Forecast Numbers That Don't Add Up

The 40% gap is structural, not a data bug. Forecasts roll up the forecast hierarchy, which by default is the role hierarchy (or the territory hierarchy once enabled). If 30% of accounts are owned by users whose Industry role does not sit beneath the division VP's role, those opportunities never enter her forecast no matter how well the rep forecasts. Roughly 30% of accounts plus some period/omitted losses explains 40%.

Commit equals Best Case because the columns are cumulative: Best Case = Commit + Best Case categories. Identical totals mean nothing is sitting in the Best Case-only category. Check the stage-to-forecast-category mapping and whether reps can override their forecast category; if every open stage maps to Commit, or overrides were bulk-set to Commit, the category collapses.

Verification order

1. Confirm the forecast hierarchy setting and draw the actual hierarchy; identify which users are not beneath the VP.
2. Report her division's opportunities grouped by owner role, and compare to the forecast total.
3. Check stage-to-category mapping, override ability, and current override values.
4. Check exclusions: no close date, close date outside the period, Forecast Category = Omitted, owners who are inactive or queues.

Do all of it with View All Forecasts in a sandbox, and reconcile the forecast to a report total before changing configuration.

Territories are the correct model if the matrix is permanent and per-account. A territory model assigns accounts to sellers independent of record ownership, which is precisely the problem here, and the forecast hierarchy becomes the territory hierarchy.

The trade-offs you must state

• Only one territory model can be active at a time, and an account that is assigned to no territory silently vanishes from forecasts. With a few thousand accounts you need assignment rules plus an ongoing audit for unassigned accounts.
• Role-based and territory-based forecasting cannot run side by side. Cutting over changes every user's forecast view and every forecast report, and effectively costs you a quarter of usable forecasting during transition. Switch at a period boundary, and rehearse in a full sandbox with production-like data.
• Territory maintenance (reps joining, leaving, changing patch) becomes a new recurring admin task.

When to fix roles instead: if only a handful of users are misaligned, create division roles under the division VP and move those users. That is far cheaper, but moving a user's role changes record visibility for them and possibly for their old subtree, so plan sharing rules before you move anyone. If 70% of accounts already roll up correctly, fix roles first, measure the residual gap, and only then consider territories.

Prerequisites

• Campaign Influence enabled, campaigns with actual members and responses, and the campaign hierarchy in usable shape.
• A chosen attribution model: first touch, last touch, even distribution, or a custom model. Only one model can be the auto-association (primary) model; the others are calculated on demand and can go stale between refreshes.
• A decision on the attribution window, since that defines which campaigns can claim an opportunity.

Sequencing constraints

• Enabling it adds Primary Campaign Source to Opportunity, which changes page layouts, permission sets, and the sales team's behavior mid-quarter. Treat it as a change-management event, not a setup checkbox.
• There is effectively no historical backfill. Influence is recorded from enablement forward, so turn it on before the next campaign wave, not after, and say so plainly rather than promising last year's ROI.

What will disappoint them: without campaign member hygiene, influence will be empty or dominated by whichever campaign happens to be added first. Pilot with one campaign, prove the members are being created reliably, then enable.