Governance·Essay·7 min·Updated 09 Apr 2026

Completeness scores: the only metric that moves

If your PIM dashboard shows one number and the rest of the team can recite it from memory, the implementation is working. If the dashboard shows fifteen metrics nobody remembers, start over.

§ 01 — what completeness is

The one metric that actually correlates with revenue.

Completeness is the percentage of required attributes that have a non-empty value. That's it. The sophistication is in what counts as required — and for whom.

A flat completeness number is nearly useless. A per-channel, per-attribute-group completeness score is the single most powerful governance metric in product content, because every retailer onboarding gate and every ad network's content scoring rewards the same thing: complete, channel-shaped data.

§ 02 — how to instrument it

Design the model, then the dashboard.

  • Group attributes into regulated · logistics · marketing · media · optional. Never blend.
  • For each group, set a minimum per channel — not one global target. The F&B and Retail playbooks spell out realistic numbers.
  • Block publish when regulated-group completeness falls below threshold. Warn, don't block, for marketing.
  • Expose the score in the authoring UI next to the SKU, not only in a weekly report nobody reads.
  • Trend it. A flat 87% score means nothing; a falling trend is the alarm.
§ 03 — thresholds

Targets that actually survive audits.

Regulated (nutrition, compliance)
98%+
block publish; legal signoff trail
Logistics (GTIN, weight, dims)
95%+
retailer onboarding gate
Marketing copy (per channel)
85%+
soft-launch OK below
Rich media (primary + alternates)
70%+
rolling workstream
§ 04 — the trap

Completeness without relevance.

The single most common failure mode: a PIM dashboard showing 94% completeness while Amazon and Zalando keep rejecting listings.

Root cause: the 'required' set was defined by whoever owned the PIM, not by the channel. If your required-attribute set doesn't map to each channel's onboarding spec, a 100% score tells you your catalog is complete for nobody.

Fix: write the channel spec as a schema, import it into the PIM as a per-channel required set, and score against that. Every mature PIM supports this; most stock implementations skip it.

The testIf your completeness score went from 72% to 94% in a quarter and nothing downstream got easier, the score is lying to you. Re-anchor it to channel onboarding requirements.