Analytics overview
Where to find storage and token analytics in AskPhi, and which ones you can see.
AskPhi tracks two kinds of usage:
- Storage — how much disk your CSV files take up (in KB / MB / GB).
- Tokens — how much AI usage your messages have consumed.
You'll find both kinds in the Management area at /manage, under
Analytics.
Personal vs organization
Every signed-in user has access to personal analytics:
- My storage — your own files.
- My tokens — your own AI usage.
If you're an organization admin or owner, you also see organization-wide analytics:
- Organization storage — every member's files.
- Organization tokens — every member's AI usage.
For the exact capability mapping, see Permissions & roles.
How analytics work
Each analytics page has two parts:
- Summary cards — totals for the selected scope and date range.
- Daily trend chart — a Plotly chart showing the past N days (default 30).
A few common controls on every page:
- Scope (only when in an organization) — choose between
active_org(just your current org) andall_orgs(across all your memberships). - From / To dates — restrict the summary to a date range.
- Days — number of days for the trend chart.
tipSet From and To when you want to compare a specific period (e.g. last quarter). Leave them empty to see all-time totals.
Where each analytics page lives
| Page | URL | Audience |
|---|---|---|
| My storage | /manage/analytics/storage | Anyone signed in. |
| My tokens | /manage/analytics/tokens | Anyone signed in. |
| Organization storage | /manage/analytics/org-storage | Admins / owners only. |
| Organization tokens | /manage/analytics/org-tokens | Admins / owners only. |
How AskPhi counts storage
- Each file's size is measured in bytes when uploaded and converted to KB / MB / GB for display.
- Adding rows manually or appending a CSV increases the count.
- Deleting a file frees its bytes immediately.
How AskPhi counts tokens
- Each AI message consumes input tokens (your question + context) and produces output tokens (the reply).
- Charts in replies count too — they're built from the same tokens that produced the text.
- Different models charge different per-token rates; analytics reports raw counts without a price.