Weekly Pipeline Review: Audit → Prioritize → Forecast

What this chain helps you accomplish

Turn a weekly opportunity export into a reviewable pipeline picture: identify incomplete or misleading records, assess evidence-backed deal health, prioritize up to three deals that genuinely need attention, plan five working days of seller actions, and prepare a defensible forecast-call narrative.

The four prompts work in sequence and produce an auditable snapshot, deal-health and priority register, deal-specific action plans, and a forecast brief with sourced amounts or explicit unknowns. The output helps AEs, managers and RevOps identify where seller effort and leadership assistance may change a real buyer decision.

A Green/Yellow/Red deal-health assessment is not the same as an official CRM forecast category. This chain preserves that distinction, avoids fictional win probabilities and reports missing data rather than manufacturing a Commit number. Promptifi provides the prompts and instructions; sellers run them in their own approved AI tool and retain control over CRM and outreach.

When to use it

• At the start of the selling week before a pipeline review or forecast meeting, using a fresh, date-stamped CRM export—not an assumption that Promptifi runs automatically every Monday.
• When an AE or manager needs to audit in-period open deals, missed customer milestones, stale close dates, qualification gaps and a lack of future buyer engagement.
• Before a manager 1:1 when the team needs to select three practical interventions rather than discuss every deal equally.
• Before reporting Commit, Best Case, Closed Won or quota attainment to leadership when the CRM forecast category, amount basis and reporting period must be reconciled.
• When RevOps or a manager suspects a filtered export, duplicate opportunities, inconsistent currencies, changing close dates or incomplete buyer evidence are distorting the pipeline.

Do not treat the output as a live CRM dashboard, an automated weekly task, a replacement for official forecast policy, or an independent statistical prediction. If the export is a subset, the chain can analyze the subset honestly but cannot claim full-pipeline coverage. If only one deal is supplied, it should produce one useful plan, not invent two more.

Before you begin

Minimum tools

Required: an employer-approved text-based AI assistant (ChatGPT, Claude, Gemini, Microsoft Copilot or a comparable tool) and a permitted CRM export or carefully copied opportunity table. No paid feature, direct CRM integration, scheduled task or web access is inherently required. Optional: a CSV/Sheets export, company forecasting definitions, access to genuine buyer notes and previous-period snapshots. The AI cannot infer live CRM values from a pasted link or account name.

Company CRM forecasting rules govern all official figures. Forecast category labels, rollup types and inclusion rules vary by CRM and configuration; the prompts do not impose one universal forecast formula. The chain is manual; optionally rerun with new data each week, but do not imply the prompts schedule themselves.

Exact starting inputs, sources, examples and missing-data rules

• {{AS_OF_DATE_AND_TIMEZONE}} — Required. Snapshot's analysis date/timezone from report generation; e.g. '2026-11-09 08:00 America/Chicago'. If missing, do not call any close date past or future.
• {{REPORTING_PERIOD_START_END}} — Required. Forecast period boundaries from employer policy; e.g. '2026-10-01 through 2026-12-31, calendar Q4'. If missing, no in-period sums.
• {{PIPELINE_OWNER_AND_TEAM_SCOPE}} — Required. Named seller/team, whether shared opportunities and splits included; e.g. 'AE West team, owned open opportunities only'. If missing, output a partial audit with undefined coverage.
• {{CRM_EXPORT_DATE_SCOPE_AND_COMPLETENESS}} — Required. Export date/time, filters, open/closed status, expected total rows, any chunks and whether COMPLETE or SAMPLE; e.g. 'CSV 2026-11-09, all team open opportunities; 18 rows'. If not verified, mark coverage UNKNOWN and withhold full-pipeline totals.
• {{OPPORTUNITY_ROWS_WITH_UNIQUE_IDS}} — Required. From authorized CRM report. Include ID, name, owner, open/closed, stage, forecast category, amount, currency, revenue measure, current/prior close dates, dated buyer interaction and buyer-confirmed next step, plus qualification data where known; e.g. '006-001, Harborview, AE, open, Evaluation, Best Case, USD 25000 ACV, close 2026-12-10, buyer review 2026-11-03'. No rows means STOP; absent individual fields become UNKNOWN.
• {{FIELD_DEFINITIONS_AND_STAGE_MAPPING}} — Required for stage interpretation. Company's stage and close-date fields; e.g. 'Negotiation means commercial review, Closed Won separate'. Unknown mapping means stage-health implications must be tentative.
• {{FORECAST_CATEGORIES_AND_ROLLUP_RULES}} — Required for official forecast math. Exact CRM category labels, whether single or cumulative rollups and which statuses count; e.g. 'Pipeline/Best Case/Commit/Closed; single categories; manager adjustments separate'. If missing, do not claim a verified official Commit figure.
• {{CURRENCY_REVENUE_METRIC_AND_SPLIT_RULES}} — Required for summation. Amount unit (ACV/ARR/TCV/bookings), currency and owner split/credit definitions; e.g. 'USD ACV, 100% owner credit, no splits'. If mixed or unknown, segregate instead of adding.
• {{QUOTA_CLOSED_WON_AND_ATTAINMENT_SOURCES}} — Optional. Same-period quota, Closed Won actuals and attainment from authorized report; e.g. 'USD 200000 ACV quota; USD 120000 closed won this quarter'. Missing/misaligned denominator means no attainment percentage.
• {{OFFICIAL_FORECAST_AND_MANAGER_ADJUSTMENTS}} — Optional. Actual dashboard Commit/Best Case figure, adjustments and snapshot date; e.g. 'official Commit 65000 USD ACV as of 08:00; no manager adjustment reported'. If missing, separate CRM category sums from official forecast.
• {{COMPANY_HEALTH_OR_STAGE_EXIT_CRITERIA}} — Optional. Actual manager policy for what counts as healthy and stage-validated; e.g. 'Proposal requires named buyer decision owner and mutually confirmed evaluation'. If none, use the chain's explicitly labeled editorial rubric.
• {{BUYER_INTERACTIONS_AND_EVIDENCE}} — Recommended. Dated buyer call/email excerpts, next action accepted by buyer, sponsor/process proof; e.g. 'D001 buyer confirmed security review on Nov 12'. If not supplied, do not mistake seller tasks for buyer progress.
• {{QUALIFICATION_GAPS_AND_SOURCE_NOTES}} — Optional. Genuine MEDDIC/MEDDPICC gaps, legal, security, budget, decision-maker evidence and sources; e.g. 'economic buyer unknown; seller note 2026-11-05'. Missing fields are UNKNOWN, not proof a deal is dead.
• {{SELLER_CAPACITY_AND_ALLOWED_ACTIONS}} — Required for realistic plans. Owner workload, approved engagement rules, resources; e.g. '2 hours/day on priority deals; no automatic email send'. If missing, ask for capacity and use small conditional plans.
• {{THIS_WEEK_BUSINESS_DAYS_AND_TIME_OFF}} — Optional. Actual Monday–Friday workdays, holidays and known leave; e.g. 'Nov 9–13, exclude local holiday'. If unknown, use relative WORKDAY 1–5, never invent dates.
• {{FORECAST_CALL_AUDIENCE_AND_REQUIRED_FORMAT}} — Optional. Sales manager, regional VP or RevOps, desired duration and definitions; e.g. '3-minute spoken report to regional VP'. If missing, create a short neutral spoken version.
• {{LEADERSHIP_HELP_AVAILABLE_OR_REQUESTED}} — Optional. Real leadership resources or permission to request them; e.g. 'security specialist can review high-risk deal if manager approves'. If absent, suggest conditional asks only when sourced issues justify them.
• {{AI_DATA_PRIVACY_AND_CRM_RULES}} — Required. Employer-approved tool, redaction, permitted notes/financial fields and CRM modification authority; e.g. 'sanitized export in approved assistant, analysis read-only'. If unclear, avoid uploading restricted rows.

Prepare the weekly snapshot safely

1. Export or copy the relevant open pipeline at one known snapshot time, retaining unique CRM IDs, original amounts and currencies. Include Closed Won only as a separate source for period attainment, not as part of open pipeline.
2. Write the forecast period, quota basis, measures and category rollup method before calculating anything. If data comes in batches, preserve the same snapshot and include every chunk before calling results complete.
3. Collect last actual BUYER interaction and next accepted buyer step for priority deals. A seller follow-up task is not customer engagement.
4. Label MEDDICC/MEDDPICC, paper process, security/procurement, executive sponsor and budget facts with dates and source locations when available. Unknown is acceptable.
5. Remove private contact details, restricted commercial data or customer secrets unless organizational AI policy explicitly permits them. Work within the employer's approved CRM and AI rules.

Primary-source guidance: Salesforce explains Forecast Categories and single versus cumulative rollups (https://help.salesforce.com/s/articleView?id=sf.forecasts3_customizing_forecasts_categories.htm&language=en_US&type=5 ; https://help.salesforce.com/s/articleView?id=forecasts3_cumulative_columns_overview.htm&language=en_US&type=5). HubSpot's sales workspace distinguishes stalled deals, stale close dates and missing future activity (https://knowledge.hubspot.com/prospecting/create-and-manage-deals-in-the-sales-workspace). Those practices inform risk checks but do not replace your company's field mapping.

How to run the chain

1. Open one new conversation in a company-approved AI assistant. Fill and paste the complete STARTING CONTEXT from Chain Steps, including a trustworthy dated CRM export. If the pipeline is large, paste labeled batches from the same snapshot with unique CRM IDs; keep all rows in scope.
2. Run Prompt 1. Review source completeness, duplicates, period boundaries, currency/revenue measure, source labels and official forecast definitions. Correct records before approving PIPELINE_SNAPSHOT_QA_v1; without source rows no audit is possible.
3. Run Prompt 2 with the full approved snapshot. Review the Green/Yellow/Red/UNASSESSED rubric, deal-specific rationale, independent forecast categories and R1–R3 top priorities. Approve DEAL_HEALTH_AND_PRIORITY_v1 or request a documented different priority.
4. Run Prompt 3 with both approved artifacts. Review the five-workday plans for realistic seller capacity, buyer-first asks, measurable proof and escalation requirements. Approve FIVE_DAY_DEAL_ACTION_PLANS_v1.
5. Run Prompt 4 using all three approved artifacts. Check the financial reconciliation against the actual official CRM report: open versus Closed Won, Commit versus health color, single versus cumulative categories, quota alignment, and any manager adjustment. Approve FORECAST_CALL_BRIEF_v1 only after reading it as the leader who will hear the numbers.
6. Before the forecast call, compare the snapshot to any new buyer replies or CRM edits that occurred after export. If materially changed, revise the earliest affected artifact and rerun its dependent prompts.
7. If a prompt invents a buyer milestone, close date, probability or official figure, quote the defective line, supply the real source and request a COMPLETE corrected version. Never approve the original alongside a conflicting revision.
8. If moving to another conversation, paste the full allowed Starting Context plus every latest seller-approved artifact required for the next prompt. There is no cross-chat assumption of CRM access or preserved approvals.
9. Use resulting recommendations manually: conduct actual buyer conversations, request leadership help, and update CRM only through your approved normal process. Running the prompts does not schedule a weekly automation.

Artifact approval and versioning

Use 'APPROVED — PIPELINE_SNAPSHOT_QA — v1', 'APPROVED — DEAL_HEALTH_AND_PRIORITY — v1', 'APPROVED — FIVE_DAY_DEAL_ACTION_PLANS — v1' and, after review, 'APPROVED — FORECAST_CALL_BRIEF — v1'. A corrected v2 supersedes v1 only when expressly approved. Seller approval is not CRM validation or buyer commitment.

What success looks like

You have a clearly scoped and reconciled snapshot; evidence-backed deal health, including unknowns; a feasible top-three weekly action plan; and a forecast narrative that identifies verified figures, challenged Commit deals and concrete leadership decisions rather than hiding uncertainty.