with. That conversation is uncomfortable, but it's a lot cheaper than months of underperforming spend.

For a subscription box client with a $45 average order and 70 percent of customers renewing at least twice, lifetime value came out closer to $110. That meant a CPA of $50, which looked alarming compared to the client's initial instinct of a $20 target, was actually well within a profitable range once the full retention curve was accounted for.

Competitor CPA figures that circulate in industry conversations are rarely apples to apples, since they depend on that specific competitor's margin structure, retention, and average order value, none of which are usually disclosed alongside the number. Anchoring your target to someone else's unverified figure is a common and avoidable mistake.

For clients with real seasonal demand swings, we set a range rather than a single fixed number, since a target that's realistic in a high-demand month can be unrealistic in a slow one, and forcing the same number year-round leads to either wasted spend or unnecessarily conservative budgets during the strong months.

Occasionally a client wants to hold a CPA target we don't believe the market supports. We'll run the campaign to their number for an agreed test period, with clear reporting on what volume that target actually produces, so the conversation about adjusting it is grounded in real data rather than a disagreement of opinion.

Brand new businesses sometimes have no purchase history to calculate lifetime value from at all. In that case we build a conservative estimate from comparable businesses in the same category, clearly labeled as an assumption to be replaced once the business has three to six months of its own real data, rather than pretending a first-principles number is more certain than it actually is at that early stage.

A CPA target is meaningless without agreeing on what counts as an acquisition, a completed purchase, a qualified lead, a free trial signup. We've seen real confusion arise from a client and an agency silently using different definitions of the same term, which makes any target essentially unmeasurable until everyone is using the same specific definition.

Broader advertising cost trends, driven by platform-wide competition and macroeconomic conditions, shift what a reasonable CPA looks like even for a business whose own unit economics haven't changed at all. We revisit targets at least twice yearly specifically to account for this external drift, rather than treating a target set eighteen months ago as still automatically valid today.

We also build a simple, shared reference document explaining the underlying lifetime value math in plain terms, so a client's broader internal team, not just the specific marketing contact we work with most closely, understands why a given target was set the way it was. This shared understanding has repeatedly prevented internal second-guessing later, when a different stakeholder unfamiliar with the original reasoning sees a CPA number in isolation and questions it without the full context needed to actually evaluate it fairly.

A blended CPA across all channels can look healthy while masking one channel that's badly overspending and another that's underspending relative to its actual ceiling. We report both numbers separately from the start, since a client making budget decisions off the blended figure alone will often keep funding a channel that looks fine only because a cheaper channel is quietly subsidizing the average.

This distinction matters most during periods of budget growth, when the instinct is to simply scale whatever's working. Scaling a channel past its natural ceiling usually raises its CPA faster than the blended number reveals at first, and by the time the blended figure moves enough to prompt a conversation, real budget has already gone to spend that was never going to hit target.

We report channel-specific CPA against channel-specific targets from the very first weekly report, precisely so this kind of quiet drift gets caught early rather than surfacing months later as an unexplained dip in the blended average that takes real digging to trace back to its actual source.