There’s a tell that gives it away in the first ten minutes of almost any ops conversation: the spreadsheet. Somewhere between the CRM and the board deck, a number gets “corrected” by hand — an export, a few adjustments everyone’s stopped questioning, a version of the truth that lives in a file called something like `pipeline_FINAL_v3_USE_THIS`.
Nobody decided to stop trusting the reporting. It happened gradually, and in a predictable order. Understanding that order matters, because trust has to be rebuilt in the reverse order it was lost — and most reporting-repair projects fail by starting at the wrong end.
How trust dies: the four-step decay
Step 1: A number is visibly wrong. A rep spots a deal in the wrong stage; a marketer sees a campaign double-counted. Small, specific, fixable — but usually not fixed, because it’s nobody’s job.
Step 2: People build workarounds. Rather than fix the source, individuals build their own views: a rep’s personal spreadsheet, a marketer’s separate attribution export. Each workaround is locally rational and globally corrosive — now there are multiple truths, each with an owner defending it.
Step 3: Meetings become number reconciliations. The pipeline review spends its first twenty minutes arguing about whose number is right instead of what to do. This is the stage most companies live in permanently.
Step 4: Leadership routes around the data. Decisions revert to gut feel and the loudest voice, while dashboards keep getting built and ignored. At this point, buying a BI tool — the most common response — is like repainting a house with a cracked foundation.
Why “better dashboards” can’t fix it
Dashboards sit at the top of a stack: data → definitions → metrics → dashboards. Trust broke at the bottom (wrong or inconsistent records, drifting definitions) and everyone experiences it at the top. Rebuilding from the top — nicer visuals over the same disputed inputs — just makes the disagreement prettier. The rebuild has to run bottom-up.
The rebuild sequence
1. Fix one number completely. Not the reporting suite — one metric. Pick the most argued-over number (usually MQLs or pipeline created) and make it bulletproof: written definition, clean underlying data, one system of record, documented calculation. This takes real work — it may involve hygiene repair and integration mapping — but scoping it to one metric makes it finishable in weeks, not quarters.
2. Get public sign-off on that number. Both marketing and sales leadership agree, in the same meeting, that this number as now defined is the number. The public agreement is the point: it converts “the data is wrong” from a general vibe into a specific, testable claim.
3. Retire the workarounds for that metric — ceremonially. The shadow spreadsheets covering that metric get deleted, and it’s said out loud that they’re deleted. Workarounds that survive quietly become Step 2 of the next decay cycle.
4. Put a maintenance ritual around it. A monthly 15-minute integrity check: does the number still reconcile across systems? Any new manual corrections creeping in? Trust decays without maintenance exactly like routing rules and scoring models do.
5. Repeat, one metric at a time. Each rebuilt metric is faster than the last, because the underlying repairs compound. Most teams need only four or five bulletproof numbers — MQLs, pipeline created, conversion by stage, velocity, and CAC-adjacent spend — to run the revenue conversation. Bulletproof five numbers beats plausible fifty.
The number to rebuild first is almost always the one leadership argues about most — usually MQLs or pipeline created — because trust returns fastest when the most contested figure becomes the reliable one. The rebuild is deliberately public: agree the definition in the room, show the query behind it, and have both marketing and sales sign off that this is now the number. The sign-off meeting is the point — it converts a disputed figure into a shared one. After that, the manual spreadsheet “adjustments” fall away one metric at a time, because there’s no longer a gap between what the CRM says and what people believe.
The cultural half of the fix
One behaviour predicts whether trust holds: what happens when someone finds an error after the rebuild. If errors are treated as evidence the project failed, people go back to hiding them in workarounds. If they’re treated as maintenance — logged, fixed at source, closed in the monthly ritual — the system self-heals. Announce that policy explicitly. Reporting trust isn’t a data-quality state; it’s an agreement about how errors get handled.
—
Attribution and reporting integrity is lens four of the Pipeline Leak Audit — including tracing your most-corrected number back to its source. Start with a free health check if you want a read on where yours is breaking.