Reconciling Three Different Lead and Pipeline Numbers

The numbers are not measuring the same thing. GA4 conversions of 900 count actions such as form submissions and demo requests, and one person can convert multiple times. LinkedIn's 1,100 leads are platform-reported raw leads, often inflated by duplicate submissions and low-intent form fills. HubSpot's 420 MQLs are contacts that passed a marketing qualification rule, while 180 sales-accepted and 46 opportunities are later stages with sales judgment. The webinar's 640 registrants are registrations, not necessarily attendees or leads.

Reconcile in this order:

1. Deduplicate contacts by email and company domain for the Q3 cohort.
2. Define one conversion event set and map platform events to it.
3. Apply the same date range and attribution lookback to each source where possible.
4. Rebuild campaign UTMs so channel, source, campaign, and content are consistent.
5. Separate raw leads, MQLs, sales-accepted leads, SQLs, opportunities, and pipeline value.

The board should not see a single blended 'leads' number without stage labels. A reasonable headline is raw leads by channel, then MQL, sales-accepted, opportunity, and pipeline created. The fact that MQL 420 exceeds sales-accepted 180 is not a data error; it is a qualification gap that the dashboard should expose.

No platform should be the sole source of truth for cross-channel budget decisions. Use HubSpot as the source of truth for contact-to-deal progression and pipeline, GA4 for web behavior and traffic, and each ad platform for spend and impressions. For budget allocation, use a consistent multi-touch model with a defined lookback, such as a 90-day position-based model that gives credit to first touch, lead creation, and opportunity creation. This fits a B2B cycle with webinars and sales follow-up better than pure first-touch or last-touch.

Set rules:

• All campaigns must use the UTM taxonomy; unmapped campaigns are reported separately, not silently assigned.
• Offline meetings and outbound calls stay in HubSpot and are included in pipeline reporting, but not blended into web conversion counts.
• LinkedIn's 30-day click number is used for platform optimization, not for board-level pipeline claims.
• GA4 data-driven attribution is useful for web diagnostics but should not be mixed with HubSpot deal attribution without a mapping.

For the board, present pipeline influenced within the lookback window plus a smaller set of controlled tests or holdout analyses. That gives a defensible answer about which channel drove pipeline while acknowledging that B2B buying is multi-touch.

The dashboard should have three layers. The executive layer shows spend, raw leads, MQLs, sales-accepted leads, opportunities, pipeline created, closed-won revenue, and cost per accepted lead or opportunity by channel. The diagnostic layer shows conversion rates between stages, average days to follow-up, campaign-level performance, and webinar attendance versus registration. The data-quality layer shows unmapped campaigns, missing UTMs, duplicate rate, contacts without a source, and deals without a campaign association.

Do not let every team define its own MQL. Publish one lifecycle definition with required evidence and one set of campaign naming rules. Assign an owner for marketing operations and a monthly close process: export the Q3 cohort, deduplicate, map campaigns, check stage regressions, and lock the numbers before the board review. Changes to attribution windows or lifecycle rules should be versioned and announced.

The dashboard should answer two questions: where is pipeline coming from, and where should the next dollar go? If the data cannot answer both, add a note rather than a false precise number. A monthly data-quality review is cheaper than rebuilding trust after the board finds three conflicting cost-per-lead figures.