Marketing Attribution in an AI World
On this page
Attribution is the question every Irish marketing budget asks: which of this actually worked? The honest answer in 2026 is more complicated than the dashboards suggest, because the customer journey now runs partly through AI assistants you cannot see into. The fix is not better tracking magic, it is first-party data, honest models, and asking customers directly. This guide is the practical version.
The 2026 reality
- Cookies are fading, so third-party tracking undercounts more every year.
- AI assistants sit in the research path: customers ask ChatGPT or Perplexity before they ever visit you. That layer is invisible to classic attribution, which is why AI share of voice has become a reporting staple.
- First-party data is the asset: the events on your own site and in your own CRM are the only ground truth left.
Build the measurement foundation
- Events that equal money. In GA4, track leads, bookings, and purchases with values, not page views. Every other event serves those.
- One source of truth. GA4 for behaviour, the CRM for revenue, and a simple monthly join between them.
- Data-driven attribution. GA4’s data-driven model distributes credit across touchpoints using your actual data, with last-click kept as a sanity check.
Capturing the AI layer
Ask at the point of conversion: how did you hear about us, with options including AI assistants like ChatGPT. It is the only reliable way to attribute AI-influenced conversions, and the answers also feed your AI share of voice programme.
The models, honestly
- Last-click: everything to the final touch. Simple, biased, useful as a floor.
- Data-driven: credit spread across the journey. Fairer, noisier, the right default.
- Self-reported: what customers tell you. Unfashionable and extremely useful.
No model is truth. Use them for direction: what to double, what to cut, and where the journey is changing.
Keep it proportionate
An SME needs the foundation, a monthly review, and decisions made on direction rather than decimal points. The same discipline applies to the ads and reporting layers, which is why we build all three together in AI marketing automation.
Book a call if your attribution currently lives in three different spreadsheets; we will consolidate it into one monthly truth.
Frequently asked questions
What is marketing attribution?
Assigning credit for a sale or lead to the channels that influenced it, so you know what to spend more on and what to cut.
Why is attribution harder in 2026?
Cookies are fading and AI assistants now sit in the research path. More of the journey happens where you cannot see it, so first-party data and self-reported attribution matter more than ever.
What should I track in GA4?
The events that equal money: leads, bookings, and purchases, with values. Everything else serves those. GA4's data-driven attribution then distributes credit across the touchpoints.
How do I capture AI-assisted conversions?
Ask at the point of conversion: how did you hear about us, and include AI assistants as an option. The honest answer to AI attribution is asking the customer.
Which attribution model should I use?
Data-driven attribution in GA4 for most businesses, with last-click as a sanity check. No model is truth; the value is in the trend and the decisions it improves.
Is attribution worth the effort for an SME?
Yes, but keep it proportionate: one source of truth, events that mean money, a monthly review, and decisions based on direction, not decimal points.