Marketing attribution – what it is and why it is harder in an AI-driven world

Definition

Marketing attribution means working out which channels and touchpoints actually contribute to a conversion, a sale or another desired outcome – across the whole user journey.

What marketing attribution tries to answer

Marketing attribution asks one question: what made the sale happen? A customer might have found you through an AI chatbot, read an article, seen an ad and finally searched for your name before buying. Which touchpoint should get the credit?

The answer determines how you allocate your budget. If you give all the credit to the last channel, you risk starving whatever actually built the interest earlier in the journey.

Attribution models – from last-touch to data-driven

There are several ways to distribute the value:

• Last-touch – all the value goes to the last channel before conversion. Simple, but misleading.
• First-touch – all the value goes to the first contact. Reflects discovery, but not what closed the deal.
• Multi-touch – the value is distributed across several touchpoints, for example linearly or weighted by time.
• Data-driven – a model that works out each channel's contribution based on actual data.

Google Analytics 4 removed the rule-based models and now uses data-driven attribution as the default. But it needs a certain amount of data to work reliably – and many smaller websites never get there.

Why attribution has become harder

Two major shifts have changed the playing field. The first is restrictions on third-party cookies – browsers such as Safari and Firefox block them by default, so tracking across websites and devices works far less well and large parts of the customer journey are left in the dark. The second is that a growing share of marketing's impact happens in channels that generate no measurable clicks.

Attribution in an AI-driven world

This is where it gets really tricky. An AI chatbot can summarise, compare and recommend your company without the user ever clicking through to your website. The last-touch model then only sees the final, measurable channel – and misses the fact that the trust was built in an AI answer.

As a result, a growing share of what drives business simply cannot be tracked with traditional tools. We look at how AI is changing search and visibility in How does AI affect SEO?.

How to work with attribution in practice

The answer is not a perfect model – there is no such thing. It is to accept the complexity and combine several signals:

• Use data-driven attribution where you have enough data.
• Supplement it with qualitative signals – asking "how did you hear about us?" often gives a more honest answer than the tracking.
• Measure trends over time instead of chasing exact attribution for every deal.
• Accept that some of the impact will never show up in the data – and make decisions with that in mind.