B2B Enterprise ABM & Demand Generation EnginePlaybook3 min readUpdated September 2026

Forecasting Marketing ROI Without Fooling Yourself

A trustworthy marketing ROI forecast shows a low, expected and high case instead of one confident number. The link between spend and pipeline is a range built on assumptions that can each be wrong in either direction, so a forecast that shows its uncertainty is more credible with a board than one that hides it.

Why forecast marketing ROI as a range instead of one number?

Instead of forecasting a single pipeline number for a given spend level, forecast a low, expected, and high case, built from the actual range you've seen in your last several quarters of cost-per-pipeline-dollar by channel. If your paid channel's efficiency has ranged between two numbers over the last year, your forecast should carry that same range forward rather than picking the midpoint and presenting it as if it were certain. This single change, showing a range instead of a point, does more to make a forecast honest than any amount of additional modeling sophistication layered on top of a single confident number.

Separate the Assumptions You're Confident In From the Ones You're Guessing At

Every forecast blends things you actually know, your historical conversion rates, your average deal size, with things you're genuinely guessing at, whether a new channel will perform anywhere close to your existing ones, whether a planned price change will affect volume. List these separately rather than blending them into one model where a shaky guess quietly inherits the same apparent confidence as a well-supported historical number. Flagging the guesses explicitly means everyone reviewing the forecast knows exactly which part of it to question first if the actual results come in different than expected.

How should you build the new-channel assumption in a forecast?

If your forecast includes a channel you haven't run at scale yet, don't forecast its performance from an industry benchmark or a competitor's claimed results. Run a small, capped test first, even a modest budget over a few weeks, and forecast the channel's contribution from that real data, adjusted for the fact that a small test often performs differently than a scaled-up version once you're spending several times as much. A forecast built on your own early signal, even an imperfect one, will be more defensible later than one built on someone else's number from a completely different business.

Revisit the Forecast Monthly Against Actuals, Not Just at Quarter Close

A forecast checked only once at the end of the quarter tells you that you were wrong three months too late to do anything about it. Compare actuals to your forecast's range monthly, and specifically note whether actuals are tracking toward the low end, the expected case, or the high end of what you projected. If actuals are consistently landing outside even the low end of your range, that's a sign your underlying assumptions, not just this quarter's execution, need revisiting before you build next quarter's forecast the same way.

Present the Forecast With Its Range Intact, Not Collapsed Into One Number

When it's time to share the forecast with leadership or the board, resist the pressure to collapse the range back down into a single confident number because a range feels less impressive to present. A forecast presented as pipeline will land somewhere between this and that, with our best estimate here, and here's what would move it toward either end, is more useful to a board making a real resourcing decision than a single number that will inevitably be wrong in one direction or the other and invite an uncomfortable conversation about why.

Keep a Running Log of Which Assumptions Were Right or Wrong

After each quarter closes, go back to the forecast and note which specific assumptions held up and which ones didn't, rather than just noting whether the total number was close or far off. A total that landed near the expected case can still hide one assumption that was badly wrong and another that happened to be wrong in the opposite direction, canceling out in the final number while teaching you nothing about which part of the model to trust next time. This log, kept quarter over quarter, becomes the most valuable input into how much confidence to put in each assumption going forward, more valuable than any amount of additional modeling sophistication applied to a fresh forecast built the same way as the last one.

To build a forecast that shows its uncertainty:

  1. Forecast a low, expected and high case from the range you've seen in recent quarters of cost per pipeline dollar by channel.
  2. List the assumptions you're confident in separately from the ones you're guessing at.
  3. Build any new-channel assumption from a small, capped test instead of an industry benchmark.
  4. Compare actuals to your range monthly and note which end of the range they're tracking toward.
  5. Present the forecast to leadership with its range intact, alongside your best estimate.
  6. After each quarter, log which assumptions held up and which didn't.
Executive Capability Standard

What Good Looks Like

A mature ROI forecasting practice presents a range built from real historical variance, flags which assumptions are guesses, and gets checked against actuals monthly rather than only once a quarter.

Building The Capability (5-Stage Skill Ladder)

1. Learn:Learn your actual historical range of cost-per-pipeline-dollar by channel over the last several quarters before building any forward-looking forecast.
2. Do Manually:Build the low, expected, and high case by hand in a spreadsheet for your two largest channels before modeling the rest.
3. Delegate:Assign a RevOps or finance partner to check actuals against the forecast monthly and flag drift outside the projected range.
4. Automate:Set up a standing report that updates the forecast's range automatically as new spend and pipeline data comes in each month.
5. Buy:Bring in outside financial modeling help if you're forecasting a genuinely new channel or a major pricing change with no reliable internal history to build from.

How to Get Started

Frequently Asked Questions

How wide should the range be in a marketing ROI forecast?

Wide enough to reflect your actual historical variance, not an arbitrarily narrow range chosen to look more precise. If your channel's cost-per-pipeline-dollar has genuinely varied by a wide margin over the last few quarters, a narrow forecast range is dishonest regardless of how confident it makes the presentation look.

What if leadership wants one number, not a range?

Give the expected case as the headline number, but keep the range visible alongside it rather than dropping it entirely. Frame the range as the tool for deciding how much risk the plan can absorb if results land toward the low end, which is usually a more useful conversation than debating the single point estimate itself.

How do we forecast ROI for a channel we've never run before?

Run a small, time-boxed test first and build the forecast from that real result rather than an industry benchmark or a competitor's reported number. Adjust for the fact that early results at a small budget often shift once you scale spend up, and say so explicitly in the forecast rather than assuming the small test's rate holds indefinitely.

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