Non-visual mirror Northstar Home · no customer data
The Silent Bestsellers
Which commercially strong products are invisible to advertising—and what is safe to repair?
Narrative sequence
- Two shelves — Internal commercial proof and advertising reach begin as separate product populations.
- Identity mismatch — Stable product IDs expose silent winners, unsupported spend and unmatched records.
- Product autopsy — One product expands into identity, stock, price and eligibility evidence.
- Policy fork — Fresh deterministic defects enter proposed repair; stale or ambiguous defects require a person.
- Exact read-back — Returned state must equal proposed state while 1-hour and later checkpoints remain OPEN.
- Economic truth — Margin and refund evidence changes priority; incremental demand remains UNKNOWN.
- Action queue — Products finish in repair, review, measure, stop or UNTESTABLE lanes.
| Evidence | Value | Status / meaning |
|---|---|---|
| Ready to compete | 62% | 84% |
| Needs attention | 23% | 9% |
| Blocked | 15% | 7% |
| Missing identifiers | 18 | human review |
| Price mismatch | 11 | read-back |
| Image policy | 7 | monitor |
| Deterministic repair | 9 | proposed |
| Returned state | 9 | exact match |
| 1-hour checkpoint | OPEN | monitor |
| Contribution evidence | UNKNOWN | holdout required |
Trivial baseline
Catalogue eligibility alone: rank products by feed status without internal sales, margin, stock or refund evidence.
Sensitivity
Re-rank the repair queue under lower contribution margin and a zero-incrementality assumption; restored eligibility remains operational evidence, not revenue.
Strongest counter-case
Restored eligibility does not prove incremental demand; organic sales may have happened anyway.
Verdict
SURVIVES: repair priority. UNTESTABLE: incremental sales without a holdout.
Synthetic example. Deterministic fictional records; not customer data or achieved customer results.