Every month, you write a check to Meta. Google gets one too. Maybe TikTok.
Some of that money is buying customers who were going to buy anyway.
You just don't know how much.
Take one of our client who built their brand to $50 million with Meta.
According to Zuck it was working. Consistent 4x return, month after month.
Then iOS 14 dropped. Revenue fell 40%.
That's nearly half the business gone.
Payroll still due. Inventory already ordered. Agencies still on retainer.
And the number on the dashboard? Still claiming 4x ROAS.
The ads weren't broken. The number was just never telling the truth.
This is a guide to figuring out which of your ads are actually causing sales — and which ones are just showing up after the fact and taking credit.
Why You Can't Just Trust the Dashboard
Here's the problem with your attribution tools.
Meta says it drove the sale.
Google says it drove the sale.
Your email platform says it drove the sale.
They're all looking at the same customer and all taking credit.
But you're not getting three sales are you? You're getting one.
It's how these platforms are designed. They're built to show you the best possible version of their own performance.
And for a long time, that was fine — because the numbers were at least directionally useful.
Then iOS 14 gave people the option to opt out of tracking. A huge percentage did.
The data got worse.
The platforms kept reporting the same numbers.
And that's when a lot of brands found out what was actually true.
So if the dashboard can't tell you, what can?
What Incrementality Actually Means
Here's the cleanest way to think about it.
Imagine your brand exists in two parallel universes.
Everything is identical — same products, same customers, same competitors, same season.
The only difference is that in one universe, you spent $50,000 on a Pinterest campaign. In the other, you didn't.
At the end of the month, universe A made $400,000. Universe B made $390,000.
That $10,000 difference — that's the incremental value of your Pinterest spend.
Not what Pinterest told you. Not what your attribution model estimated. The actual, real-world difference your money made.
That's it. That's incrementality. The revenue that exists because of your spend, not just alongside it.
You can't actually run that experiment. You only have one universe.
But there are ways to get close — and that's what the rest of this guide covers.
Why Most Brands Don't Do This
You've heard of incrementality testing. You may have even done it once.
That's the problem.
Most brands treat it like a box to check. Run a test, get a number, move on.
But channels change. Algorithms shift.
What was incremental six months ago might not be today. Incrementality isn't a one-time audit — it's a practice.
And here's the honest reason most brands don't do it consistently: the dashboard feels like enough.
You open Shopify. Numbers look okay.
You check Triple Whale. ROAS looks fine.
You tell yourself the machine is working. And it might be.
But you're looking at what the platforms want you to see, not what's actually happening.
The brutal version of this plays out in slow motion.
A brand builds its entire operation around one channel. Hires people for it. Brings in agencies. Builds workflows.
The channel reports strong returns, so the brand doubles down.
Then something breaks — an algorithm change, a policy update, an iOS — and suddenly the revenue that the platform was "driving" turns out to have been largely fictional.
The real number was always in the bank account.
We watched this happen with a client we work with closely.
They'd built an entire cost structure on Meta's back.
When Meta's reported performance stopped matching reality, there wasn't a clean way to unwind it.
You can't fire an agency in a week. You can't cancel inventory that's already on the water.
The cost of not testing isn't just a bad month. It's everything you built while you weren't paying attention.
Four Ways to Run Incrementality Tests
You can't run the parallel universe experiment. But you can get close.
Here are the four methods that work in practice — ordered from most rigorous to easiest to start with.
1. Geo-Split Testing
Pick a channel you want to test. Run it in some markets. Don't run it in others. Keep everything else the same.
At the end of the test period, compare revenue between the two groups. The difference is your incrementality signal.
This is the most controlled method available to most brands.
It's not perfect — markets are never truly identical — but it's close enough to make real decisions with.
Best used when you're evaluating a new channel before committing serious budget to it.
2. Holdout Groups
Instead of splitting by geography, you split by audience.
Take a percentage of the people your campaign would normally reach and deliberately exclude them from seeing it.
Then compare. Did the people who saw the ads buy more than the people who didn't? By how much?
This is how you find out whether your retargeting is actually converting people — or just following around customers who already decided to buy.
3. On/Off Testing
The bluntest instrument. Turn a channel off for a defined period and watch what happens to revenue.
We did this with a client testing AppLovin. It came in showing strong early numbers.
Rather than take the platform's word for it, we turned it off and watched. Did revenue hold? Did it dip? By how much?
It's not a clean test — too many variables move at once — but it's fast, it costs nothing, and it tells you something real.
If revenue barely moves when you turn a channel off, that's a signal worth paying attention to.
4. Platform Attribution Modeling
The easiest method and the least reliable.
Most platforms now offer some version of an incrementality or lift measurement tool built in.
You're trusting the platform to grade its own homework, which has obvious limitations.
Use it as a starting point, not a final answer.
If a channel's own incrementality tool is struggling to show lift, that tells you something. If it's showing strong lift, verify it with one of the methods above before scaling.
Turning Test Results Into a Number You Can Act On
Running the test is step one.
But a test result by itself — "revenue dipped 8% when we turned off AppLovin" — doesn't tell you how to adjust your budget.
You need a way to compare channels against each other on the same scale.
That's where the incrementality coefficient comes in.
The idea is simple.
Every channel has a reported ROAS — the number the platform shows you.
And every channel has a real-world multiplier that reflects how much of that reported ROAS is actually true.
Multiply them together and you get your iROAS: your incremental return on ad spend.
iROAS = Reported ROAS × Incrementality Coefficient
Here's what that looks like in practice across a typical channel mix:
No reported ROAS, but downstream lift is measurable
Look at that table for a moment.
Google Branded Search is reporting the highest ROAS in the mix — 8x.
It looks like your best-performing campaign.
But strip out the customers who were already searching for your brand name and would have found you through organic results anyway, and you're left with an iROAS of 2x.
Meanwhile Meta is reporting a modest 2x. Looks weak on paper.
But Meta is working higher up — it's putting your brand in front of people who didn't know they wanted you yet. Account for that, and the real multiplier is 6x.
This is why budget decisions made from reported ROAS alone can be so badly wrong.
You end up over-investing in channels that take credit well and under-investing in channels that actually build demand.
Your coefficients won't be identical to these — they depend on your category, your customer, and your testing results.
Start with your best guess, run the tests to calibrate, and adjust over time. The precision matters less than the direction.
iROAS Is Only Half the Picture
Now you know which channels are actually driving revenue.
But there's one more variable that determines where you should actually spend.
When do you get the money back?
A channel with a 6x iROAS sounds better than one with a 4x iROAS. And it is — eventually.
If you have the runway to wait, influencer wins. If you need the cycle to keep spinning next month, Google Shopping wins — even though it's the "worse" channel on paper.
But if the 6x takes four months to recoup and the 4x takes 30 days, they're not the same decision.
Not if you have payroll due next month. Not if your inventory is already tying up working capital.
Not if your payment terms mean cash comes in slower than it goes out.
Most brands are operating with tighter cash cycles than they realize.
On average: 30 days to collect cash from a sale, 45 days of inventory sitting in a warehouse before it moves, 15-day delays in receiving payments.
Stack those up and your actual available budget is often 20 to 30 percent lower than your P&L would suggest.
So the real question isn't just "which channel is most incremental?" It's "which channel is most incremental given my current cash position?"
Think about it as a simple matrix. On one axis: how incremental is the channel? On the other: how fast does it pay back?
- High incrementality, fast payback — spend here first, every time
- High incrementality, slow payback — spend here when you have the cash reserves to wait
- Low incrementality, fast payback — useful when cash is tight and you need the cycle to keep moving
- Low incrementality, slow payback — hardest to justify; needs a specific strategic reason
We've seen this play out with clients: a 4x return with a 30-day payback versus a 6x return with a 120-day payback.
In a cash-constrained environment, the 4x wins almost every time. Not because it's better. Because the business needs the cycle to keep spinning.
Knowing your cash position isn't a finance exercise. It's what makes incrementality data actionable.
How to Structure Your Budget Around What You Know
Once you have iROAS numbers and an honest read on your cash position, the question becomes: how do you actually divide the money?
One framework we've seen work — after a client spent five years getting this wrong and then right — is a simple split:
- 50% goes to proven channels — ones with at least six months of data, predictable returns, and completed incrementality tests
- 30% goes to scaling experiments — expanding what's already working into new campaign types or audiences
- 20% goes to future bets — entirely new channels with no performance history yet
The logic is straightforward.
Your proven channels fund the experiments.
Your experiments fund the bets.
Nothing gets bet on before it's earned its place in the portfolio.
And crucially: a channel doesn't graduate into the 50% bucket just because it reported good ROAS for a few months.
It gets there when it's been tested — when you actually know it's driving incremental revenue, not just showing up in the attribution window.
One More Thing: Don't Build on Sand
There's a final variable that most brands ignore until it's too late.
How fragile is your revenue?
A 6x iROAS is impressive.
But if that return depends entirely on one platform, one algorithm, one ad account that could get banned tomorrow — it's not a business. It's a bet.
Every channel carries a different fragility level.
Google Shopping tends to be stable — search volume-based, predictable, hard to disrupt overnight.
Returning customers and your email list are reliable and compound over time.
Meta is high-value but volatile — especially if your product category is sensitive to policy changes.
The goal isn't to avoid fragile channels. It's to make sure you never depend on them.
The rule we use: only invest in a high-risk, slow-payback channel — like influencer, or a new platform you're testing — when your more reliable channels are already covering your fixed expenses.
That way, if something breaks, the business doesn't break with it.
When Meta's iOS hit landed, the brands that survived it cleanly were the ones that had already built something underneath.
Not because they predicted iOS 14.
Because they'd been building a portfolio instead of a single point of failure.
Putting It All Together
By now you have three variables for every channel:
- iROAS — how much revenue it actually drives
- Payback speed — how fast you get the money back
- Fragility — how likely it is to disappear
The way to combine them is simple. Channels with high iROAS, fast payback, and low fragility get the most budget. Everything else gets discounted accordingly.
Here's what that looks like with real numbers:
Notice what happens. Google Branded Search — the campaign with the highest reported ROAS in your account — drops to the bottom of the priority list.
And influencer, despite having genuinely strong incrementality, stays in the medium tier until you have the cash cushion to wait for it.
This is the point.
Not to follow a formula blindly, but to have a framework that forces the right questions.
What is this channel actually doing? When do I get the money back? What happens to my business if it disappears tomorrow?
Where to Start
You don't need to build the whole system at once.
Pick one channel you're not sure about. Turn it off for two weeks and watch what happens.
That's your first incrementality test.
It costs nothing. It takes two weeks. And it will tell you something real that no dashboard will.
Then build your iROAS table — even a rough one with your best guesses. Run one holdout test to calibrate it. Adjust.
Do this once a quarter, not once ever.
The brands that win the next five years won't be the ones with the biggest budgets or the best creative.
They'll be the ones who actually know what their money is doing. Your competitors are still reading dashboards and believing them.
That gap is the opportunity.
If you're running an eCommerce brand and your financial setup isn't giving you numbers like these — iROAS by channel, real payback periods, contribution margin that actually accounts for ad spend — we can help.
UpCounting handles accounting and fractional CFO work specifically for DTC brands. The kind where your books are built to make decisions from, not just file taxes with. Get in touch if you want your numbers to tell you the truth.
.jpg)
.jpg)
