Spotting the Saturation Point
Every performance channel has a point where extra budget brings barely any customers while the cost per lead and per customer keeps rising. Whoever doesn't know that point pours money into a channel that's already full. The saturation model shows where the limit lies - and turns the reallocation into a sober calculation instead of a gut decision.
Why "more budget = more customers" is weaker
The linear assumption breaks at the limit
At first a channel scales almost linearly: more budget, more clicks, more customers. But markets are finite. Beyond a certain point the primary metric barely rises anymore, while CPL and CAC increase progressively and the ROI collapses. Whoever keeps thinking linearly pays for customers who no longer come.
What the saturation model makes visible
The model maps the curve via α · (1 − e^(−γ · x)): α is the limit the metric approaches, γ the steepness. That makes visible where the channel flattens - and from when every extra dollar works worse than elsewhere.
How the saturation analysis works
- Set up a channel with a saturation limit. Bring the paid channel into the model with its budget and its saturation limit (α).
- Ramp up the budget and read the curve. Raise the budget step by step and watch how the primary metric flattens while CPL and CAC rise and the ROI gives way - the point where it tips becomes visible.
- Check the sensitivity. Use the sensitivity analysis (±10% and ±25%) to see which lever has the biggest pull - often a better rate brings more than additional budget.
- Reallocate or lift α. Shift the budget beyond the limit into a channel with headroom - or enlarge the market via a parallel push channel so α shifts upward and the channel breathes more freely again.
- Compare variants. Set the reallocation against the starting state via "compare plans" and pick the more viable distribution.
Make a channel's saturation limit visible and reallocate: try the tool
Talking points for the conversation
- "The channel isn't bad - it's full." Separates a saturation problem from a performance problem.
- Show rising CAC, don't assert it: the curve makes visible why the next dollar gets more expensive than the last.
- Reallocate rather than pile on: budget beyond the limit brings more in a channel with headroom - an efficiency argument, not a cutting one.
- Show the breathing via halo: a push channel can lift the limit - an alternative to simply switching off.
Common thinking traps
- Taking the saturation limit for an exact dollar amount. α and γ are estimated curve parameters, not measured. The model shows the shape of diminishing returns, not the cent-precise tipping point.
- Confusing saturation with poor performance. A saturated channel isn't weak - it's maxed out. Whoever switches it off for that reason loses the customers it still brings at the right budget.
- Assuming the lift of α is certain. That a push channel enlarges the market is modeled, not guaranteed. The shift of α is an assumption about synergy, not an assured effect.
Frequently asked questions about the saturation point
What is the saturation limit in this model?
The value (α) the primary metric of a channel approaches. The closer the budget is to α, the less each additional dollar brings - the metric flattens while CPL and CAC rise. The curve follows the form α · (1 − e^(−γ · x)).
How do I recognize that a channel is saturated?
By the combination of a flattening primary metric and rising cost per lead and per customer with falling ROI. Raise the budget step by step and the point becomes visible from which additional spend shows barely any effect.
Which lever brings more than additional budget?
The sensitivity analysis shows that. It varies the most important parameters by ±10% and ±25% and makes visible which lever acts most strongly - often a better rate rather than more budget in an already full channel.
Can the saturation limit shift?
Yes. If parallel push channels enlarge the market, α shifts upward - the channel can take more budget again without flattening early. The model factors this interplay in, but doesn't guarantee it.
Is the saturation limit a measured value?
No. α and γ are assumptions the model computes forward - from benchmarks or your own empirical values. The model shows the shape of diminishing returns, not a point measured on your data.
