When product growth destroys its economics

Revenue rose, more customers arrived and sales beat the target. So did the implementation queue, urgent support requests and service costs. At month-end, less cash remained than before the successful campaign.

Growth amplifies the model you take to market. If new customers receive value through expensive manual work, more sales first scale that work. Before raising the budget, examine the next group of customers separately from the accumulated base.

Average CAC can reassure you for too long

Consider a hypothetical SaaS. Referrals bring 100 customers at ₽8,000 each. A paid channel brings another 100 at ₽24,000 each. Average CAC — the cost of acquiring a paying customer — is now ₽16,000.

That is the correct average for both groups. It does not mean additional customers will also cost ₽16,000. The latest available group cost ₽24,000, and further expansion may change the cost again.

Use an agreed cost scope, such as advertising and the team effort attributable to those sales. Otherwise, a costly channel involving lengthy negotiations will look more efficient than it is.

Compare new cohorts over the same horizon

Suppose that in the first six months, each referral customer contributes ₽30,000 before acquisition costs, while a paid-channel customer contributes ₽20,000. Service costs are included; fixed company costs are not yet deducted.

The first group left ₽2.2 million; the second has yet to recover ₽400,000 of acquisition costs. The overall picture may remain positive while hiding deterioration in the latest cohort. This does not prove the channel is permanently unprofitable: repeat payments may follow. But future revenue must be validated or explicitly kept as a forecast.

Also check whether the customers are really comparable. Paid advertising may have attracted a different usage scenario that requires more support and offers fewer reasons to return.

Find the customers worth acquiring again

ABCDX segmentation analyses the existing customer base through product fit, satisfaction and margin. Groups A and B help investigate successful customer–offer combinations; C and D reveal weaker fit, while X points to needs beyond the product’s current use.

The practical question is what you can learn about a new customer before the deal to anticipate a suitable scenario. Labels such as A or C do not answer that. You need contextual traits that can be checked and connected to subsequent use.

Do not extrapolate from a few successful contracts to the whole market. With a small base, the profile remains a hypothesis. And weak product fit is not a judgement of the customer: your offer, pricing or service model may be the problem.

A service has capacity limits

Economic models often assume constant unit costs. In reality, customer 101 may require another specialist, separate infrastructure or a new process. The spreadsheet looks good before that threshold and quite different after it.

Test growth against a specific scarce resource. How many implementations can the team finish each week? How long do customers wait for first value? What happens to support quality at peak load? When sales outpace implementation, new contracts temporarily become a queue of promises.

The next move may be standardising one scenario, charging for a clearly scoped launch or tightening sales qualification. Abruptly cutting support for existing clients is not a universal answer: obligations, trust and retention still matter.

Increase budgets according to a testable rule

Before expanding the channel again, agree on acceptable acquisition cost, successful-launch rate, contribution, payback time and service load. Set the evaluation horizon and the stopping point.

Then increase investment in bounded steps and watch new cohorts. The team can see when the old model stops explaining reality. Growth becomes manageable when the company can explain why the next customer moves it towards its goal, accounting for every resource needed to fulfil the promise.