Review credit usage
The usage log is the record of what your account actually spent. It is the first place to look when a balance moved more than expected, and the only place that attributes spending to a specific site.
Where it is
My Account → POM AI Management on account.pom.es, below the credit and domain cards.
It lists the most recent operations first, 50 to a page, with pagination for older entries. It is per account, so it covers every authorized domain that used your key.
What each entry records
| Column | Meaning |
|---|---|
| Domain | The site that made the request |
| Action | The kind of operation performed |
| Tool | The POM AI tool group involved |
| Endpoint | The service endpoint that did the work |
| Credits used | The amount charged for that operation |
| Date | When it happened |
Only completed, charged operations appear. Estimates are not logged, refused requests are not logged, and batch expiry is not logged because it is not an operation.
That last point matters when diagnosing a balance drop: if the log does not explain it, check the batch expiry dates instead. See Credit expiration.
Reading it usefully
Group by domain first. On a shared subscription this is the question that matters — which site is spending. An unfamiliar domain in this column deserves immediate attention; see Understand domain authorization limits and, if you cannot explain it, Rotate a license key safely.
Then look at the shape of the run. A bulk operation appears as a burst of similar entries in quick succession — one per image, one per translated element, one per generated section. A single unexpectedly large entry is a different problem from fifty small ones.
Compare against what you confirmed. If you accepted an estimate of 400 credits, the entries around that time should add up to roughly that. A multi-stage run appears as many entries, not one.
Common patterns and what they mean
| What you see | Usual cause |
|---|---|
| Many small entries in a burst | A batch job: images, translations or sections |
| A long run of translation entries, more than the posts you selected | Attached taxonomy terms were translated too |
| Steady consumption from a domain you forgot about | An old site still holding the key |
| A large entry from an image tool | The higher-quality profile or 2K resolution |
| Consumption at hours nobody works | Worth investigating; check who has access to that site |
Using it to size a plan
The log is the only honest basis for choosing a plan. Rather than estimating what you might use:
- Work normally for one full billing period.
- Read the total consumed in the log for that period.
- Add the growth you actually expect, not the growth you hope for.
Then compare with Single, Agency and Unlimited. A plan chosen this way is usually smaller than the one chosen by intuition.
Using it to set domain limits
If several sites share the subscription, the log tells you what each one normally consumes. Set each domain's spending limit somewhat above its normal figure, in Manage AI Domains on the same screen.
A limit set from real data stops a runaway batch without blocking ordinary work. A limit guessed in advance tends to be either useless or an obstacle.
Keeping a routine
- Check the log at least once per billing period.
- Check it after any bulk operation, to confirm the run cost what you agreed.
- Check it immediately after an interrupted run, to see which stages completed and were charged.
- Check it before renewing or changing a plan.
When something looks wrong
Do not retry the operation to see if it happens again — that spends more credits and adds noise. Collect the details listed in Information needed to review a charge and raise it with support.