Paid Acquisition CAC/LTV Dynamics in Mature Markets

The two curves that govern paid acquisition

A paid acquisition engine is governed by two curves. The first is the response curve of the channel: as the company spends more on the channel, the marginal cost of acquiring the next customer rises. The shape of the response curve is determined by the saturation of the addressable market (the higher the saturation, the steeper the curve), the cost-per-click of the channel (the higher the CPC, the steeper the curve), and the conversion rate of the channel (the lower the conversion rate, the steeper the curve). The response curve is a curve, not a number; the company’s operating decisions are made on the slope of the curve, not on the point estimate of the channel’s reported ROAS.

The second is the retention curve of the cohort: as the cohort ages, the cumulative revenue per customer rises, but at a declining rate. The shape of the retention curve is determined by the churn rate (the higher the churn, the steeper the decline), the price increase cadence (the higher the cadence, the higher the cumulative revenue), and the upsell/cross-sell motion (the more aggressive the motion, the higher the cumulative revenue). The retention curve is also a curve, not a number; the company’s operating decisions are made on the slope of the curve, not on the point estimate of the cohort’s reported LTV.

Why naive ROAS is misleading

Naive ROAS is the ratio of the channel’s reported revenue to the channel’s reported spend, computed at the end of the reporting period. The number is, by construction, a backward-looking number: it tells you what the channel did, not what the channel will do. The number is also a channel-local number: it tells you the channel’s ROAS, not the company’s ROAS on the cohort the channel produced.

The bias is in the direction of over-optimism. A channel that converts a customer on day 1 and the customer churns on day 90 has a 90-day ROAS that is much higher than the cohort’s 24-month LTV/CAC. The channel is credited with a customer that the company loses, but the loss is on a different time horizon and a different reporting system. The two never reconcile in the channel’s report.

The bias is amplified in mature markets. A mature market is, by definition, a market where the addressable market is approaching saturation. As the company spends more, the marginal CAC rises; the cohort churns faster (because the company is converting less-qualified buyers); the LTV is lower. The channel’s reported ROAS is, on a 90-day horizon, flat or rising; the company’s 24-month LTV/CAC is, on the same horizon, falling. The two never reconcile in the channel’s report.

The right metrics for a category leader

The right metric is the marginal CAC: the cost of acquiring the next customer, computed on the next dollar of spend. The marginal CAC is the slope of the response curve at the current spend level. A company that is operating at a marginal CAC equal to the cohort LTV is at the breakeven point on the next dollar of spend; a company that is operating at a marginal CAC less than the cohort LTV is making money on the next dollar; a company that is operating at a marginal CAC greater than the cohort LTV is losing money on the next dollar. The operating decision is to spend up to the point where the marginal CAC equals the cohort LTV, and to stop spending at that point.

The right metric for the cohort is the 24-month cohort LTV, discounted to present value. The 24-month horizon is long enough to capture the typical B2B SaaS renewal cycle (annual contract, with 80%+ renewal at the first renewal, 90%+ at the second, 95%+ at the third), and short enough to be operationally useful (the company can see the cohort’s performance in time to make the next quarter’s decisions). The present value is the right basis because the company’s operating decisions are made in present-value terms, not nominal terms.

The operating cadence

For a category leader, the right operating cadence is quarterly. Once a quarter, the company recomputes the marginal CAC by channel and the 24-month cohort LTV by cohort. The output is a one-page operating report: the marginal CAC by channel, the 24-month cohort LTV by cohort, and the implied operating decision (spend more, spend less, hold) for each channel.

The cadence is quarterly because the marginal CAC and the cohort LTV are both slow-moving. The marginal CAC changes with the response curve, which changes with the saturation of the addressable market, which changes on a quarter-to-quarter horizon (the company is spending on a steady cadence, the addressable market is growing on a quarter-to-quarter horizon, the saturation is changing on a quarter-to-quarter horizon). The cohort LTV changes with the retention curve, which changes with the cohort’s age, which changes on a quarter-to-quarter horizon. The operating cadence is the cadence of the underlying curves, not the cadence of the channel’s reporting.

A worked example

Consider a category leader with a 10,000-customer addressable market, a 20% market share, and a paid acquisition engine that has been operating for 24 months. The channel’s reported ROAS on a 90-day horizon is 4.0x; the company’s 24-month LTV/CAC on the same cohort is 2.5x. The channel is reporting a 60% overstatement of the company’s return.

The cause is the channel’s reporting horizon. The channel is reporting on the 90-day window after the click; the company is paying on a 24-month window. The 24-month window is the right window because the cohort’s churn is non-trivial in months 4–18, and the 90-day window does not see the churn. The company is, in effect, paying for customers that the channel reports as profitable but the company’s P&L reports as breakeven.

The operating response is to recompute the marginal CAC. The marginal CAC is the cost of acquiring the next customer, computed on the next dollar of spend. The company runs the recomputation on the next quarter’s spend plan: the channel’s marginal CAC is 1.4x the cohort LTV, the channel is operating above breakeven on the next dollar, and the channel should be cut by 30% over the next two quarters. The cut is the operating response; the recomputation is the operating cadence.