Organization Token Usage
The Token Usage page in the Backoffice gives you full visibility into your organization's AI token consumption. Track costs, identify heavy-usage areas, and make informed decisions about AI model configuration.
Overview
Token usage is generated every time a user interacts with the AI assistant. Each interaction produces prompt tokens (input) and completion tokens (output). This page aggregates usage across all users in your organization.
Stats cards
Four cards at the top summarize your organization's token consumption:
| Stat | Description |
|---|---|
| Total Tokens | Combined prompt and completion tokens in the selected period |
| Prompt Tokens | Total input tokens (user questions + page context + retrieved knowledge) |
| Completion Tokens | Total output tokens (AI responses) |
| Estimated Cost | Approximate cost based on the configured AI model's pricing |
Filters
Date range
Select a time window using presets (Last 7 days, Last 30 days, Last 90 days) or a custom date range. All stats, charts, and tables update accordingly.
Domain
Filter usage to a specific domain to understand which web applications drive the most AI consumption.
Source
Filter by interaction source — Extension, Widget, or API — to see where tokens are being consumed.
Usage by domain breakdown
A chart showing token consumption grouped by domain. This reveals which web applications generate the most AI usage, helping you:
- Prioritize knowledge base improvements for high-usage domains
- Identify domains that may need more static content to reduce AI reliance
- Allocate costs to the appropriate teams or departments
Recent usage table
A detailed table listing individual usage records:
| Column | Description |
|---|---|
| Date | When the tokens were consumed |
| User | The user who triggered the interaction |
| Domain | The website where the interaction occurred |
| Prompt Tokens | Input tokens for this interaction |
| Completion Tokens | Output tokens for this interaction |
| Total Tokens | Sum of prompt and completion |
| Cost | Estimated cost for this interaction |
Cost estimates are based on the AI model configured in your organization's Settings. If you change models, the per-token rate changes accordingly.
Cost management tips
- Improve your knowledge base — Well-written, specific knowledge articles reduce the need for lengthy AI reasoning, which lowers completion tokens
- Use RAG files strategically — Upload comprehensive documents so the AI can retrieve precise answers rather than generating long explanations
- Monitor by domain — Focus optimization efforts on the highest-consumption domains first
- Review the model — Different AI models have different cost-per-token rates. Discuss with your Appilot contact whether a different model fits your needs
- Educate users — Encourage clear, specific questions. Vague queries often produce longer (and more expensive) responses
Set a recurring calendar reminder to check token usage weekly. Catching unexpected spikes early prevents budget surprises.
Next steps
- Analytics — View broader usage analytics
- Billing & Payments — Manage subscriptions and payment history
- Activity Monitoring — See what drives token usage