Measurement

How to measure marketing ROI when your systems disagree

Published by Zubair Afzal, responsible editor

First published September 8, 2026Updated October 8, 20268 min read

How does your own site measure up?

A free SEO audit: an instant preview with no sign-up.

Why ROI arguments happen

Why marketing and finance disagree on marketing ROI

Marketing and finance usually disagree about definitions, not the formula. What people dispute, often without noticing, are four things underneath. What counts as return? What counts as cost? Which period does the money belong to? Would the sale have happened without the spending?

That is why better dashboards don’t settle the argument. If marketing divides revenue by media spend and finance divides gross profit by media, fees, production and a share of salaries, they will get different numbers forever, and both will be correct.

So the useful step is agreeing definitions, and that happens in a meeting, not in a tool.

What makes up the return

The four numbers behind marketing return, and who controls each

Gross profit from marketing broken into four multiplied numbers. — Recorded enquiries
The only number most marketing reports measure directly.
Close rate
Controlled by sales. A change here moves reported marketing ROI by itself.
Average order value
Controlled by pricing and product mix, and often seasonal.
Gross margin
Controlled by finance. Using revenue instead makes every channel look profitable.

The method

How to measure marketing ROI in six steps

The first two steps are conversations. Skip them and the other four settle nothing.

Agree what counts as return
Choose revenue, gross profit or contribution after variable costs, write it down, and say whether it includes repeat purchases within a set window. Gross profit is the safer default: a channel can bring in revenue at a margin below its own cost and lose money while still showing a positive return on ad spend.
You get: One written definition of return, signed off by finance
Agree what counts as cost
Include media, agency fees or staff time, production, tools and any platform minimums. Internal hours are the most common omission, which makes in-house work look free and skews every comparison against outside suppliers. Decide once, apply it everywhere, and write down what you exclude.
You get: A cost basis, with exclusions listed
Fix your tracking before you analyse anything
Check what counts as a conversion and whether it fires once or every time, what happens when a visitor declines cookies, and whether a checkout on another domain breaks the session. Analysis built on broken tracking gives confident, precise, wrong answers, with no warning.
You get: A written definition of every tracked event and where it fires
Match revenue to the month you spent the money
Link revenue to the month of the spending, even if the money arrived later. Mark recent months as incomplete and say how long they usually take to mature. For any considered purchase, this often flips the conclusion: judged by calendar month, a slow channel always looks worse than a fast one.
You get: A report grouped by spend month, with recent months flagged as incomplete
Check the numbers against your accounts
Compare the number of sales your business recorded with what your marketing systems claim. They won’t match. Know how big the gap is and which way it runs, and state it in your report. A marketing number never checked against the accounts won’t stand up in a budget meeting.
You get: A stated gap between the marketing and finance totals
Run one holdout test a year
Pick one significant line of spend, pause it in one region or segment for a set period, keep everything else the same, and compare results with a matched group. This is the only step that shows what your spending caused, as opposed to what happened alongside it.
You get: One documented test with a matched control group and a written conclusion

Common mistakes

Four common marketing ROI mistakes

Each gives a number that is consistent, repeatable and wrong, which is why they last so long.

Platform totals were added together.: Each ad platform reports every conversion it can link to itself, so one purchase touched by two platforms gets counted twice, by design. Add the columns together and you get more sales than the business made, and a return inflated by exactly that overlap.
Revenue was used where margin belonged.: Dividing revenue by spend makes almost every channel look like a success, because it ignores the cost of delivering what you sold. At a twenty per cent margin, a campaign returning four times its ad spend loses money once you add fees and production. People quote the number their tool shows by default.
A slow channel was judged on a calendar month.: Search and brand activity often bring in revenue months after the spend; channels that capture existing demand bring it in within days. Judge both on the same calendar month and the slow channel gets cut. Then the fast channel slowly declines, because nothing is creating the demand it was capturing. Matching revenue to the spend month prevents this.
Nobody asked what would have happened anyway.: Branded search is the clearest example. It converts well and costs little, but some of those people were going to find you regardless. The reported figure is accurate about what happened and says nothing about what caused it. Only a holdout test can separate the two.

Measurement methods

Four ways to measure whether marketing worked

These answer different questions, cost different amounts and fail in different ways. A solid measurement setup uses several.

Platform reporting, analytics attribution, asking the customer and holdout tests compared on the question each answers, the cost, and how each fails
Dimension	The question it can answer	What it costs to run	How it fails
Platform reporting	Did activity on this platform happen alongside sales it could see?	Nothing extra. It is already set up and already in your meetings.	It overlaps with every other platform and can’t see anything outside itself.
Analytics attribution	Among the touchpoints we recorded, how should we split the credit?	Setup and upkeep, plus the discipline to tag every campaign the same way.	It misses touchpoints it never saw, and cookie consent and data thresholds affect it.
Asking the customer	What does the buyer think brought them to you?	One form field and the patience to read free-text answers.	Individual answers are unreliable. People favour the most recent thing and whatever is easy to name.
Holdout tests	What changes when this activity stops, compared with a matched group?	Lost revenue for a set period, plus the discipline to leave the test alone.	Expensive and slow, and it only answers the one question it was set up for.

What to stop doing

ROI measurement tasks that waste time

Businesses spend hours on every one of these, and none of them changes a decision.

Trying to make platform totals match to the penny. They were never meant to agree. Know the size of the gap, state it and move on.
Rebuilding the attribution model because the current one makes you look bad. Changing the rule rewrites your history, and unless you reset targets at the same moment, comparisons stop meaning anything.
Reporting a return for a period that isn’t complete. For a considered purchase, last month isn’t a result yet, and publishing it as one guarantees an argument two quarters later.
Quoting an industry benchmark for return. The businesses, definitions and time period behind those numbers are almost never stated, which makes them unusable.
Building a custom attribution system before checking your tracking. Clever analysis of unreliable data just produces confident nonsense faster.
Counting a lead as a return. A lead is the start of a sales process with its own conversion rate. Treating it as revenue takes the argument somewhere finance can’t follow.

Questions

Marketing ROI questions from marketing and finance

What is the formula for marketing ROI?

Marketing ROI is return minus cost, divided by cost. The sum is never the problem. "Return" and "cost" each have several sensible definitions, and the people arguing are usually using different ones.

Write both definitions down before you calculate anything. Return is normally gross profit from revenue you can link to marketing. Cost is normally media plus agency fees or salaries, production and tools; media alone understates it. Only an agreed pair of definitions ends the argument.

What is the difference between ROAS and ROI?

ROAS is revenue divided by advertising spend. ROI is profit minus total cost, divided by total cost. They answer different questions, and on the same campaign they can point in opposite directions.

A campaign returning four times its ad spend, at a twenty per cent gross margin with fees and production counted, is losing money. Trouble starts when you quote one number while your audience is asking about the other.

Why do our platform conversion totals add up to more sales than we made?

Each platform counts every conversion it can link to itself, with its own attribution window and counting rules, and none knows what the others counted. So one purchase touched by two platforms gets counted by both.

Some platforms also report modelled conversions, estimating sales that can’t be tied directly to a click. That is documented behaviour, but it means platform totals were never meant to be added together. Your accounts are the only reliable record of how many sales you made.

Our sales cycle is nine months. How do we measure anything?

Group revenue by when you spent the money, not by calendar month, linking each sale back to the period of the spend even if the invoice arrived much later. The current quarter is always incomplete; label it that way.

Meanwhile, track early signs that tend to predict the final result, such as qualified enquiries, new opportunities and deals moving through sales stages. Report them separately from results, so nobody mistakes activity for revenue.

How do we account for word of mouth and offline influence?

Your analytics can’t. Nothing records a chat in the pub or a recommendation in a group chat; any model that seems to include them is reshuffling credit between the touchpoints it did record.

Two partial fixes help. Ask on your enquiry form how people heard about you, and treat the answers as a rough guide. And run an occasional holdout test, which captures offline influence automatically in the difference between the two groups.

Is turning a channel off to test it not just losing revenue?

It does cost something. In return you get the only evidence that separates what your spending caused from what would have happened anyway.

Scoped well, it is cheaper than it sounds: pause the channel in one region or segment for a set period and leave the rest running. A few weeks of partial pause is a small price for a budget line that otherwise runs every month with no end date.

References

Sources for how to measure marketing ROI

The official documents and research this page is based on. Platform rules change, so check the source before you act on a detail.

Keep reading on how to measure marketing ROI

Start here

Agree your ROI definitions before the next review meeting

Most of this gets settled in one session with marketing and finance in the room: what counts as return, what counts as cost, and how big the gap to the accounts is. It ends an argument that keeps coming back.

Try the free UTM Campaign URL Builder

Want to see how your own site stacks up? Get my free SEO audit

Agree what counts as return

Agree what counts as cost

Fix your tracking before you analyse anything

Match revenue to the month you spent the money

Check the numbers against your accounts

Run one holdout test a year