> ## Documentation Index
> Fetch the complete documentation index at: https://manual.botric.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# How credits work
> A complete reference for what Botric charges, when, and how the balance is spent.
Every action that costs compute costs credits. This page is the full picture; [Settings → Credits](/account/credits) is where you see your own numbers.
## The balance
Your organization has one balance, made of two pools:
| Pool | Refills | Expires |
| ------------------ | ---------------------------------- | --------------------------------------------- |
| **Plan credits** | At the start of each billing cycle | Yes — unused plan credits don't carry forward |
| **Add-on credits** | When you buy a pack | No — they survive cycle resets |
**Total available = plan credits remaining + all add-on packs remaining.**
Spending draws from plan credits first, then add-on packs. That ordering is deliberate: it means the credits you paid extra for are the ones that survive the reset.
## What costs what
| Action | Unit | Notes |
| -------------------- | --------------------- | -------------------------------------------------- |
| Site Audit | Per run | Independent of page count |
| GEO / AI Visibility | Per prompt | Billed on the full tracked cap plus add-on prompts |
| Chat Agent | Per 200 messages | Charged in batches |
| Topic discovery | 20 credits per run | Content Agent |
| Blog generation | 200 credits per blog | Refunded if generation fails |
| Prompt Research | 20 credits per prompt | Maximum 10 prompts per run |
| Creating a workspace | Site Audit + GEO cost | The first runs are triggered automatically |
Per-unit prices for Site Audit, GEO and chat messages are shown live on the Credits page, since they can change.
## The GEO billing rule
This is the one that surprises people.
A GEO run bills your plan's **full tracked-prompt cap plus every add-on prompt** — not the number of prompts currently in the list. Botric refills unlocked tracked slots to the cap on each run, so you're charged for the cap either way.
Worked example, on a plan with a cap of 20 tracked prompts and 5 add-on slots filled:
```
25 prompts × per-prompt GEO rate = credits per run
```
The **Run Analysis** button shows the exact figure before you commit.
### Tracked prompt caps by plan
| Plan | Tracked prompts per site |
| -------- | ------------------------ |
| Free | 5 |
| Bit | 10 |
| Byte | 20 |
| Infinity | 25 |
## Workspace creation
Creating a workspace triggers a Site Audit and a GEO run, so it's charged as the sum of both. Onboarding shows the required amount and warns existing subscribers when their balance is short; the server enforces the limit regardless.
## Refunds
A failed run is refunded automatically. You'll see the debit and the matching refund in your transaction history, and the item is marked as refunded in its own UI — a failed blog shows **Failed · refunded**.
## When you run out
Botric blocks the run before it starts and shows a dialog with:
* What the action **requires**
* What's **available**
* The **shortfall**, broken down by component
From there you can buy an add-on pack or upgrade. Nothing is charged partially — a run either has the credits it needs or doesn't start.
## Budgeting
A useful rhythm: one GEO run and one Site Audit per week, plus two or three blogs a month. Size your plan around that and buy add-on packs for one-off pushes rather than raising your tier permanently.
Ways to spend less:
* Run GEO weekly rather than daily — AI answers don't move fast enough to justify more.
* Use [Prompt Research](/discoverability/prompt-research) to trial prompts instead of burning a tracked slot and a full run on a guess.
* Re-run Site Audit after you've shipped a batch of fixes, not after each one.
Balance, history and top-ups.
Credit grants by tier.