Multi-Touch Attribution & Marketing Mix Modeling (MMM)Playbook3 min readUpdated September 2026

Forecasting Marketing ROI When Channels Don't Behave Linearly

The simplest way to forecast marketing ROI is to take a channel's current return and multiply it by however much more you plan to spend. It's also usually wrong, because almost no channel returns the same marginal value at twice the spend that it returned at its current level. Here's a worked example of where that simple forecast breaks and what a more honest one looks like instead.

Vendors Covered in this Article

Disclosure: We may earn a commission if you buy through some links on this page. It doesn't change what we recommend.

The linear forecast, and why it's tempting

Say a channel currently returns three dollars of attributed revenue for every dollar spent, at a monthly budget of ten thousand dollars. A linear forecast for doubling that budget to twenty thousand dollars simply assumes the same three-to-one return holds, producing a clean, easy-to-present sixty thousand dollar revenue forecast. It's a tempting number to put in a budget deck because it's simple to explain and easy to defend in the room.

The trouble shows up later, once the budget is approved on that basis and the actual results come in well below the forecast. At that point the conversation shifts from evaluating the channel to explaining why the forecast missed, which is a much harder conversation to have well.

Where the assumption actually breaks

Most paid channels have a finite pool of high-intent audience at any given moment. The first ten thousand dollars of spend typically reaches the people most likely to convert. The next ten thousand dollars has to reach further into a colder audience to find enough volume, and that audience converts at a lower rate. The result is a real return that's noticeably below the linear forecast, not because anything went wrong operationally, but because the channel was never going to sustain its current efficiency at double the volume.

Building a forecast that accounts for this

Look at your own channel's history for evidence of this curve rather than assuming a specific shape. Compare periods where spend was meaningfully higher against periods where it was lower, and check whether return per dollar held steady or declined as spend rose. If you have at least a few data points across different spend levels, you can sketch a rough diminishing-returns curve instead of a flat line, even without formal statistical modeling.

Even a rough version of this curve, three or four points plotted by hand, is more useful for forecasting than a single average return figure, because it shows you the direction and roughly how steep the drop-off has been historically, which is the piece of information a flat average throws away entirely.

A simple way to sanity check any forecast before presenting it

  • Ask what audience the incremental spend would actually reach, and whether that audience looks as strong as your current one
  • Check whether the channel has ever been run at the higher spend level before, and if so, what happened to return per dollar
  • Build a range instead of a single number: a conservative case using a lower marginal return and an optimistic case closer to the current rate
  • Present the range, not just the midpoint, so the decision-maker sees the uncertainty instead of a false sense of precision

What this means for how you talk about ROI internally

A forecast presented as a range, with the assumption behind it stated plainly, holds up much better under scrutiny than a single confident number that turns out to be wrong once the budget actually gets spent. It also protects your own credibility: a leader who consistently forecasts using the simple linear method and consistently misses on the high side will eventually have their numbers discounted by default, which makes every future budget conversation harder than it needs to be.

When a channel genuinely doesn't show diminishing returns yet

Not every channel is near its ceiling. A channel you've only run at a small scale, or one where you've deliberately under-invested relative to the audience size available, can sometimes sustain close to linear returns for a while as spend increases. The way to tell the difference isn't guessing, it's checking your own historical data for the pattern before assuming either a flat return or a steep drop-off applies to your specific situation.

Executive Capability Standard

What Good Looks Like

The standard is forecasting ROI from your own channel's historical response to spend changes, presented as a range, rather than a single number that assumes flat returns at any scale.

Building The Capability (5-Stage Skill Ladder)

1. Learn:Pull your own channel's return-per-dollar at different historical spend levels to see whether the pattern shows meaningful diminishing returns.
2. Do Manually:Build forecasts as a conservative-to-optimistic range by hand for any budget request involving a meaningful spend increase.
3. Delegate:Have a growth or performance marketing analyst own building and defending these forecasts before they go into a budget conversation.
4. Automate:Set up a recurring pull of return-per-dollar by spend level so the underlying curve updates automatically as new campaign data comes in.
5. Buy:Bring in a marketing mix modeling specialist if you need a more rigorous, statistically grounded response curve than a simple historical comparison can provide.

How to Get Started

Disclosure: We may earn a commission if you buy through some links on this page. It doesn't change what we recommend.

Frequently Asked Questions

Why does a linear ROI forecast almost always overstate results?

Because most channels have a limited pool of the highest-intent audience, and additional spend has to reach further into a colder, less likely-to-convert group. A linear forecast assumes the same efficiency holds at any spend level, which rarely matches how channels actually behave once you scale them.

How much historical data do we need to build a non-linear forecast?

A few data points at genuinely different spend levels are enough to start sketching a rough curve. You don't need a formal statistical model to notice whether return per dollar has historically held steady or declined as spend increased.

Should we present a single ROI number or a range to leadership?

A range that shows a conservative and an optimistic case, with the assumption behind each stated plainly, tends to hold up better than a single confident number, especially once actual results come in and get compared against the forecast.

About the numbers

This guide doesn't quote a sourced benchmark. Figures in it are estimates or general guidance, so check them against your own numbers.

Related Guides