The Attribution Lie: Why Your Platform Metrics are Fiction

The uncomfortable truth about campaign revenue

Your ESP tells you last month's campaign generated $2.3 million in revenue. Three other campaigns also went out that week, and they each "generated" similar numbers. Your paid search platform claims credit for $1.8 million. Add it all up and the total attributed revenue across your channels exceeds actual company revenue.

This isn't a reporting glitch. It's attribution working exactly as designed — and that's the problem.

How attribution actually works

Attribution models don't measure causation. They assign credit. When your ESP sends a campaign, it sets a conversion window — typically 3 to 14 days. Any purchase made by a recipient within that window gets credited to the campaign. Didn't click? Didn't open? Doesn't matter. They were "influenced."

Your paid search platform does the same thing. So does social. So does display. They're all counting the same purchases, from the same customers, as their own wins. The customer bought once. Each platform took credit. Your CFO gets a consolidated report that doesn't reconcile to reality.

Your best customers create the biggest attribution lies.

This is the part most marketers find uncomfortable. Your Hero and Super Hero segments — your most frequent, highest-value customers — show the highest attributed revenue in any campaign report. They also have the lowest incremental lift.

Why? Because they're buying anyway. Every campaign you send them finds a purchase in the window and claims it. But if you'd withheld the campaign, they'd have bought anyway. You spent campaign budget on a sale that was already happening.

SIVV's measurement work consistently shows this pattern. When we run holdout testing on VIP segments, incrementality rates are frequently in the 10–20% range. On win-back and onboarding campaigns — where the campaign is actually changing behaviour — incrementality regularly exceeds 60%.

The budget allocation problem

Attribution doesn't just give you inaccurate numbers. It actively misdirects your spend. If your highest attributed revenue campaigns are the ones sent to people who'd buy regardless, you keep scaling them. If a win-back campaign to lapsed customers shows lower attributed revenue (because the absolute revenue per customer is smaller), it looks like a weaker performer — even though every dollar it generates is genuinely incremental.

Over time, attribution optimises your marketing spend toward customers who need the least help and away from customers where campaigns actually make a difference. It's a slow, invisible deterioration in marketing effectiveness.

The scientific alternative

Holdout testing — withheld control groups assigned before a campaign runs — is the only way to separate correlation from causation in marketing measurement. It's the same logic pharmaceutical trials use. You can't know if a drug works without a control group. You can't know if a campaign works without one either.

Before each campaign, SIVV splits the target audience. The treatment group receives the campaign. The holdout group receives nothing for that campaign — but continues as normal in all other respects. After the campaign window closes, revenue is compared between both groups. The difference, adjusted for size, is the true incremental lift.

The result reconciles to your P&L. Incremental revenue plus baseline revenue equals actual company revenue. No double-counting. No fiction.

What to do with this

You don't have to rip out your existing measurement stack. Attribution still has a role — it tells you about channel coverage and touchpoint sequences. But it shouldn't be the number you use to justify budget or evaluate campaign performance.

Start with your highest-spend campaigns. Run a holdout on the next one. Compare the incremental number to what attribution reports. The gap will tell you everything you need to know about where your marketing budget is actually going.

The most common response when clients see their first holdout result is "I wish I'd known this sooner." The second most common is an immediate restructure of their campaign calendar away from loyalty segments and toward acquisition, onboarding, and win-back — where campaigns genuinely change behaviour.

That's not a measurement exercise. That's a revenue exercise.

‍

About SIVV

SIVV is the customer intelligence and decisioning platform built for marketers who want to know what's actually working.

We sit above your marketing platforms, combining:

• Sophisticated customer intelligence (churn prediction, lifecycle segmentation, propensity modelling)
• Scientific campaign measurement (randomised control groups for every campaign)
• True incremental revenue reporting (revenue that reconciles to your actual business performance)

Clients across telecommunications, retail, gaming, entertainment, and travel use SIVV to:

• Measure true incremental revenue for every campaign
• Identify which audiences and offers drive real lift
• Make budget allocation decisions based on incremental ROI
• Optimise marketing performance based on what actually moves the needle

Stop optimising against attribution. Start measuring incrementality.

Learn more at sivv.net or contact us to discuss how incremental measurement can transform your marketing performance.