How do I report marketing contribution when a B2B sale has many touches and decision-makers?
Short answer
Report four columns instead of one number: sourced, influenced, assisted, and AI-influenced revenue, each with a stated weight and a visible assumption. Set the weights once, write down the reasoning, and never adjust them to hit a target. Weights that move on demand are why most marketing contribution reports get dismissed in the first joint review with sales.
Why a single number always loses the room
A long B2B sale involves three to eight people on the buying committee and 15 to 40 marketing touches. Any single contribution number is going to fail the moment someone asks how you got there.
Report only sourced revenue and marketing looks small. Report only influenced and marketing looks like it's claiming deals sales closed. Four columns with different weights makes the assumptions visible, which moves the conversation from "who deserves credit" to "which content type produces which outcome."
The four columns
| Column | What it counts | Weight |
|---|---|---|
| Sourced | First-touch page that captured the lead | 100% of deal value |
| Influenced | Pages consumed by contacts before closed-won | 30 to 50% |
| Assisted | Pages viewed by non-buyers on the same account | 10 to 20% |
| AI-influenced | Deals naming an AI surface as first touch in the survey | 100%, deducted from Direct or Branded |
The rule that protects the report
Set the weights once, document why, and freeze them.
The temptation every quarter is to nudge a weight up when the number looks weak. Do that once and finance learns the report is an argument rather than a measurement, and every future number gets discounted. A report with unflattering numbers and fixed weights is worth more than a flattering one with movable weights.
Pick your influenced weight from your average touches per deal: under 8 touches means 45 to 50%, 8 to 20 means 35 to 45%, over 20 means 30 to 35%. Fewer touches means each page carried more of the persuasion.
Why the AI-influenced column can't be optional
B2B buyers now educate themselves inside AI surfaces before visiting your site, then arrive as Direct or Branded traffic, which credits the wrong channel and understates marketing.
Our own state of AI search research, covering over 22,000 answers and 115,000 citation events, found Google AI Mode returns a citation-rich answer in 97.4% of matched B2B queries and Perplexity mentions the tracked brand in 41.7%. The research also found that once your brand appears in an AI answer for a prompt, next-observation mention probability is 83.2% on ChatGPT versus 9.9% after a miss, so AI-search presence compounds into branded search over weeks.
Without this column, that compounding shows up as unexplained brand strength and gets credited to nobody. Pew Research found that when an AI summary appears, users click a source link in only about 1% of visits, which is why the touch leaves no trace in any system you control. See How can I estimate how much AI-influenced traffic is hidden in my Direct channel?.
The three joins the report requires
Contact-to-page. Every page viewed by every contact over 12 months.
Contact-to-account. The buying committee grouping.
Account-to-opportunity. Deals joined back to accounts so revenue attributes to prior touches.
Most stacks have this data sitting idle rather than missing. See How do I connect Search Console data to CRM opportunities and revenue? for the GSC half.
The format for a joint review with sales
One page, three rows.
Top row. The four numbers with the weight stated in a footnote under each.
Middle row. Each number split by intent bucket.
Bottom row. Delta from last quarter, with one line of commentary naming what changed.
That commentary line is what turns the meeting from a negotiation into a strategy conversation, because it explains the movement before sales has to infer it.
Automate the monthly version
Assembling this manually takes 3 to 5 days a quarter, and the version sales sees usually differs from the version finance sees because they were built from different exports on different dates.
Running it as one agent removes that entirely. A Branch node computes each of the four columns on its own path with its own weight, so the logic per column is explicit rather than buried in a single prompt, and the frozen weights live in a Code node where they can be reviewed rather than renegotiated.
The monthly agent:
Start (schedule, 1st of month) → HubSpot Get CRM Objects (closed-won, 4 quarters, with first-touch, last-touch, and survey fields) → HubSpot Search Contacts (page-view history per associated contact) → Branch (four paths: sourced, influenced, assisted, AI-influenced) → Code (apply each column's frozen weight and divide influenced credit across touching pages) → GA4 AI Traffic Overview → share-of-voice recipe → funnel-coverage recipe → Prompt LLM (produce the four columns with the weight stated in a footnote under each, plus one line of commentary on what changed) → DOCX export → Send Email to CMO, VP Sales, and CFO.
Sending the identical document to all three from one source is the point. Most contribution disputes are not disagreements about method, they're two teams comparing numbers that were never computed the same way, and a single DOCX export attached to the run ends that argument permanently.
FAQ
Related answers
- How do I measure organic-influenced pipeline without claiming false precision?
- How should I value content that influences a deal but doesn't create the lead?
- How do I separate first-touch, assisted, and last-touch organic revenue?
- Which attribution model should I use for a long B2B buying journey?
Want a marketing contribution report that survives sales scrutiny? Start a free Analyze AI trial and wire GA4, GSC, and your CRM into one reporting agent.