Analytics
What this page is — a read-only dashboard over the documents you can access: speed, approval waits, signatures, quotation outcomes, what expires, what volume looks like, what AI costs — plus two AI cards that answer questions across the portfolio.
What it is for — so "contracts take too long" becomes a number attached to the step where the time goes.
The problem it solves — complaints that contracts take too long cannot be acted on, because nobody knows which step the time is lost in.
Route: /org/papers/analytics · Permissions: View cycle-time / win-rate / bottleneck analytics. (and .menu for
the sidebar); Use AI drafting/extraction/Q&A for documents. for the two AI cards and the Negotiation health strip.
1. What it is
Everything except the AI cards is computed from your documents, approvals, signature requests, quotations and AI run records. Nothing on the page changes data. Project isolation applies — two colleagues with different project access see different totals.
The date range (default: last 90 days) filters almost everything — but not everything:
| Card | Filtered by |
|---|---|
| Avg cycle time, Cycle time by type | Documents finalized in the range |
| Approval bottlenecks | State entries in the range |
| Signature turnaround | Requests in the range |
| Quotation win rate | Quotations sent in the range |
| Volume mix chips | Documents created in the range |
| AI usage | Runs in the range |
| Expiring (6 months), Expiry forecast | Ignores the range — always the next six months from today |
| Documents created per week | Ignores the range — always the last 12 weeks |
2. Why you would use it
- Slow steps get a name and a number. "Documents wait 71 hours in legal approval" moves a conversation that "legal is slow" does not.
- Renewals get a calendar. The expiry forecast shows the month carrying the most value at risk, not just the most documents.
- AI spend is itemised. Runs and credits per utility, so the credit balance has an explanation.
- Portfolio questions get answers. "Which contracts auto-renew next quarter?" without opening any of them.
3. How each number is computed — and where it misleads
| Figure | Computed as | Watch for |
|---|---|---|
| Avg cycle time | Days from creation to the first move into a state whose name contains "final"; median beside it | Finalize alone does not count. Sealing a PDF writes no such move, so on most organisations this reflects very few documents. Read the document count in the subtitle |
| Approval bottlenecks | Average hours per visit to each state whose name contains approv or review | A state named legal_signoff never appears, whatever its kind |
| Signature turnaround | Hours from sending to completion, completed requests only; decline share beside it | — |
| Quotation win rate | Accepted ÷ sent | — until a quotation was sent in range |
| Value at risk | The sum of the type's currency fields; falling back to fields named total, amount or contract_value | Documents with no value count in bars but add nothing to the amount |
| Entries in bottlenecks | Each time a document entered the state | High entries with a short average usually means rework loops |
4. Step by step
- Set From and To. Changing either reloads everything; there is no Apply button.
- Read the four KPI cards, then cycle time by type, then bottlenecks.
- Check the count behind every average before quoting it.
- Pick a project in the top bar, then use Portfolio Insights or Negotiation bottlenecks (AI).
- Refresh after work elsewhere — the page does not auto-refresh.
5. Field reference
KPI cards
| Card | Big number | Subtitle |
|---|---|---|
| Avg cycle time | Average days | Median and document count |
| Signature turnaround | Average hours | Requests sent and % declined |
| Quotation win rate | Percent accepted | Accepted of sent |
| Expiring (6 months) | Documents | Value at risk |
AI cards
| Card | Utility | Cost | Reads |
|---|---|---|---|
| Portfolio Insights | papers_portfolio_insights | 400 est. · 1,600 max | Each document's title, type, state and expiry, captured fields, and open obligations — a capped slice of large portfolios, and it says so |
| Negotiation bottlenecks (AI) | papers_negotiation_rollup | 500 est. · 2,000 max | Redlines, suggestions, rejections, rounds and cycle times — the 30 most contested documents per run |
Both cover the project selected in the top bar; with none selected they cover only organisation-level documents, which is usually nothing. Results are not stored — leaving the page discards them. Failed runs are not charged.
Other cards
| Card | Holds |
|---|---|
| Negotiation health | A signpost to documents, where per-document stall scores live |
| Volume mix | Chips by family and by state; weekly bars for 12 weeks |
| Expiry forecast | Month bars — count · value at risk |
| AI usage | Utility code, runs, credits — busiest first |
6. Worked example
A head of legal operations prepares a quarterly review for 1 April – 30 June.
Approval bottlenecks shows legal_approval — 48 entries · 71.5 h against 30 documents: documents bounce back and re-enter review, not just queue. That goes on slide one.
Avg cycle time reads 12.4 d · median 8.0 d · 3 docs. Three documents is not a trend: the organisation finalized 41 documents in the quarter, but its types have no state named "final…", so 38 were never counted. They leave the number off the slides, and rename the finished state to include "final" for next quarter.
Expiry forecast shows October with 5 documents and ₹12,00,000 at risk, and November with 9 documents and ₹80,000 — October matters more.
With Vendor Contracts selected they ask Portfolio Insights: "Which contracts auto-renew next quarter?" — three matches, each with the renewal date and why. 400 credits.
AI usage explains the credit drain: papers_doc_qa 84 runs; papers_type_from_document 6 runs but 7,200
credits.
7. The admin contract
| Must be true | Where | What breaks without it |
|---|---|---|
| The role holds View cycle-time / win-rate / bottleneck analytics. | Role editor | Could not load analytics |
| The role holds Use AI drafting/extraction/Q&A for documents. | Role editor | No AI cards |
| A project is selected for the AI cards | Top bar | They analyse organisation-level documents only |
| Lifecycles name finished states with "final" | Lifecycle | Cycle time misses those documents |
| Approval states are named with "approv" or "review" | Lifecycle | Bottlenecks miss them |
| Monetary values are currency fields | Fields | Value at risk under-reports |
| Documents carry validity expiry dates | Behaviour, documents | The expiry forecast is empty |
8. Downstream
Nothing changes when you read this page. To act on what it shows: renew from the Obligations page, fix slow approvals on the type's Lifecycle tab, and feed negotiation playbook fixes into the Clause library.
9. Don't confuse this with…
| Report builder | Choose your own dimensions and measures |
| Obligations | Individual dates to act on |
| Ask the repository | Answers from document wording, not fields |
10. Troubleshooting
| Symptom | Cause |
|---|---|
| Everything reads zero | The date range excludes your documents |
| Cycle time counts almost nothing | Finalize is not a "final" state move |
| An approval state is missing | Its name lacks "approv" and "review", or nothing entered it in range |
| Value at risk looks too low | Values are not in currency fields or the fallback names |
| The AI cards found nothing | No project selected in the top bar |
| The expiry forecast ignores my dates | It always looks six months ahead from today |