Economic health
This workflow is not healthy under the current plan assumptions.
The reviewed workflow produced 18 useful answers for Customer A and cost
an estimated $41.20 in provider spend. The customer pays $49/month on the
active plan snapshot, leaving $7.80 before product, hosting, support, and
payment costs.
Plan revenue
$49.00
Workflow cost
$41.20
Retry spend
27%
Margin left
$7.80
Evidence table
What the existing export can already prove.
Signal
Observed
Source
Decision impact
Workflow provider cost
$41.20
cost field
margin risk
Usable outcomes
18 answers
completed status
value delivered
Retried calls
11 retries
retry count
27% spend
Failed runs
6 failed
failed status
no user value
Plan snapshot
price_starter_v1
active price id
rate locked
Data gaps
What would make the next report more reliable.
Missing outcome reason
Completed and failed statuses exist, but the export does not explain whether failures came from retrieval, context length, model refusal, timeout, or parsing.
Missing context size
Provider cost is available, but context-window tokens and retrieved-file size are not separated. That limits the confidence of a context-cap recommendation.
Partial customer segmentation
Customer ID exists, but plan tier and account type are not complete for every row. Segment-level margin should not be automated yet.
Recommended decision
Add a soft cap now. Do not migrate to a gateway yet.
The evidence is strong enough to justify a customer-level warning and a
workflow-level review, but not strong enough to enforce hard blocking or
reprice all plans from one snapshot.
Do now
Warn Customer A before the workflow consumes 60% of monthly plan revenue.
Inspect next
Review the retry path and identify why 27% of provider spend repeats work.
Instrument only if usage continues
Add outcome_reason, retry_reason, and context_size to future events.
