Pipeline Reporting Built on Daily Deal Snapshots

You cannot answer "what did this quarter look like on day thirty" from live CRM records, because the fields have been overwritten since. Salesforce keeps stage and amount changes in OpportunityFieldHistory, HubSpot keeps property history, and both are awkward to query and get pruned. The practical consequence is that nobody can measure slip, meaning how much of a shortfall was deals moving out rather than deals being lost, and slip is usually what the board is actually asking about. How it works: A daily job writes every open opportunity into a snapshot table in your warehouse, one row per opportunity per day, keyed on date. It runs before your CRM's overnight automation, not after. From roughly sixty days of snapshots the reporting starts being useful; from two quarters it starts being predictive. What snapshots make possible: Stage conversion measured on the entry cohort rather than the exit cohort. This matters more than it sounds. Counting conversions from deals that left a stage this month flatters your win rate whenever deals sit, because the ones still stuck are invisible. Counting the cohort that entered the stage produces a number that moves when reality moves. Slip and push tracking, by rep and by segment: which deals moved close date, how many times, and how many days each move added. A rep who pushes the same deal three times is a different management conversation than a rep who loses cleanly. Coverage measured against your own trailing win rate rather than the three-times rule somebody repeated at a conference. If you close 31 percent, three-times coverage is optimism. Forecast bias per rep. After two quarters of snapshots you can state, per person, how far their commit historically overstates or understates the outcome. That single number makes a forecast meeting shorter. Reports we ship: A period waterfall, opening balance through created, won, lost, slipped in, slipped out, expanded and contracted, that reconciles to the closing number without a plug figure. Stage-to-stage conversion by entry cohort. Aging by stage with an explicit rule for when an opportunity is dead regardless of what its owner says. Forecast versus actual with per-rep bias applied. Built as dbt models in your warehouse and surfaced in whatever your team already opens: Looker, Lightdash, Metabase, Omni, or native CRM reports sitting on the same definitions. We do not resell a BI licence and have no interest in which one you chose. We will not build: Activity dashboards. Calls dialled and emails sent are inputs, and putting them on a management screen produces the behaviour you measured rather than the behaviour you wanted. If you want them for coaching, build them yourself; we will not present them as performance reporting. A forecast-accuracy report for a team that does not maintain close dates. It will be arithmetically correct, completely uninformative, and quoted anyway. The board deck. We build the data layer it stands on and we will sit in the meeting where it is questioned, but the narrative is yours. Too small for this: Under about fifty opportunities a quarter, a maintained spreadsheet answers every question above and costs nothing. We will tell you that on the first call. With under two quarters of usable history we can start the snapshot job now so the reporting exists later, but we will not model conversion rates on a sample that thin and call it a forecast.