First-Touch Says 3% While Sales Quotes the Research
Halcyon sells HR software to mid-market companies. Fourteen months ago its three-person content team built a topic cluster around employee retention: one pillar page, 22 cluster posts, and three gated assets. The cluster now gets 41,000 organic sessions a month, up from essentially zero. Deals take about 4.5 months and involve six or more touches. The default attribution model in HubSpot is first touch, which credits the blog with 3% of closed-won revenue; paid search and outbound take most of the credit. Sales reps say prospects quote the retention research on calls, and branded search for Halcyon has grown 30% in the same period. The CFO wants to cut the content budget by 40% next quarter because '3% does not justify three people.' The team can change HubSpot attribution reports without engineering help but cannot run a geo holdout test.
1. Why does first-touch attribution report 3% when sales sees the cluster in deals? Explain the mechanism and propose a measurement approach the team can actually run.
Read the worked answer
First touch credits the earliest known interaction. Blog readers are anonymous until they identify themselves - usually by requesting a demo, downloading a gated asset, or clicking a retargeting ad. Those later touchpoints are what get recorded, so the blog's influence is invisible by construction. With a 4.5-month cycle and six or more touches, the cluster does mid-funnel work: it supplies the vocabulary buyers use on calls and drives branded search. First-touch is structurally the worst-suited model for that job.
What the team can run without engineering:
- Compare models. In HubSpot's attribution reports, put first touch, even-weighted, and U-shaped (roughly 40% first, 40% conversion, 20% middle) side by side. Do not champion the biggest number - present the range, which is the honest output.
- Blog-influenced pipeline. Deals where any contact had a blog session or blog form submission in the 90 days before close, regardless of position. Label it as a 'touched' measure, not causal.
- Self-reported attribution. Add 'How did you first hear about us?' with a content/blog option to the demo form. Small sample, but it is the only direct evidence and it usually surprises finance.
- Leading indicators: branded search volume, returning organic visitors, direct traffic, and the share of new contacts whose first-ever session was a blog page.
Frame the finding explicitly: first touch is a floor on the blog's contribution, not a measurement of it.
2. The CFO wants a 40% cut. Build the argument and a defensible compromise. What evidence would make you agree with the cut?
Read the worked answer
Argue on pipeline, not percentages. Present: team and tool cost against blog-influenced pipeline and closed-won; conversion from blog-sourced contacts to MQL versus the site average; time-to-close for deals that touched the cluster versus those that did not, explicitly labeled as correlation; branded search up 30%; and the replacement cost of 41,000 monthly sessions - rebuilding the cluster from zero would take another 14 months and roughly the same budget.
Concede what is true. Most of the value sits in a handful of URLs, and a 3% first-touch figure proves nothing on its own. Defend specific pages, not 'content' as a category.
Compromise: cut net-new production by 40%, not maintenance. Keep the pillar page and the top 10–15 cluster posts refreshed twice a year, keep the three gated assets, and pause new cluster expansion for two quarters. That captures most of the saving while protecting what already ranks. Publish the measurement plan - model comparison plus blog-influenced pipeline - in the same document, with a review date attached, so the decision is revisited on evidence.
Evidence that would make me agree with the full cut: after the two-quarter pause, blog-influenced pipeline falls proportionally or more with no recovery; or blog-sourced contacts convert to SQL below the site average and sales stops citing the research; or branded search declines despite continued paid spend. If the cluster genuinely is not contributing, cut it - but 3% first touch is not that evidence.