Learn why your Pipedrive deals are won and lost

**Objective:** Analyze the deals won and lost between `{from-date}` and `{to-date}` (default: the last 90 days), optionally only in the `{pipeline-name}` pipeline, and explain why deals are won and lost.

**Step 1: Pull the closed deals**

* Call `pipedrive-get-deal` with `status` = won, and keep deals whose won-date falls in the period.
* Call `pipedrive-get-deal` with `status` = lost, `expected-close-date.from` and `expected-close-date.to` set to the period. Deals carry no lost date, so the expected close date stands in for it. Say so in the report.
* If `{pipeline-name}` is given, keep only rows whose pipeline matches it. Set `max-records` high enough (e.g. 500) to see the whole period.

**Step 2: Analyze**

* Win rate by count and by value (won / (won + lost)), per currency.
* Average won deal size, and average sales cycle in days (won-date minus created-date).
* Win rate and totals by owner.
* Lost deals by the stage they were lost in (the stage field on a lost deal).
* Lost reasons: group similar free-text reasons (e.g. "price", "too expensive") and count each group.

**Step 3: Report**

1. Headline metrics table.
2. By owner table.
3. Lost by stage table.
4. Lost reasons, grouped, with counts and value lost.
5. Three findings and one recommendation per finding, each backed by a number from the tables.

**Rules & Guidelines:**

* Read-only. Do not call `pipedrive-update-deal` or any other add or update tool.
* Never add values across currencies.
* Count lost deals with a blank lost-reason as "No reason recorded" and report how many there are.

**Validation:**

* If fewer than 10 deals closed in the period, give the numbers but say the sample is too small to draw conclusions.
* If `{from-date}` is after `{to-date}`, swap them and say so.