> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.pivotal.app/llms.txt.
> For full documentation content, see https://docs.pivotal.app/llms-full.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.pivotal.app/_mcp/server.

# Forecast launches

Pi forecasts a launch date for every onboarding with a target. The estimate appears next to the target on the onboarding detail page as a **Forecast** chip. Green means on track. Amber means slipping inside one phase. Red means slipping by more than a phase.

## How Pi builds the forecast

The model reads three inputs per onboarding:

* **Task pace.** Average days between task completions on this onboarding.
* **Phase velocity.** Median days other customers spent in each phase, weighted toward customers on the same plan.
* **Slippage history.** This customer's own pattern of finishing phases early or late.

Pi runs the forecast nightly at 2 a.m. UTC and on every phase change. A new onboarding gets a forecast after the first task closes; before then, the chip shows **No forecast yet**.

## The Forecast chip

| Field                | What it shows                | Notes                               |
| -------------------- | ---------------------------- | ----------------------------------- |
| Estimated date       | Pi's predicted actual launch | Updates nightly and on phase change |
| Slippage             | Days off target, signed      | Positive = late, negative = early   |
| Confidence           | Low / Medium / High          | High needs at least 4 closed tasks  |
| Last updated         | Timestamp of last run        | Hover the chip to see it            |
| Contributing factors | Top 3 inputs ranked          | Click the chip to expand            |

Click the chip to open the explanation drawer. Pi lists the three factors that moved the estimate the most since the last run, with arrows showing whether each pushed the date earlier or later.

## Reading the colors

* **Green.** Forecast within 3 days of target on either side.
* **Amber.** Forecast slips 4 to 14 days past target.
* **Red.** Forecast slips more than 14 days past target, or one full phase median.

Workspace admins can change the thresholds in **Admin > Workspace settings > Forecast**.

## Gotcha: a fresh phase reset can swing the forecast

Resetting an onboarding to an earlier phase (for example, dropping from UAT back to Integration) wipes the recent task pace, so Pi reverts to a low-confidence estimate built from the workspace median. The chip flags **Confidence: Low** until the customer closes four more tasks. Don't act on the dollar value of the slippage until confidence climbs back.

## Gotcha: target-less onboardings

An onboarding without a `target_launch_date` shows the estimated date but no slippage or color. Set the target in the onboarding header and the chip fills in on the next run. See [Target launch dates](/product/onboardings/target-launch-dates).

## Asking Pi about a forecast

The composer answers prompts like "why is Acme's forecast slipping?" or "which onboardings are red right now?". Pi pulls from the same forecast table the chip uses, so the numbers match.

## Related

* [What's at risk](/product/ask-pi/whats-at-risk)
* [Target launch dates](/product/onboardings/target-launch-dates)
* [The phase model](/product/onboardings/the-phase-model)

Email **[help@pivotal.app](mailto:help@pivotal.app)** with a screenshot of where you got stuck and the customer or onboarding id from the URL.