Your Marketing Attribution Says 62% Came From Direct

It is Monday the 15th, 9:04 AM. Your VP Marketing opens the attribution dashboard before the exec meeting. $1.4M in sourced pipeline last month, $868K tagged Direct. $210K tagged Organic Search with no landing page recorded. $164K tagged Paid Social with a UTM string that stops at utm_source=facebook. $98K tagged Referral pointing at t.co, $60K under a channel labeled Other.

She opens the CFO's spend view in a second tab. $184K spent on paid media in June. $92K on content and SEO. $34K on events. $28K on partnerships. The dashboard says $868K of the pipeline came from Direct, which the CFO reads as free. He walks into the meeting at 9:30 and asks why paid returned $164K on $184K spent and content returned zero on $92K.

She pings the demand gen lead. The demand gen lead pings the marketing ops contractor. The marketing ops contractor answers Thursday. The Direct bucket includes every lead where the first touch cookie expired, every hand-raiser from a sales-led outbound sequence, every referral from a partner without a tracked link, and every prospect who typed the domain after seeing a LinkedIn ad two weeks ago. Nobody knows the split. The board deck ships on the 22nd.

Attribution is a function. Most Series B and C teams have not staffed it because the first attribution report at Series A was HubSpot's default source field on 40 deals a quarter. The pipeline grew to 340 opps a quarter across nine channels, four regions, and three product lines. The function lives in the gap between the VP Marketing who owns the number, the demand gen lead who owns paid, the content lead who owns organic, the RevOps analyst who owns the pipeline report, the sales ops lead who owns opportunity source, and the CFO who signs off on next quarter's spend. On the org chart it sits under Marketing Ops. In practice it sits inside a HubSpot property nobody has audited since Series A.

The 62% nobody can decompose

Pull the pipeline sourced report from the last quarter. Filter by original source. Count opps tagged Direct, Other, or Unknown. Count opps with a UTM string that stops at utm_source. Count opps where the first-touch timestamp predates the account creation date by more than 90 days. Most teams past Series B find 45 to 70 percent of pipeline in unhelpful buckets, 20 to 30 percent with partial UTMs, and 10 to 15 percent with a first-touch cookie older than the current quarter.

Walk the Direct bucket. The first opp is a $220K deal from a director at a Fortune 500 who typed the domain into a browser. He also downloaded a benchmark report three weeks earlier from a LinkedIn ad, opened two nurture emails, and joined a partner webinar on the 4th. HubSpot recorded him as Direct because the cookie expired between the webinar and the demo request. The second opp is a $140K deal from an SDR-sourced outbound cold email that the AE marked as Marketing sourced because the prospect had visited the pricing page the week before.

The team that should own this knows it is broken. The VP Marketing runs the paid mix off the same dashboard the CFO reads and knows the Direct bucket is a soup. The demand gen lead runs paid spend against a Google Ads conversion column that fires on any form fill. The content lead publishes twelve posts a month against a report that credits organic zero because the first-touch cookie lives 30 days. The RevOps analyst rebuilds the sourced report every Monday from a HubSpot export and a Slack thread with the sales ops lead. Nobody watches every opp against a stitched touch history from cookie, IP, email, LinkedIn ad log, webinar registration, and CRM record.

Hiring a marketing analyst is the slow answer

The textbook fix is a senior marketing analyst or a director of marketing operations. Loaded comp in the US runs $140K to $190K a year. Months one through two go to auditing the HubSpot source rules, mapping the UTM taxonomy, and interviewing every channel owner on how they tag campaigns. Months three through six are when the multi-touch model gets stood up, the sourced report gets rebuilt with weighted attribution, and the Direct bucket shrinks from 62% to 18%.

The fractional version is faster to start and stops at the same wall. Seven to eleven thousand a month buys ten to fourteen hours a week of senior marketing ops time. The first month rewrites the UTM standard and ships a cleaner source rule. The 340 opps a quarter that need a stitched touch history stay in the Direct bucket because a fractional operator cannot audit 340 buyer journeys a quarter and also run the campaign roadmap.

Both versions assume the work is a person auditing a report on a cadence. The work itself is stitching every touch across ad platforms, the website analytics feed, the CRM record, the email tool, the webinar tool, the content library, the partner portal, and the LinkedIn ad log against every opportunity the day it converts. On 340 opps a quarter that is 30 to 50 hours a week of senior marketing ops work. No single hire clears that pile while also running the paid mix and the next campaign brief.

What a fractional AI attribution function does

Hand the HubSpot or Salesforce record, the Google Ads and LinkedIn Ads spend feed, the GA4 event log, the Segment event stream, the email tool send log, the webinar registration list, the partner portal, and the last four quarters of closed-won and closed-lost records to a fractional AI agent. The agent does the work a senior marketing analyst, a marketing ops director, and a data engineer would do together. The cadence is daily on touch stitching, per-event on the campaign launch, weekly on the sourced report, and per-deal on the closed-won attribution brief.

Every opp stitched against a full touch history the day it converts. The $220K Fortune 500 opp gets the LinkedIn ad view on the 22nd, the benchmark download on the 24th, the two nurture email opens on the 3rd and the 7th, and the partner webinar on the 4th stitched to the demo request on the 12th. The Direct bucket loses the deal and paid social gains a $220K credit the June spend report can read.

Every campaign launch tagged against a standard UTM taxonomy before it goes live. The demand gen lead launches a paid campaign on Tuesday. The agent audits the UTM string before the campaign turns live, flags the missing utm_content and utm_term fields, and rewrites them against the taxonomy. The June utm_source=facebook stub does not reach production.

Every closed-won deal briefed with a weighted attribution snapshot. The $140K SDR-sourced deal that also touched the pricing page gets a weighted split at 65% outbound and 35% organic. The AE stops fighting the sales ops lead over the source field. The demand gen lead sees the pricing page assist and prioritizes the next landing page test.

Every paid channel scored against a live blended ROAS, not a platform-reported one. Google Ads reports $6.20 return per $1 on its own conversion column. The stitched view lands at $2.80 blended once the outbound overlap and the referral leakage strip out. The CFO reads a number that matches the pipeline report.

Every Direct-tagged opp rescored against the last 180 days of touches. The historical Direct bucket gets re-attributed weekly against the full touch log. The 62% shrinks toward 15 to 22% because the buyer journeys resolve. Content stops reporting a zero contribution when it drove the second touch on 34% of closed-won.

The unit economics of a broken source field

A Series B company at $22M ARR running $340K a month in blended marketing spend is burning three specific things. The VP Marketing, the demand gen lead, the content lead, the RevOps analyst, and the CFO spend a combined 30 to 55 hours a week on the sourced report loop against a fully loaded hour of $180 to $310. That is $22K to $65K a month of senior time on work a stitched view clears. Ten to fifteen hours a week come back inside the first sprint.

The spend allocation is the second line. A budget mix built off a 62% Direct bucket underinvests in the paid social channel that sourced $220K and overinvests in the events line that shows a $60K credit tied to a stale first-touch cookie. On $340K in monthly spend the misallocation runs 15 to 25 percent, or $50K to $85K a month of paid budget landing on the wrong channel. A cleaner mix returns 20 to 40 percent more sourced pipeline against the same spend inside one quarter.

The board line is the third. A CMO who cannot explain why paid returned $164K on $184K last quarter and content returned zero on $92K loses the case for next quarter's budget. The CFO cuts the content line and shifts spend to paid social. Six months later the organic pipeline drops 30% and the paid mix hits diminishing returns. The reset costs $180K to $340K in redirected spend plus a two-quarter drag on the number while content ramps back.

A 14-day sprint to stand up the agent runs in the low to mid five figures. Ongoing cost lands closer to one senior analyst than a director hire. The touch stitching runs in week one. The weighted sourced report runs in week two. The blended ROAS view runs off a live feed before the sprint closes.

What changes after the sprint

Picture the same Monday the 15th, 9:04 AM moment, thirty days after the sprint ships. The VP Marketing opens the attribution dashboard. $1.4M in sourced pipeline last month. 18% tagged Direct, all with a stitched touch history showing a genuine no-ad, no-email, no-partner entry point. Paid social credited $340K against $184K spent, a 1.8x blended. Content credited $290K against $92K spent, a 3.2x blended. Events credited $110K against $34K spent. Referral credited $95K against $28K spent. The Other bucket sits at $12K.

The Fortune 500 deal shows the LinkedIn ad view on April 22nd, the benchmark download on April 24th, the nurture opens on May 3rd and May 7th, the partner webinar on May 4th, and the demo request on May 12th stitched in one view. The SDR-sourced deal shows a 65% outbound, 35% organic split with the pricing page assist named. The CFO walks into the exec meeting on the 22nd and approves a 15% paid social lift and a hold on the events line.

If your attribution dashboard currently ships with more than half of pipeline in Direct, Other, or Unknown, the version where every opp lands with a stitched touch history and the CFO reads a blended ROAS that matches the pipeline report is fourteen days away. Attribution is a function. You can hire against it, you can retain a fractional marketing ops operator for it, or you can scope a sprint and have it running this month. The work is the same. The math is not.