Attribution was previously considered as the Holy Grail of digital marketing. I know I considered it as that. When I started my long career in advertising technology, the notion of “everything is trackable” was the greatest promise I fell for. 

Marketers poured billions into advertising, under the assumption that they could precisely tell which ad drove results. investors poured billions into marketing tech platforms offering various methods of tracking the user. Platforms focused on last-click, first-click, multi-touch attribution (MTA), or more recent AI driven models to track users. The goal was always the same: trace the path of the customer and assign value accordingly. 

But the system is crumbling. Privacy laws, platform restrictions, and the fundamental flaws in attribution itself exposed marketers into a state of uncertainty. Today, the question isn’t just whether attribution is accurate, but whether it should continue to exist at all. 

Marketing leaders who once obsessed over assigning credit are now facing a really harsh wake up call: The traditional way of measuring advertising performance is no longer viable. Instead, we are in a new era – one where incrementality and algorithmic attribution will shape the future of marketing measurement.

Filling the gaps with nonsense

It’s crazy to think that up to a few years ago, marketers had seemingly endless data at their disposal. You could map out the entire population of the internet with most programmatic DSPs from the comfort of your desk. Marketers were able to track users across devices, analyze every touchpoint, and build highly detailed custom journeys. Privacy regulations starting with GDPR in Europe, CCPA in California, and Apple’s App Tracking Transparency (ATT) changed the game completely. 

These policies placed strict limitations on tracking users without explicit consent, making most common attribution models significantly less reliable. Advertisers who once relied on user-level data found themselves flying blind, especially since users were increasingly opting out of tracking. The deprecation of third party cookies in chrome (coming soon to a browser near you!) will only tighten the grip on data availability, making it nearly impossible to track cross platform activity.

As a result, attribution has become less of a science, and more of an educated guess – throwing a dart on a board with advertising platforms to decide “who should get the credit for this conversion?”. Marketers who previously relied on multi-touch attribution (MTA) models now work using fragmented, incomplete datasets. But rather than call it quits – MTA tech vendors are resorting to filling the gaps using pseudo-science logic, trying to reconstruct the user journey in some way that would make sense to advertisers. The illusion of precision is breaking, and brands are beginning to see that the promise of  “perfect attribution” has always been a pipe dream. 

The Walled Gardens and Overlapping Claims

Let’s assume that privacy wasn’t an issue. Even before the privacy restrictions, attribution faced another massive challenge: the walled garden problem. 
Today’s dominant ad platforms – Google, Meta (Facebook), TikTok – all operate in silos, restricting the flow of cross-platform data. Each of these platforms claims credit for the same conversion, leaving advertisers over-attributing and double counting results. 
For example, if a user saw an ad for a pair of shoes, then later searched for those on Google and clicked a paid search result before purchasing. Who gets the credit?

• Facebook will say its ad played a critical role in the purchase.
• Google will say that without its search ad, the user wouldn’t have converted.
• Meanwhile, the e-commerce brand running the campaign has no way to know which claim is more valid.

This duplication creates a distorted view of marketing effectiveness. The ad platforms are not reporting all of their 1st  data externally. The result? Marketers have to make budget decisions based on incomplete data.

Turning back time

In an effort to fix the issues listed above, once privacy restrictions, policies, and laws got put into place, we saw futile efforts to continue tracking users by advertisers/platforms sharing data in a “clean room”. But the most baffling direction we saw to fixing the issues caused by technology from the 90s (attribution) was to use technology from the 50s (Media Mix Modeling).

MMM made its strange resurgence into today’s world with some additional features and Bayesian inference (fancy word to describe a way of updating prior knowledge). 

But MMM came from a world where marketers operated in a slow moving media space. No one was running 70 ad variations in the New York times, and the New York Times was not algorithmically modeling the number of sales they generated for you, for it to create even more ad variations, or target specific ads to its readers in segment A vs. readers in segment B. 

The world in which MMM lived in was not doing real time bidding, changing the assigned value to media and therefore changing the potential efficiency curve down to the second of the day. 

MMM while great for very specific use-cases, specifically as a scenario playground, taken with a massive grain of salt, was never to be an alternative to attribution. (which is why MMM fails to meet expectations).

What does a world without attribution look like?

Making a much needed shift to Incrementality

As traditional attribution fails, marketing leaders are shifting their focus towards incrementality. Unlike attribution, which tries to assign credit to specific touchpoints, incrementality measures the true impact of campaigns over results. 

Incrementality offers marketers the answer to the basic question: what happened when we started, increased, decreased, or stopped spending a specific campaign?

For example, if an advertiser pauses their Facebook ads and sees no change in conversions, that’s a strong sign that Facebook wasn’t generating incremental results. On the other hand, if conversions would to drop significantly, that would suggest that Facebook ads were contributing substantial incrementality. 

The shift towards incrementality represents a fundamental paradigm shift. Instead of trying to trace every step of a user’s journey, marketers are now measuring real business outcomes. 
 

Algorithmic attribution: The future of optimization

While incrementality measurement is gaining traction, ad platforms themselves are also evolving. In a world where traditional attribution fails, platforms like Google, Meta, and TikTok are embracing a new approach: algorithmic attribution.

Instead of relying on user tracking across multiple sites, algorithmic attribution leverages first-party data, AI, and deep learning models to estimate the impact of advertising more effectively.

Ad platforms already have massive amounts of first-party data—they know who engages with ads, who converts, and what content is most effective. By combining this data with AI, platforms can optimize ad delivery in real-time, automatically improving targeting and adjusting spend based on expected value.

For example:

• Google’s Performance Max campaigns already use AI to determine where and how ads are shown without marketers needing to specify channels.
• Meta’s Advantage+ Shopping campaigns automate creative selection, audience targeting, and bidding based on predicted performance.
• Amazon’s ad ecosystem is built on closed-loop attribution, where purchase data feeds directly into campaign optimization.

This algorithmic attribution model shifts the focus from "who gets credit for a sale?" to "how can the platform maximize results?" The future isn’t about tracking individual users but leveraging AI to dynamically improve marketing effectiveness within platform ecosystems.

Can you prepare for the Attribution-Free era

Marketing measurement is changing, and brands that fail to adapt will waste millions chasing inaccurate data. To prepare for this shift:

• Start thinking in terms of business outcomes (i.e. VALUE), rather than channel-based ROAS. 
• Add incrementality to your tech stack and start measuring incrementality continuously. Running a few inconclusive tests is not enough. 
• Embrace AI-driven automation, but remain skeptical of platform self-reporting. Use multiple sources of truth to verify impact.
• Invest in creative testing and audience insights. With attribution becoming less reliable, creative quality and audience targeting will matter more than ever.

The end of attribution; beginning of innovation

Attribution, as marketers once knew it, is dying. The future belongs to AI powered incrementality measurement and AI-driven algorithmic attribution. Instead of obsessing over which touchpoint deserves credit, the smartest brands will focus on maximizing actual business outcomes.

The question isn’t whether attribution will disappear—it’s whether your marketing team is ready to thrive without it.