Štěpán
Founder,
Digital Ants
Incrementality: How many of your conversions would have happened without ads?
Your campaigns love to take credit for things that would have happened without them. Incrementality is a type of measurement that shows the difference between a conversion your ad created and one it just claimed. It tells you whether you're paying unnecessarily for results you'd have got for free.
| Campaign | In report | Incre- mentallyIncrementally |
|---|---|---|
| Brand search | 420 | 31 |
| Remarketing | 310 | 88 |
| App campaign | 540 | 190 |
| Prospecting | 230 | 221 |
| Total | 1 500 | 530 |
Illustrative numbers. On the left, what you see in the report. On the right, what the campaign actually delivered. You pay for the difference too.
The Non-incremental Fairy Tale
Imagine two friends who predict the weather for you. The first only says "it'll rain" on days when it actually rains. The second says "it'll rain" every single day. Both have near-perfect accuracy on rainy days — but only one of them is actually useful to you. An ad channel works the same way: it claims credit for every nearby order, regardless of whether it influenced it at all.
Non-incremental conversion
The customer had already made up their mind. The ad just got in the way — and claimed the credit.
Incremental conversion
Without the ad, the purchase wouldn't have happened. This is the only conversion that marketing actually created.
Why does this matter to you?
When a large share of conversions are non-incremental, you're paying for results you would have gotten anyway. Money meant for new growth is instead subsidising demand that already existed.
Wasted budget
Part of your media spend buys no incremental growth.
False confidence
Campaigns look profitable, so you keep funding them — needlessly.
Hidden stagnation
You've hit a ceiling that doesn't show up in the report at all.
How this corrupts your numbers
Non-incremental conversions don't just lie about one metric — they corrupt the entire logic of the report at once. They credit a campaign with conversions it didn't cause, inflating volume while artificially deflating CPA. The campaign then looks like your cheapest source of customers. And that's exactly why you put more money into it.
CPA as reported
250 CZK
1,500 conversions for the budget
Actual CPA after clean-up
708 CZK
530 incremental conversions
Same budget. When you remove conversions that would have happened without the campaign, the real cost per acquired customer can be several times higher than the reported figure. Optimising toward the reported number then leads you in the wrong direction.
Does this apply to you?
For some, incrementality is a minor issue; for others, it's critical. The more you recognise yourself in the list below, the higher the chance that part of your performance is really just existing demand repackaged.
Strong, well-known brand
People search for you by name. Ads then often just get in the way of a conversion that would have happened anyway.
Repeat customers with frequent purchases
Recurring orders from loyal customers are easily attributed to remarketing — even when they would have bought on their own.
High share of branded search
Many people already know your name. A click on a brand ad just replaces a direct visit — and you pay extra for it.
Short purchase journey
Impulse or quick purchases happen before ads even have a chance to influence anything.
High share of remarketing and app campaigns
Channels targeting people with existing intent automatically harvest ready-made conversions.
Reporting without CRM integration
When you can't match a conversion to a specific customer, you can't tell new from returning.
Strong, well-known brand
People search for you by name. Ads then often just get in the way of a conversion that would have happened anyway.
Repeat customers with frequent purchases
Recurring orders from loyal customers are easily attributed to remarketing — even when they would have bought on their own.
High share of branded search
Many people already know your name. A click on a brand ad just replaces a direct visit — and you pay extra for it.
Short purchase journey
Impulse or quick purchases happen before ads even have a chance to influence anything.
High share of remarketing and app campaigns
Channels targeting people with existing intent automatically harvest ready-made conversions.
Reporting without CRM integration
When you can't match a conversion to a specific customer, you can't tell new from returning.
The Usual Suspects
None of these campaigns is "bad". But all of them tend to claim credit they don't entirely deserve. It's worth keeping a close eye on them.
Brand search
high riskTargets people who are already searching for your name. Most of them would have reached you via an organic result or direct entry for free. Ads here often just buy clicks you'd have got without them.
Remarketing
high riskTargets people who already showed interest — they were in the cart, on a product page, on your site. Many would have completed the purchase anyway. Remarketing loves to claim that conversion as the last touch.
Google App Campaigns
medium to high riskA black box that optimises on installs and in-app events. Left unchecked, it easily harvests users who would have found and installed the app themselves — and reports them as its own performance.
Broad "performance" formats
medium riskAutomated campaigns optimising on last click tend to take credit for conversions set in motion elsewhere. Attribution then tilts toward the channels that harvest, not the ones that create.
How to actually measure incrementality
Good news: you don't have to guess. There are methods to properly isolate the lift. They vary in complexity, but the principle is always the same — compare a world with ads to a world without them.
Tracking connected to CRM and shop system
The foundation of everything. When you can match a specific conversion to a specific customer, you can tell new from returning and first purchase from fifth. Without this data foundation, all other tests are blind.
Geo-holdout tests
You turn off ads in some regions and keep them running in others. The difference in revenue between the on and off areas is the pure incremental lift — like a medical trial with a control group, but with cities instead of patients.
Conversion lift studies
Platforms like Meta or Google can randomly exclude part of your audience from ads and compare how much more the exposed group converted. You measure the actual difference the ad made — not what attribution claims credit for.
Controlled pauses and gradual scaling
You deliberately pause a suspect campaign and watch what it does to overall revenue. If nothing happens, you've just found a budget that was contributing nothing.
Tracking connected to CRM and shop system
The foundation of everything. When you can match a specific conversion to a specific customer, you can tell new from returning and first purchase from fifth. Without this data foundation, all other tests are blind.
Geo-holdout tests
You turn off ads in some regions and keep them running in others. The difference in revenue between the on and off areas is the pure incremental lift — like a medical trial with a control group, but with cities instead of patients.
Conversion lift studies
Platforms like Meta or Google can randomly exclude part of your audience from ads and compare how much more the exposed group converted. You measure the actual difference the ad made — not what attribution claims credit for.
Controlled pauses and gradual scaling
You deliberately pause a suspect campaign and watch what it does to overall revenue. If nothing happens, you've just found a budget that was contributing nothing.
Štěpán · Founder,
Digital Ants
I'll show you this on your own campaigns
30 minutes, video call, no sales pitch. I'll show you where you're most likely paying for non-incremental conversions — and what to do about it.
You'll leave the call knowing:
- Where the biggest distortions are hiding
- Which campaigns to test first
- Whether measurement is worth tackling for you at all