The Next Customer Does Not Inherit the Average.

The channel can remain profitable on average while every additional dollar destroys value. That is not a contradiction. It is what happens when a historical average is asked to price a future edge.

The first customers were not random samples from an endless market. They were the people easiest to reach, quickest to recognize the offer, and least expensive to persuade.

A growth model that treats their economics as a permanent property of the channel does not forecast scale. It memorializes the favorable beginning.

Model One: The Average Customer

The average model compresses every acquired customer into one composite figure. Total acquisition spend is divided by total customers. Total contribution is divided by the same population. If the resulting relationship clears the company’s threshold, the channel receives more capital.

This is a useful description of what already happened. It becomes dangerous when it is used as the price of what happens next.

Consider a fully illustrative example. A company spends $100,000 and acquires 1,000 customers. Average acquisition cost is $100. Their first-year contribution before acquisition cost averages $180, leaving $80 per customer against the initial spend. The cohort looks strong, and the next budget increase appears obvious.

Now add $50,000. The increment acquires 300 customers, or roughly $167 each. Their first-year contribution averages $140. The new spend produces $42,000 in contribution against $50,000 in acquisition cost: an illustrative loss of $8,000 before shared overhead.

Blend the cohorts, however, and the channel still looks healthy. Total contribution is $222,000 against $150,000 in spend. Average acquisition cost is about $115, and average first-year contribution is about $171.

The successful center has absorbed the failing edge.

Model Two: The Marginal Customer

The marginal model refuses that absorption. It asks what the next dollar buys, not what every previous dollar bought together.

Analysis: acquisition curves commonly bend in two places. The visible bend is cost. Reach expands, response weakens, frequency accumulates, and the next customer requires more spend. The less visible bend is quality. Wider distribution may reach people with weaker need, lower urgency, heavier incentive dependence, smaller purchases, higher service demands, or shorter retention. The business can pay more to acquire a customer who contributes less.

These bends do not need to occur in every channel or at the same point. That uncertainty is the reason to model the increment separately. If management knows only the blended result, deterioration becomes visible late—after the profitable base has hidden several weak rounds of expansion.

The right unit is not “paid social,” “partners,” or “search.” Those labels are too broad for a capital decision. The unit is the next defined block of spend under specific conditions: audience, geography, offer, incentive, placement, sales capacity, and time window. Change the conditions and the curve may reset. Leave them unchanged and history should not be allowed to flatter the next purchase.

Scale Changes the Buyer

The average model assumes more volume reveals more of the same customer. The marginal model allows scale to change who arrives.

That distinction matters beyond advertising efficiency.

A lower-intent cohort can lengthen sales work, require more explanation, use more support, choose a smaller entry offer, or leave before acquisition cost is recovered.

A discount can improve conversion while teaching the model to find customers whose demand depends on the discount.

A new distribution partner can widen reach while weakening the context that made earlier buyers understand the value.

Analysis: the commercial risk is not merely rising acquisition cost. It is identity drift inside the customer base. The company believes it is scaling a proven market when it may be purchasing a different market with different economics.

OrionPilot’s Strategy Summary can make that distinction explicit before weekly campaigns multiply: preserve the proven audience and its evidence, then name the hypothesis behind the next audience, the cost ceiling it must meet, and the customer-quality signals that can revoke the spend. The campaign is then funding a test of marginal demand, not celebrating a blended past.

Put the Stop Rule Before the Budget

A serious growth budget should contain its own refusal. Before the next increment launches, define the smallest spend block that can be judged, the contribution window, the quality indicators, and the threshold that stops expansion. Do not wait for the channel average to fail. By then, new losses may have grown large enough to contaminate a still-profitable history.

The executive question changes from “Is this channel working?” to “Under what conditions does the next tranche deserve to exist?” That wording moves authority from the dashboard to the investment decision. It also makes a pause interpretable. Capital withheld at the edge is not fear of scale. It is evidence that the company can distinguish a large profitable base from one more unprofitable customer.

The center of the field can remain green for a long time after the outer ring stops carrying water. The disciplined decision is made before that ring is planted.