SEO Attribution: Proving ROI Without Last-Click
Why last-click buries SEO ROI and which multi-touch models actually defend the budget in your next CFO meeting, with real numbers from BigQuery and GA4.
The CFO opens the GA4 report, sees 12% of revenue credited to organic, and asks the killer question: why are we spending eighty thousand a month on SEO? The honest answer is not in the default dashboard. Last-click attribution credits organic only when it slams the door shut, ignoring the median 47-day B2B journey that started with a well-ranked blog post. If you keep defending SEO with last-click numbers, you will lose the budget fight, and you will deserve to lose it. The channel is not broken. The yardstick is.
To size the damage, I ran a BigQuery analysis joining GSC with GA4 events for a SaaS client: 38% of assisted conversions started on informational queries that never showed up as source/medium on the final conversion. Those users entered through a post about Search intent: 4 types and how to map them on the SERP, came back three weeks later typing the brand, and GA4 stamped everything as direct. Multiply that across six months and you have a fully wrong narrative reaching the boardroom. Anyone not measuring assisted conversions with a 90-day window is flying blind.
The honest path starts with three models running in parallel: GA4 data-driven attribution (DDA) as baseline, a Markov Chain model in Python or R to validate DDA, and a simplified Shapley model for low-sample channels. Across four projects where I deployed this stack, SEO climbed from 12-18% (last-click) to 31-42% (multi-touch) of credited revenue. That is not magic, that is correct accounting. Before you get there, audit your conversion events: too many teams want to redesign attribution on top of broken tagging, and no model will save that.
Top-of-funnel content is where last-click does the most damage. A piece like Featured snippets: how to structure content for position zero can pull 8,000 sessions a month with zero direct conversions, then feed 23% of pipeline sixty days later. Cross it with Cohort analysis applied to organic content and you discover which editorial clusters generate paying customers at 90, 180, and 365 days. I keep a per-client sheet that ties landing URL to closed-won deals through a persisted user_id. Without that stitching, any ROI conversation collapses into opinion. Dreamdata, HockeyStack, or a well-built Looker Studio on top of BigQuery all get you 80% of the way.
To defend investment, translate attribution into projected revenue. Use the forecasting frame from SEO Forecasting: how to project results with confidence with the CTR benchmarks in CTR benchmark by position: updated 2026 data to estimate incremental revenue per position gained. Real example: e-commerce client, query with 22,000 monthly searches, position 7 to 3, CTR jumping from 2.1% to 8.4%, average order R$ 340, assisted conversion rate 1.8%. That is roughly R$ 95k per month in incremental revenue attributed on a 60-day window. That number lands on the slide with source, formula, and confidence interval. CFOs do not argue with auditable math, they argue with vibes.
The closer is cultural. Multi-touch attribution only survives if growth teams stop comparing channels on different rulers. Paid Search should not get last-click credit while SEO gets a one-day view-through window. Standardize the window, standardize the model, publish the methodology internally. Pair it with Honest SEO KPIs: beyond rankings and traffic and SEO dashboards: what to show the CFO vs the marketing team so each stakeholder sees the slice that matters to them without distorting the whole. Practical takeaway: this week, run a side-by-side last-click vs DDA comparison in GA4 for your top 20 organic URLs. Take the delta to your next meeting. If it is under 15%, your attribution is already decent. If it is bigger, you just found the budget you were about to lose.