Blog · Data quality and reconciliation
The ten questions a head of sales operations asks before anything is reported, does the CRM reconcile to the ledger, does every account have one owner, do the territories' quotas sum to the target, are the splits summing to one, is the pipeline clean, has any identifier drifted, which reps' data can be believed, did the definitions change, did the export match its control total, and what is on the exception list, each with the check that answers it, the identity behind it, and why sales operations owns the checks that everyone else's numbers rest on.
Every leader's numbers rest on ten checks that sales operations owns. If the checks hold, the lists and the tables mean something; if they do not, nobody's numbers do. This guide is the ten questions a head of sales operations asks, the check for each, and what happens when one fails.
| # | The question | The check | Identity | On failure |
|---|---|---|---|---|
| 1 | Does the CRM reconcile to the ledger? | Five mismatch types, listed | Attributed revenue equals ledger | Unowned revenue listed; coverage greyed for affected reps |
| 2 | Does every account have one owner? | Assignment check on the date | Exactly one owner per active account | Double-owned accounts excluded and listed |
| 3 | Do the quotas sum to the target? | Territory quota identity, after moves | Σ quotas = target | Plan flagged; attainment not computed |
| 4 | Do the splits sum to one? | Crediting identity; overlay apart | Σ shares = 1 per line | Lines listed before the commission run |
| 5 | Is the pipeline clean? | Five hygiene checks | Open pipeline = clean + flagged | Flagged share shown on every forecast |
| 6 | Has any identifier drifted? | Vanished, appeared, paired, threshold | Map applied before measures | Run stopped past threshold; pairs proposed |
| 7 | Whose data can be believed? | Data quality score per rep | Passing ÷ records | Dependent measures greyed under the floor |
| 8 | Did the definitions change? | Version on every table | Version matches or the change is noted | Movements flagged as definition changes |
| 9 | Did the export match its control total? | Validation line 2 | File total = source total | Upload rejected or marked unreconciled |
| 10 | What is on the exception list? | All failures, with owner and deadline | Every failure is a line | Unresolved at deadline: excluded and told |
Nine first: if the export is wrong, nothing after it matters. Then six, because a re-keyed base breaks everything. Then one and two, the book. Then three and four, the plan and the credit. Then five, seven and eight. Then ten, which is the list of everything that failed.
| Check | Result |
|---|---|
| 9 | Ledger $4,182,400 = control; CRM activities 48,211 = control. Pass |
| 6 | 3 new accounts, 1 rep re-keyed (R-19 → R-04), mapped. Pass |
| 1 | 210 ledger customers with no CRM owner, $1.9m. Listed; coverage greyed for the region |
| 2 | 2 accounts with two owners. Excluded, listed |
| 3 | Σ quotas $14.0m = target after two moves. Pass |
| 4 | 2 lines with splits at 1.4. Listed for the commission run |
| 5 | 36% of weighted forecast flagged. Shown on the forecast |
| 7 | R-04 at 41%; coverage and untouched greyed for R-04 |
| 8 | v2026.3, unchanged. Pass |
| 10 | 7 items, 3 owners, deadline Thursday |
Ten lines, five minutes, and the head of sales operations knows exactly which of the CRO's lists to trust this week.
Checks run by nobody. Every leader's numbers rest on nothing.
Failures fixed in the report. The source is still wrong next month.
Check nine skipped. A half-month export analysed as whole.
No exception list. The failures are known and unowned.
Covirage runs the ten on every upload and produces the exception list with owners. The metrics governance solution describes the setup, and the validation report guide covers the six lines the first checks are read from.
Because they are about the data, not the sales result, and the person who runs the CRM, the assignments and the quota file is the one who can fix a failure. The CRO reads the lists; sales operations makes sure the lists are built on a book that reconciles.
Minutes, when they are computed on every upload. The first time, a week, because the failures are discovered. After that, the checks run, the exception list is short, and the head of sales operations reads ten lines on Monday before anyone else reads anything.
The dependent measures are greyed or suppressed and the failure is on the exception list with an owner and a deadline. A coverage figure on a rep whose accounts have two owners is not shown as a coverage figure; it is shown as a check that failed.