Sales management
Explain a pipeline change from two snapshots with AI
Reconcile record-level entries, exits and amount changes before writing a weekly pipeline narrative.
The change in a total is easier to explain when every deal has a place in the bridge.
Editorial workflow guide · AI-assisted preparation · Not a hands-on product test. How we work
Freeze the population and amount basis
Take two approved snapshots with stable opportunity IDs, original amount, currency, stage, owner, expected close and relevant report filters. Use one agreed reporting currency and rate basis for this comparison, or reconcile conversion separately. Record the snapshot times and whether the report counts all open work or a selected forecast period.
Ask AI to match record IDs and identify missing fields. Do not match only by opportunity name: a rep can rename a deal while it remains the same record. Keep an unmatched row unresolved until the report owner decides whether it is new, removed, inaccessible or missing from the export. A disappearing record is not automatically a lost opportunity.
Sponsor
SPONSORED
AI FOR YOURSales team.Less busywork.More selling.
Your team knows sales.
We help put AI to work.
Work a bridge that balances
A fictional opening pipeline contains D1 at $10,000 and D2 at $20,000, totaling $30,000. In the closing snapshot, D1 is $12,000, D2 has left the selected open population, and new D3 is $8,000. The bridge is opening 30,000 plus 2,000 for D1, minus 20,000 for D2’s exit, plus 8,000 for D3: closing $20,000.
That arithmetic does not explain why D2 left. Check its documented stage and filter history. It may be won, lost, reassigned outside the report or moved to another period. Label the reason unknown until supported. Keep entry and exit reasons separate from amount changes on records present in both snapshots.
Choose one treatment for simultaneous changes
A deal might change amount and leave the population on the same day. State whether the exit uses its opening amount or a reconstructed amount immediately before exit, and avoid counting both inconsistently. If the source cannot support a detailed event sequence, use a simpler opening-to-closing bridge and disclose the limitation.
HubSpot’s sales analytics documentation describes reports and filters that can affect what a team sees. Confirm those definitions before reconciling exports. AI should organize the explanation and propose evidence questions; a spreadsheet or script should calculate totals and residual differences. The model’s narrative must not fill a missing reason with a plausible story about buyer intent.
Publish the reconciliation and the uncertainty
The final review includes opening total, entries, exits, common-record changes, closing total and any unreconciled difference. Link a small record-level schedule so the manager can verify the headline. Highlight data corrections and filter changes separately from actual commercial events. Do not count newly visible records as newly generated revenue.
Test the method using fictional renamed, duplicated, reassigned and currency-changed records. Have the report owner approve the categories before applying it to real weekly reviews. This is a proposed reporting discipline with illustrative amounts, not a tested CRM integration, forecast guarantee or evidence that AI caused pipeline growth.
A prompt to adapt
Match these approved snapshots by stable deal ID using the supplied population and amount basis. Build opening + entries - exits + common-record amount changes = closing, with a residual check. Preserve unknown exit reasons and separate filter/currency changes. Show formulas and record evidence; do not infer wins, losses or buyer intent.
Use fictional or anonymized examples while testing. Add your approved business rules before using a draft.Source documents
Primary sources support the references above. These workflows are editorial suggestions; confirm current product availability, pricing, permissions, and integration details directly.
Questions or corrections?
See our editorial policy for how we distinguish workflow suggestions, vendor statements, and tested results.