Rule 1 — every deal has an exit criterion for its current stage
The stage isn't the point; the exit criterion is. 'Discovery' means nothing on its own — 'discovery complete when we have a signed mutual action plan and named economic buyer' is a filter. Write the criterion once per stage on the CRM stage itself. Every Monday, look at every deal in a stage and ask 'does it meet the exit criterion for the NEXT stage?' If yes, move it. If no, push the close date, don't leave it.
Rule 2 — no deal survives two consecutive slip weeks
If a deal's close date slips two Mondays in a row without a specific customer-side reason (procurement queue, board meeting, budget cycle) recorded in the notes, it goes to a 'nurture' stage and stops counting against the current-quarter forecast. This is the single biggest lever against sandbagging AND happy-ears optimism. Reps hate it for one quarter and love it for the rest of their career.
Rule 3 — the forecast is the sum, not the vibe
A weighted forecast is the sum of (deal value × stage probability). The stage probability is set by the pipeline, not by the rep. If your close rate from 'proposal' is 42%, then 'proposal' deals contribute 42% of their value to the forecast. Reps do not adjust probability per deal — they adjust the STAGE per deal. This one rule kills 'best case / worst case / commit' theatre and replaces it with a number a CFO can plan against.
Rule 4 — every review names the next customer commitment
Not 'I'll send a follow-up' — that's a rep commitment. A customer commitment is 'they've agreed to book a security review with their IT lead on the 15th'. If a deal has no next customer commitment, it isn't progressing. The rep either books one this week or the deal moves to nurture. This is the single fastest way to identify deals that are one-sided conversations.
Rule 5 — the pipeline is a habit, not a report
Run the ritual at the same time every week (Monday 9:15am works well), under 30 minutes for a book of 25 deals, with the CRM open on the shared screen. Do not skip it for anything short of a national holiday. Consistency is the point — the value compounds over quarters, not weeks. After 8–10 weeks you will notice your forecast is boring and accurate, which is the goal.
Pipeline hygiene isn't a CRM feature — it's a five-minute ritual with five rules. Songbird enforces the ritual with saved views and stage exit criteria, but the discipline is the product.