Completeness and readiness
Completeness measures each product against its family's contract: which attributes are required, and, where the family says so, for which channels. Because an attribute can be required for some channels only, the answer depends on which channel you ask about.
Read Channels, locales and completeness for the concept. This page is about using it.
Use it as a work list, not a score
"The catalog is 87% complete" is not something anybody can act on. These are:
- Which products are below 100% for the German web shop?
- Which attributes are missing most often across them?
- Which products became incomplete this week?
Get there by filtering the product grid on completeness, then grouping by family. Missing data clusters by family far more often than it scatters. One family whose supplier feed lacks a field accounts for most of the gap, and fixing that one feed fixes hundreds of products at once.
How the number is produced
| Measured against | The attributes the product's family marks Required, narrowed by Required channels where set |
| Measured when | You assign a family (Assign family measures in the same call) or run Recompute completeness on a product or a selection in the grid |
| No family | No score. The product has nothing to be measured against; it sits in the Without a family queue |
| Never measured | A product that arrived through an import with a family but has not been measured; its own queue in the grid |
| Stored | On the product, in the Completeness section of its detail page, until the next measurement |
The score does not move on its own. After an import or a bulk change, measure the products you touched, or the grid shows last week's number.
The weekly loop
- Filter products where completeness for your primary channel is below 100%.
- Group by family. Find the largest cluster.
- Look at which attribute is missing across that cluster.
- Ask why: is the data absent upstream, is a mapping wrong, or is the attribute marked required when it should not be?
- Fix the cause, then recompute.
Step 4 is the one that matters. Filling 300 products by hand treats the symptom; fixing the field mapping treats the cause and stops it recurring.
When completeness is stuck at a number
If a family sits at exactly the same percentage for weeks, an attribute is usually marked required that nobody can ever supply. That is a modelling decision to revisit, not a data problem. See Families.
Required should mean "we cannot publish without this". Anything aspirational marked required turns completeness into noise, and once people stop trusting the number they stop looking at it.
Completeness is not quality
It measures presence. An attribute filled with -, tbd or siehe Datenblatt counts as complete.
For catching wrong rather than missing data, see Data quality in practice.
Before a launch
For a new channel or language, filter on completeness for that channel rather than overall. A catalog that is 100% for your German shop can be far from ready for an English punchout catalog, and completeness per channel is the report that tells you so before your customer's procurement team does.