Your Dashboard Is Lying to You

Hey everyone,

Welcome back for another bite to chew on.

Here is something we have been thinking about a lot lately.

Every platform says it drove the sale. Meta says it. Google says it. TikTok says it. You add up all the claimed conversions and it is 3x your actual revenue.

But when you look at the P&L, revenue is flat. Maybe even down.

The dashboards are lying.

It is not a conspiracy. It is a structural problem. Privacy changes like iOS 14.5 and cookie deprecation broke the signal that attribution relied on. The tracking pixels that used to stitch the customer journey together are now blind spots. And every platform filled those blind spots with modeled data that conveniently credits itself.

The sharpest operators we talk to are not trying to fix attribution. They are replacing it entirely with something called causal measurement.

Let's get into it.

On the Menu:

If you have ever looked at your Meta dashboard, your Google dashboard, and your TikTok dashboard on the same day and wondered how all three are claiming credit for the same sale, you are not alone.

The measurement gap is real. And it is costing brands real money every single day they rely on broken attribution to make spending decisions.

We came across a resource from Lifesight that lays out the shift from legacy attribution to causal measurement better than anything else we have seen. It is called Your Guide to Modern Measurement: The Causal Revolution, and it breaks down exactly why the old models fail and what to replace them with.

The results brands are seeing after making this shift are hard to ignore:

Here is what is inside the guide:

If you are spending six or seven figures a month on paid media and still making decisions based on platform-reported ROAS, this is worth 15 minutes of your time.

Get Your Free Guide to Modern Measurement →

Every brand running paid media at scale has felt this. The dashboards say things are working. The bank account says otherwise.

Here is the scenario. A customer sees your Meta ad on Monday. Clicks a Google Shopping ad on Wednesday. Opens a TikTok retargeting ad on Friday. Buys on Saturday.

Meta claims the sale. Google claims the sale. TikTok claims the sale. Your actual revenue is one sale.

When you add up platform-reported conversions across channels, the total can be 2x to 5x your real revenue. This is not an edge case. It is the default state of multi-channel advertising in 2026.

The result is that brands systematically overestimate channel performance across the board, which leads to overspending on channels that look productive but may not be driving incremental revenue.

iOS 14.5 was not just an Apple privacy update. It was a structural break in how digital attribution works.

Before ATT (App Tracking Transparency), platforms could follow a user across apps and websites with reasonable accuracy. After ATT, roughly 75-85% of iOS users opted out of tracking. That is not a small signal loss. That is most of your signal.

Platforms responded by building probabilistic models that estimate conversions. The problem is that these models are optimized to make the platform look good, not to give you an accurate picture.

Cookie deprecation compounds the problem. Third-party cookies are going away. The cross-site tracking that multi-touch attribution relied on is disappearing.

The foundation attribution was built on no longer exists.

Bad measurement does not just give you wrong numbers. It gives you wrong decisions.

If Meta tells you a campaign is returning 4x ROAS but the true incremental ROAS is 1.5x, you will keep scaling that campaign. You will move budget toward it and away from channels that might actually be driving more incremental revenue.

Over time, this compounds. Brands end up with entire media mixes optimized for vanity metrics rather than actual business outcomes. They are spending more and growing less, and the dashboards keep telling them everything is fine.

The first step out of this trap is acknowledging that platform-reported metrics are not ground truth. They are estimates from interested parties.

The question is not whether attribution is broken. It is what you replace it with.

Multi-touch attribution (MTA) was designed for a world where you could track individual users across touchpoints. It assigns fractional credit to each interaction in a conversion path.

That model assumed near-complete visibility into the customer journey. With iOS 14.5, cookie deprecation, and cross-device complexity, that visibility is gone.

MTA also has a fundamental design flaw: it can only credit channels it can see. Offline channels, brand effects, word of mouth, and organic search all get zero credit in most MTA models, even if they drove the conversion.

This systematically biases budget toward trackable digital channels and away from everything else.

Causal media mix modeling takes a completely different approach. Instead of tracking individual users, it analyzes aggregate data to isolate the causal impact of each channel on business outcomes.

Think of it like a controlled experiment at the portfolio level. The model accounts for seasonality, promotions, market trends, competitive activity, and other confounding variables. What is left is the estimated causal contribution of each media channel.

The key difference: causal MMM does not care about cookies, pixels, or device IDs. It works with the data you already have — spend, impressions, revenue, and external signals.

Modern causal MMM platforms can update weekly or even daily, giving you something close to real-time guidance on where your spend is actually productive.

Causal MMM on its own is powerful. But the sharpest teams calibrate their models with incrementality testing.

Incrementality tests use controlled experiments — geo-based holdouts, audience splits, or channel blackouts — to directly measure the lift a specific channel or campaign generates.

The results feed back into the MMM model, making it more accurate over time. Think of incrementality testing as ground truth checkpoints that keep your model honest.

Without this calibration step, even a well-built MMM can drift. With it, you have a self-correcting system that gets smarter the more you use it.

Knowing that measurement is broken and knowing what to do about it are two different things. Here is the practical playbook.

Do not try to measure everything at once. Pick the channel where you are spending the most — usually Meta or Google — and run a structured incrementality test.

The simplest version: pick matched geographic regions, turn off spend in the holdout regions for 2-4 weeks, and measure the difference in conversions compared to regions where spend continued.

This gives you a direct read on how much revenue that channel is actually driving versus how much would have happened anyway.

The results are often surprising. Brands regularly discover that their highest-ROAS campaign on paper is driving far less incremental revenue than expected. And sometimes a channel they undervalued turns out to be doing heavy lifting they could not see in the attribution dashboard.

Once you have incrementality data, feed it back into your measurement model.

If your MMM says Meta is driving 35% of revenue but your incrementality test shows the true number is closer to 22%, adjust the model. That recalibration step is what separates brands doing real measurement from brands running models as theater.

Each test makes the model more accurate. Over time, you build a measurement system that reflects reality, not platform self-reporting.

The whole point of better measurement is better decisions.

Once you know the true incremental return on each channel, you can reallocate budget toward what is actually working. Not what the dashboards claim is working — what is actually driving revenue.

Brands that make this shift typically find 15-30% of their media spend was going to channels or campaigns with low incremental value. Reallocating that spend to genuinely productive channels is where the ROI improvement comes from.

This is not a one-time exercise. The best operators run incrementality tests quarterly and update their models continuously. Measurement becomes a living system, not a static report.

If you want a deeper dive on how to implement this, the Lifesight guide walks through the full framework step by step.

Get Your Free Guide to Modern Measurement →