Marketing attribution tries to answer which channels and pieces of content caused a customer to buy. It assigns credit for a conversion to one or more touchpoints using a model, such as first touch, last touch or a data-driven approach. For SaaS companies with long, multi-person buying cycles, attribution is never perfect, but without some version of it, organic search usually gets less credit than it earns.
Common models
GA4 now offers data-driven attribution and last-click variants; Google removed first-click, linear, time-decay and position-based models from GA4 in 2023. CRM and dedicated attribution tools still offer them.
- First touch: all credit to the first recorded interaction.
- Last touch or last non-direct click: all credit to the final interaction before converting.
- Linear: credit split equally across touchpoints.
- Position-based: most credit to the first and last touches, the rest spread between.
- Data-driven: credit assigned by an algorithm that compares converting and non-converting paths.
Why SEO is often under-credited
A buyer might first find you through a comparison page, return weeks later by typing your URL, and finally request a demo after a sales email. Last-touch models give organic search nothing. Visits lost to ad blockers, consent choices and cross-device journeys break the chain further, and most of the lost visits get reported as direct.
AI assistants add another blind spot. A buyer may learn about you in a chat answer, then search your brand name. The assistant gets no credit and the branded search gets it all.
A practical approach for SaaS
No single model is correct, so use several sources and look for agreement.
- Record first-touch and last-touch source on every lead and opportunity in the CRM.
- Add a free-text “How did you hear about us?” field to sign-up and demo forms.
- Report pipeline from opportunities whose first landing page was an organic page.
- Watch branded search demand over time as a signal of awareness built elsewhere.
- Where budget allows, run holdout or geographic tests for paid channels.
What to avoid
Avoid choosing the model that makes your own channel look best. Agree definitions with sales and finance first, then apply them consistently, so a change in the numbers reflects a change in the business rather than a change in method.
Example
Two models, two stories
Hypothetical company Northdesk looks at last quarter’s new customers. On last touch, most are credited to direct traffic and sales outreach, with organic search close to zero.
On first touch, a large share started on comparison and integration pages. The self-reported field on the demo form often names “Google” or a specific comparison. The team now reports both views side by side to leadership rather than picking one.
Common mistakes
Relying on last click alone
It systematically credits the final step and ignores the research that came before, which is where most SEO sits.
Trusting one tool’s numbers
Ad platforms, analytics and CRM each count differently and each tends to credit itself. Reconcile them against closed revenue.
Changing the model to fit the story
Switching models each quarter makes trends meaningless. Pick definitions and keep them.
Questions
Use more than one. First touch shows what brings people in, last touch shows what closes, and self-reported attribution fills gaps that tracking misses.
Only partly. Some assistant visits arrive as identifiable referrals, but much of their influence is invisible to tracking. Self-reported fields and branded search trends are the best current proxies.
