Attribution and incrementality: why you need both and when to use each
By Paco Herranz · November 5, 2026 · 6 min read
Attribution tells you who touched the sale. Incrementality tells you whether the sale would have happened without you.
They are two different questions and they get mixed up, because both end up in a report with euros and channels. Attribution splits the credit for sales that have already happened among the channels that took part. Incrementality measures how many of those sales your marketing actually caused, by comparing them with a group that didn't get the campaign.
They don't compete: each one serves a different decision. The problem starts when you use attribution to decide something only incrementality can answer, such as how much budget a channel deserves next year. If you want the detail of how incrementality is measured, it is in what incrementality is and how to measure it. Here we go to the practical side: what each one decides and how to use them together.
What each one answers
| Attribution | Incrementality | |
|---|---|---|
| Question | Which channels took part in this sale and how much credit do I give each one? | How many sales would not have happened without this campaign? |
| How it is calculated | With a model that splits each conversion among touchpoints: last click, linear, position-based or data-driven | By comparing a group that gets the campaign with a control group that doesn't |
| What it needs | Following the customer journey: clicks, opens, visits and purchases tied to the same person | Holding out the control group before launch and crossing both groups with real sales from every channel |
| When you get the answer | Almost in real time, while the campaign is live | When the campaign closes, once there are enough sales to compare |
| Which decision it serves | Adjusting on the fly: which creative, keyword or sequence works best within a channel | Deciding budget: whether a campaign or channel deserves to be repeated, scaled or cut |
| Where it goes wrong | It credits sales that would have happened anyway, especially loyal customers, retargeting and brand | It doesn't say which part of the journey worked, only whether the whole drove sales |
The last row is the one that matters. Each method fails exactly where the other gets it right, which is why you can't pick one and forget the other.
An example: the same Black Friday campaign seen both ways
A brand invests €30,000 in Black Friday across three channels. The figures are illustrative and rounded, but the pattern is the one that shows up again and again when you measure.
| Channel | Spend | Attributed sales | Attributed ROAS | Incremental sales | Incremental ROAS |
|---|---|---|---|---|---|
| Customer email | €3,000 | €135,000 | 45x | €20,000 | 6.7x |
| Retargeting | €12,000 | €90,000 | 7.5x | €15,000 | 1.3x |
| Social prospecting | €15,000 | €75,000 | 5x | €40,000 | 2.7x |
With attribution, the report says all three channels pay off and prospecting is the weakest. With incrementality the picture changes:
- Email is still profitable, but far less than it looks. It reaches customers who were going to buy anyway; its real value is in the lapsed customers it wakes up, not the loyal ones.
- Retargeting barely earns back what it costs. It chases people who had already decided and takes the last click.
- Prospecting is the channel that brings the most new sales. Attribution penalises it because its buyers take longer and go through other channels before buying.
If you set the 2027 budget with the first table, you cut what grows the most and reinforce what adds the least.
When to use each one
Use attribution for quick, small decisions, within a channel and with the campaign live:
- Which creative or email subject line to keep.
- Which keywords or audiences to push up or down this week.
- At which step of the journey people drop off.
Use incrementality for the big decisions, the ones that move budget:
- Whether a campaign deserves to run again next year.
- How much money goes to each channel in the annual plan.
- Whether a discount brings new purchases or gives margin away to people who were going to buy anyway.
How to combine them
The most practical approach is to use incrementality to calibrate attribution. Once or twice a year you measure the channels and campaigns that weigh the most with a control group, and for each one you work out how much of what was attributed was incremental. In the example above, email turns 15% of what it is credited with into incremental sales; retargeting, 17%; and prospecting, more than 50%.
With those factors you correct day-to-day attribution. You keep the speed of attribution to adjust campaigns, but with numbers that look like what each channel really contributes. And at the next peak season you measure again, because the factors change with the offer, the customer base and the competition.
How to start this Black Friday
You don't need a perfect attribution model to measure incrementality. Three steps are enough:
- Pick the campaign that weighs the most in your season budget.
- Before launch, leave 5-10% of the audience out, chosen at random. That group gets nothing on any channel.
- When it closes, compare the real sales of both groups, adding up store, ecommerce and any other sales channel, not just the conversions each platform reports.
Step 3 is where it usually breaks: each customer's sales are spread across systems that don't talk to each other. That is why in FLYDE the control group is set when you build the audience, it is the same for every channel and it is crossed automatically with real sales. The result comes with lift, incremental revenue and ROAS, margin and the confidence interval, next to the omnichannel attribution of the same campaign. You get both answers in one place: that is how our incrementality solution works.
Want to know whether your data would let you measure like this today? The 30 questions of the self-assessment will tell you.
And if you'd rather see it on your own channels, request a demo. You bring the data. FLYDE finds the way.