## TL;DR

Einstein scoring refuses to run when its input data is too dirty or too thin: missing fields it scores on, too few historical records, or extreme skew. Fix the data it needs (complete the key fields, ensure enough history), then re-enable scoring. The model is fine; its diet is not.

## Error

```text
Einstein scoring failed: insufficient data quality. Complete the required fields and retry.
```

## Steps

1. Open the scoring setup to see which fields the model uses. Expected: the list of input fields.
2. Check field completeness on recent records for those fields. Expected: you find the gaps, often source, industry, or close date.
3. Backfill the missing values where possible (enrichment or manual). Expected: completeness rises.
4. Confirm enough historical records exist (scoring needs a minimum volume of closed records). Expected: you verify the volume threshold is met.
5. Re-run scoring setup. Expected: the model builds and scores new records.

## When to use

- Einstein scoring setup or refresh fails on data quality.
- Scores disappear after a data import.
- New orgs or new record types with thin history.

## When not to use

- Scores that seem wrong (model behavior, different question).
- Non-Einstein scoring tools.

## Tool compatibility

- Salesforce Einstein Lead Scoring and Opportunity Scoring.
- Field completeness reports.

## Variant phrasings

### insufficient data for scoring

Volume or completeness; check both.

### scoring disabled: data quality

Same fix; clean the inputs.

## Why it happens

Scoring models need representative, complete history. CRMs with sparse fields or young datasets cannot support the model, so it declines to run rather than produce garbage.

## Edge cases

- Skewed data (99 percent one outcome) also fails quality checks; the model needs both outcomes represented.
- Sandbox data is usually too thin; expect scoring to work only in production-like data.
- Field-level security hiding fields from the scoring user looks like missing data.

## Provenance

Resolved from the public thread: https://vectle.com/posts/pst_nczmjFE_raJABZTfi_NG4Q