Digital advertising built its reputation on measurability. Every click could be tracked, every conversion traced back to an ad, and every dollar attributed to a result. For a long time, that promise mostly held.
It holds less well today. Safari and Firefox block third-party cookies by default. Apple's App Tracking Transparency requires apps to ask permission before tracking people across other companies' apps, and many people decline. Privacy regulations in many countries limit how personal data can be collected and used. Each change removes some of the signals that traditional tracking depended on.
The effect is that the numbers in advertising dashboards increasingly rely on estimates and modelling. That is not a reason to stop measuring. It is a reason to measure differently.
Why one number is no longer enough
Most advertisers still rely heavily on attribution, the reports that assign credit for each conversion to the ads a person saw or clicked. Attribution remains useful for day-to-day decisions, but it has limits that have grown more visible.
| Gaps in tracking. | When some conversions cannot be linked to an ad, the channels that are easiest to track get more credit than they deserve. |
|---|---|
| Each platform marks its own homework. | Platforms report the results of their own ads, and their figures frequently add up to more conversions than actually happened. |
| Correlation, not cause. | Attribution shows which ads were present before a purchase, not whether the ads caused it. Someone who was going to buy anyway can still be counted as a success. |
A combination of methods
The most reliable approach is to use several measurement methods together, each covering the others' weaknesses.
First-party data and server-side tracking.
Collecting data directly from your own customers, with their consent, and sending conversion information from your own systems to the platforms (often called server-side or conversion API tracking) restores some of the signal lost from browsers and apps.
Incrementality testing.
Controlled experiments compare a group that sees your ads with a similar group that does not. The difference shows what the advertising actually added. Many platforms offer tools for this, and simple geographic tests, running ads in some regions and not others, can work well too.
Marketing mix modelling.
Marketing mix modelling uses statistical analysis of sales and spend over time to estimate the contribution of each channel, without tracking individuals. Once mostly used by large advertisers, it has become more accessible through open-source tools such as Google's Meridian and Meta's Robyn.
Attribution for daily decisions.
Attribution still has a role for quick, tactical choices within a platform, provided its results are checked against experiments and models.
Making it work in practice
A measurement approach only helps if it shapes decisions.
- Agree what success meansfor the business, such as revenue, profitable customers or qualified leads, rather than platform metrics alone.
- Invest in your own datawith clear consent, so that you rely less on signals controlled by others.
- Run experiments regularlyespecially before large budget shifts, so you know which channels truly drive results.
- Compare methods, and look for agreement.When attribution, experiments and models point the same way, you can act with confidence. When they disagree, that is worth investigating.
- Accept useful uncertainty.A range you can trust is more valuable than a precise figure that is wrong.
Less precise, more reliable
The era of perfectly tracked advertising is ending, if it ever truly existed. The advertisers who adapt will stop looking for one perfect number and build a measurement approach that combines their own data, controlled experiments and modelling. The result is less precise on any single day, but far more reliable for the decisions that matter.
