Data quality
Field coverage and the known defects in this ERP snapshot, with what each one changes in an answer.
Field coverage
Fields the business depends on, lowest coverage first
| Field | Populated | Coverage | Effect on answers |
|---|---|---|---|
| Wedding band gender | 0 / 306 | 0.0% | Unlabelled bands must not be redistributed |
| Order pickup date (completed) | 879 / 1,077 | 81.6% | Life-cycle timing depends on this |
| Diamond type (lab vs natural) | 1,413 / 1,413 | 100.0% | Lab/natural splits must show an unknown bucket |
| Diamond cut grade | 1,413 / 1,413 | 100.0% | Cut-quality questions are mostly unanswerable |
| Diamond landed cost | 1,413 / 1,413 | 100.0% | Margin analysis covers only the costed subset |
Client identity
Duplicate resolution
Customer records
1,100
Rows in the ERP
Resolved clients
1,100
0 records merged
Records merge on matching email, or on phone plus surname. Phone alone is not used: the studio’s own number appears on unrelated records, so matching on it would collapse strangers into one client. Counts of clients use the resolved figure; counts of records do not.
Cancellations
Soft-deleted orders
Cancelled orders
121
Excluded from every order figure
Sentinel dated
0
Cancelled on 2000-01-01
Cancelled orders stay in the table with a deletion timestamp set, so any query without that filter overstates order counts. The sentinel date is a placeholder, not a cancellation date — exclude those rows from any time series.
Payment anomalies
First payment before order date
Payments before order
0
Of 1,458 orders with payments
Share
0.0%
Of orders with payments
These are deposits taken before the order was formally created. Raw day-differences go negative, and averaging them signed drags a salesperson’s collection time toward zero. Timing figures clamp at zero and keep the raw value so the anomaly stays visible.
Known limits
Questions this snapshot cannot answer correctly
Recent cohorts are right-censored
The snapshot ends Jun 29, 2026. The engagement-ring-to-wedding-band return rate looks like it collapses in the latest years; the median gap from ring to band is 10 to 12 months, so those cohorts have not had time to return. Compare cohorts only on an equal observation window.
No cost on stones or ring styles
Non-diamond stones and style details carry no cost column, and there is no cost field on the order. True order profitability cannot be computed from this ERP. Retail on the order line is the only figure available, and it is retail.
Diamond cost covers part of the book
Landed cost is present on roughly 73% of centre stones. Margin computed from that subset is not representative of the portfolio, and it swings widely year to year.