Straight-through processingSTP
The share of users decided automatically, with no human involved.
If 90 out of every 100 verifications get an automatic approve or decline, and 10 go to a person, the straight-through processing rate is 90 percent.
The proportion of cases fully adjudicated by automated policy without analyst intervention, tracked as a rate ops teams use to size the review queue and staff accordingly. Adjudicated means a final decision was reached, and analyst intervention means a person looked at the case.
Straight-through processing (STP) is calculated by dividing the cases decided automatically by all cases in the period. Several things move the rate:
- Model and rule quality, since more accurate decisions leave fewer cases uncertain.
- Thresholds, since a wider review band sends more cases to people.
- Document and data coverage, since unsupported documents often end up in review.
- Capture quality, since poor images lead to more uncertain results.
Teams use the rate in a few ways:
- Capacity planning, to estimate how many analysts the review queue needs.
- Cost tracking, to see what each verification costs once review time is included.
- Performance tuning, to find causes of cases that fall out of automation.
A high rate is not automatically a good one. If it rises because thresholds were loosened, fraud can slip through, so teams read it together with fraud and pass rates.
Straight-through processing sits next to terms that are easy to mix up:
- Manual review is the human work that handles the cases outside the automated share.
- Pass rate is the share of genuine users approved, which is a different measure.
- Decisioning engine is the component that produces the automated decisions.
- Term
- Straight-through processing (STP)
- Abbreviation
- STP
- Category
- Platform
- Slug
straight-through-processing