Your platforms are lying to you. And you split your budget on that lie.

A client asked us the most honest question anyone can ask about digital marketing, and didn’t even realize it:

“If I double my budget, which channel do I grow? And if they cut it in half, which one do I drop?”

It sounds like a budget question. It’s really a measurement question. Because to answer it you’ll go look at what each platform says it contributed — and right there, without anyone warning you, you’ve already lost.

Try something before reading on: add up what Meta, Google, TikTok and your email tool report this month. It’ll come out to quite a bit more than your Shopify shows. Sometimes almost double. That’s not a glitch in your dashboards: it’s the system doing exactly what it was designed to do.

Why the sum comes out double

Every platform is both judge and player.

Think of one real sale, just one. The person saw your ad on Instagram, the next day searched your brand on Google, and bought after opening an email from you. All three claim that sale: Meta by its view-through window, Google by the last click, email by the open. None of them lies on its own spreadsheet — each applies its own rule, and that rule is written to justify its own budget.

That’s why the arithmetic doesn’t add up. You didn’t sell double: you counted the same sale three or four times. The over-counting isn’t a bug. It’s the design.

And back to the client’s question: if every channel pins on medals for sales it didn’t generate alone, how are you supposed to know which one to grow or cut? You’re deciding budget with a rule of three over numbers you already know are inflated.

“But I use Google Analytics”

The natural move is to look for a neutral referee. It almost always ends up being GA4. Two problems.

The first you already sense: asking a Google tool to measure how Meta performs is asking a player to referee its rival’s match. GA4 has its own rule too — last non-direct click, by default. It’s not neutral: it’s one more rule, with an owner.

The second almost no one says out loud: following a person from one platform to another barely works anymore. Third-party cookies are gone, mobile identifiers got lost, and every platform is a walled garden that won’t let you look outward. Multi-touch is limping. Much of what you see today is modeled conversions: estimates dressed up as certainty. They look precise. They’re not.

So no — this isn’t fixed by switching dashboards. It’s fixed by switching the question.

The question that actually helps

It’s not “who gets the credit.” It’s: what would have happened if that channel didn’t exist?

That’s incrementality, and it changes everything. The sale a channel claims is one thing; the sale it actually generated is a very different thing. The person who was already going to buy and clicked your branded Google ad on the way isn’t a sale Google produced: it’s your sale, where Google set up a tollbooth.

Look at it that way and the budget question answers itself:

• The channel you grow is the one that, if you switch it off, drags your sales down — not the one with the prettiest ROAS.
• The channel you cut is the one whose real return, once you finally measure it right, is close to zero.
• And where to put the next dollar isn’t decided by history: it’s decided by where that dollar still pays off.

No single tool measures well

Unpopular part: there is no single tool. There are three ways to measure, and the trick is knowing that each one lies differently, so you know when to trust which.

Platform data (and multi-touch in general) is fast and granular: good for moving campaigns day to day. But it over-counts. Use it to operate, never as the verdict on where the money goes.

MMM (econometrics) looks at every channel with the same yardstick, including what leaves no click trail, and shows you where you’re already over-investing. It’s impartial because it doesn’t depend on anyone’s tracking. The cost? It’s slow and needs history to calibrate.

Experiments — turning a channel off in some markets and leaving it on in others, or running a holdout — are the only way to measure cause and effect for real. They’re the referee. Everything else is a hypothesis; the experiment hands down the verdict.

The full picture isn’t picking one: it’s using MMM to split the budget, experiments to calibrate, and platform data for daily tactics. Read, that is, with healthy suspicion.

What to do on Monday (if you sell on Shopify)

Let’s bring it down to earth.

Start with an experiment, not a dashboard. A holdout or a geo-test on your most expensive channel answers “where am I burning money?” better than any attribution, and it doesn’t depend on any platform telling you the truth.

Look at POAS, not ROAS. Return on profit, not on revenue. Selling a lot while losing margin is an expensive way to feel good.

Think at the margin. The question isn’t “which channel has the best ROAS,” it’s “where does my next dollar pay off most.” A channel with spectacular numbers may be saturated: every extra dollar there barely moves the needle anymore.

Aggregators are fine, but they’re not the referee. The ones that integrate with Shopify tidy up your operation and are worth having. But the referee is the experiment; the rest are well-dressed hypotheses.

The heart of it

There is no perfect impartial platform, because the problem was never the platform. Each one is optimized to sell you more of itself. The experiment is optimized to tell you the truth.

When you build your measurement around the experiment, the client’s question stops being a trap. “If I double, which do I grow? If they cut, which do I drop?” gets an answer — one that doesn’t depend on what each platform wants you to believe.

At Colabor we help brands set up exactly that: measurement that tells you honestly where each dollar should go. If you’re staring at your dashboards and the numbers don’t add up, book a call with us and let’s look at it together.