Attribution Alignment Between Sales and Marketing

In most B2B pipelines, the source a rep types on the opportunity and the last-touch attribution the CRM calculates disagree on 30 to 50 percent of closed-won deals. Neither side is lying. The rep is describing the conversation that moved the deal; the system is describing the last measurable click. The work here is not building a better model. It is getting both teams to sign one written definition of source and then enforcing it at the field level. The deliverable: An attribution charter, four to six pages, signed by whoever owns marketing spend and whoever owns quota. It states what counts as a source, when each stamp is written, and who may change it afterwards. First touch is stamped at contact creation and locked. Opportunity source is stamped at opportunity creation and editable only by the deal owner, with a reason, and only before the deal reaches the stage where forecasting starts. Every ambiguous case that surfaced in the audit is written down with the ruling, so the argument is settled once instead of quarterly. Self-reported data: An open-text "how did you hear about us" on the demo request, read by a human weekly for the first two months, then collapsed into a picklist built from what people actually typed rather than what the marketing team hoped. This is the only instrument that sees podcast mentions, Slack communities, conference hallways and a colleague's recommendation from a previous job. It is imprecise, and it is still better than any model for channels that produce no click. Enforcement: Validation rules in Salesforce that block the move to Closed Won when the source reason is empty. Required properties gated on deal stage in HubSpot. The field removed from inline list editing so it cannot be bulk-overwritten by someone tidying a view. Campaign Influence configured with one model and one only, because two models in the same org guarantees two versions of the number in two decks. Reconciliation: Each month, three figures printed side by side: what the ad platforms report, what the CRM says on last touch, what the reps entered. We do not force them to agree and we will not build a report that pretends they do. We publish the delta and name its causes: view-through windows, cross-device, platform-modeled conversions, and the rep who types "referral" for everything. After two cycles the executive team stops asking which number is right and starts asking which number answers the question in front of them. That shift is what you are buying. Tools, where they earn it: HubSpot attribution reports, Salesforce Campaign Influence, and warehouse-level modeling in Dreamdata or HockeyStack when volume supports it. GA4 is used as a channel-shape check and never as a revenue source. Where we stop: We do not make media budget decisions for you. We do not run incrementality tests inside this scope, though we will say plainly when a geo holdout is the thing you actually need and attribution will not answer your question. We will not build a multi-touch model that assigns fractional credit summing to one hundred percent and also reconciles to platform-reported conversions; those are different accounting systems and one of them has to govern a decision. We do not rebuild your CRM here. This is wrong for you if: Your business closes fewer than about forty deals a year. The sample cannot carry a touch-weighted model and anyone selling you one is selling noise; self-reported answers and a clean first touch are the honest instruments at that size. It is also wrong if the model is expected to support a budget decision that has already been made.