# Wallet & pay-as-you-go

> https://docs.errora.org/en/billing/wallet

Every organization has a **wallet in Toman** and pays per unit of what it actually uses.
There are no plans, seats or tiers.

## Price list

Rates per unit:

| Service | Unit | Rate (Toman) |
| --- | --- | --- |
| Error / Log | per event or log record | 10 |
| Tracing | per span | 0.1 |
| Session Replay | per recorded session | 150 |
| Cron Monitoring | per check-in | 1 |
| Uptime Monitoring | per check | 1 |
| Profiling | per profile | 20 |
| Metrics | per metric bucket | 0.5 |
| AI Fix | per auto-fix run | 20,000 |

Live list: `GET /api/v1/pricing`. The pricing page's sliders let you estimate a monthly bill
from your expected volumes.

## Signup credit and the credit line

- New organizations start with **100,000 Toman** of credit.
- Services keep running below zero down to a credit line, so a traffic spike never silences
  your monitoring in the middle of an incident.
- Past that line, ingestion is **paused**: the endpoint refuses new data until the wallet is
  topped up. Nothing already stored is deleted.

## How metering works

Usage is counted as data arrives and the wallet is debited a few seconds later, in the
background. That is why the balance moves in small steps rather than with every single
event — and why metering never slows down or rejects what your app is sending.

## The usage page

**Usage** shows:

- Current **balance** and whether the organization is suspended
- **Cost and units per service** for the period
- **Daily and monthly breakdowns**
- The **wallet ledger** (every debit and credit) and payment history

```text
GET /api/v1/pricing
GET /api/v1/organizations/{org_id}/usage
GET /api/v1/organizations/{org_id}/wallet
GET /api/v1/organizations/{org_id}/wallet/transactions
```

Reading billing requires the **`org:billing`** capability. The balance is also available to
AI agents over [MCP](/en/products/mcp) via `get_wallet_balance`.

## Balance alerts

Set a threshold and a channel, and `wallet.balance_low` fires when the balance crosses it —
email, SMS, Mattermost, Telegram, Bale or a webhook.

```text
GET/POST/PATCH/DELETE /api/v1/organizations/{org_id}/balance-alerts
```

Set at least one. Running out of credit stops ingestion, and the first symptom is usually
"why is the dashboard quiet?".

## Controlling spend

1. **Sample traces and profiles** — they are the highest-volume services.
2. **Lower `replaysSessionSampleRate`** — replays are the most expensive unit.
3. **Watch tag cardinality on metrics** — every tag combination is its own bucket.
4. **Use the project sample rate** for a firehose you can't fix at the source.
5. **Prefer `issue.created` over `event.received`** for alerts — not a billing item, but it
   is the same instinct.

## Troubleshooting

| Symptom | Cause / fix |
| --- | --- |
| Ingestion stopped | Balance is past the credit line — [top up](/en/billing/payments) |
| Balance moves in bursts | Settlement runs on an interval, not per event |
| Usage looks higher than expected | Check which service dominates on the usage page — usually tracing spans or replays |
| Numbers differ from your own counters | Errora meters what *arrives*, after SDK sampling and before the project sample rate |