Attribution Dispute Between First-Touch and Primary Campaign Source
Cobalt Ridge sells enterprise software with a seven-month sales cycle. Marketing runs 14 programs a month: LinkedIn lead-gen, Google search, webinars, field dinners, content syndication, and always-on nurture. Salesforce sets 'Primary Campaign Source' when a lead is created; Marketo's Revenue Explorer dashboard uses first-touch attribution. The CMO's dashboard shows webinars driving 42% of pipeline, while the CFO's Salesforce report shows paid search at 51%. The gap widened after the team added a 'Campaign Influence' custom object populated only when a rep manually adds a campaign. Last quarter 310 opportunities closed-won, but only 190 had any campaign association. Three field dinners cost $240,000 and generated 28 opportunities, 9 closed-won; the current dashboard credits them with $0 because attendees were imported without program membership. The board wants a defensible answer on marketing's contribution.
1. Why do the CMO and CFO see different program winners? Explain the effect of first-touch, Primary Campaign Source, manual Campaign Influence, and missing program membership.
Read the worked answer
They are answering different questions with different denominators. The CMO's first-touch view credits the first known program that touched a person. That overcredits early demand creation and can make webinars look strong if many buyers first appear at a webinar. The CFO's Salesforce report uses Primary Campaign Source, which is set when the lead is created and often reflects the demand-capture channel, such as paid search. It is static: later webinars and field dinners cannot change it. Manual Campaign Influence is a third model, and because reps populate it only sometimes, it undercounts programs that sales does not log. The 190 of 310 closed-won opportunities with any campaign association means 120 wins have no campaign chain at all; any model is only describing 61% coverage. The field dinners show $0 because imported attendees were never added to a program, so there is no membership to attribute. The gap is therefore partly model choice and partly data coverage. Neither dashboard is a defensible measure of incremental contribution. Before choosing a winner, fix the missing memberships and opportunity contact roles, then compare models on the same complete dataset.
2. Recommend a defensible attribution approach for a seven-month enterprise sales cycle. Specify the model, data requirements, and how to handle offline field dinners.
Read the worked answer
Model. Use a position-based or W-shaped model for this cycle: credit the first touch, the opportunity-creation touch, and the middle touches. A simple starting split is 40% first touch, 40% opportunity creation, and 20% spread across middle touches, with the exact weights agreed by sales and finance. Do not use a single-touch model as the board truth. If Marketo Measure or another attribution product is licensed, use it for multi-touch; otherwise build campaign influence in CRM with a documented model and automate it. Keep first-touch and last-touch as diagnostic views, not the headline.
Data requirements. Every program must create campaign membership, including offline events and content syndication. Require opportunity contact roles so the model knows who is on the buying committee. Capture UTM parameters and a hidden source field on forms; sync offline attendee lists into a Marketo program before or at the event. Record program cost so cost-per-opportunity can be calculated. Deduplicate contacts and reconcile Marketo programs to CRM campaigns.
Field dinners. Create a Marketo program and a CRM campaign for each dinner, import attendees with consent, add them to the campaign, and associate them to open opportunities. Record the $240,000 cost across the three dinners. Because attendees are selected, do not claim the dinner caused the deal; use a matched comparison or a post-event lift question. Report influenced pipeline and closed-won with coverage, plus cost per opportunity and cost per win. If coverage is only 61%, present a range and a data-quality plan, not a false precision.
3. The board asks what the three field dinners returned. Give the calculations, the caveats, and how you would present overall marketing contribution.
Read the worked answer
Field dinner math. 28 opportunities from $240,000 gives $8,571 cost per opportunity. 9 closed-won gives $26,667 cost per win. The win rate is 9/28 = 32.1%. Compare that with the baseline win rate for similar opportunities and with average deal size; without those, the return is incomplete. The caveat is selection bias: attendees were chosen, so they may have been more likely to buy anyway. There is also no control group and the current attribution shows $0 only because membership is missing, not because the dinners had no effect. Present the result as 'influenced' with a range, not as causal proof.
Overall contribution. First fix coverage: 190/310 = 61.3% of closed-won opportunities have any campaign association. Report that coverage explicitly. Then present a range: directly attributable closed-won under the chosen model, plus modeled influence for the unassociated 38.7%, clearly labeled as an estimate. Show pipeline, closed-won, cost per opportunity, and cost per win by program, with the model weights and assumptions in an appendix. Validate with a holdout or geo test where possible. The board message should be: 'Here is what we can defend, here is the coverage gap, and here is the investment needed to close it.' That is more credible than declaring webinars or paid search the winner from two incompatible dashboards.