Ask Pi

What's at risk

How Pi ranks customers by risk, which signals matter, and how to act on the list.
View as Markdown

Pi keeps a live ranked list of customers and onboardings most likely to slip, churn, or stall. The ranking feeds the Risk lane widget on the Workbench and answers prompts like “what should I look at today?”. Each entry has a reason attached, so you can decide whether to act or dismiss.

The signals Pi watches

Pi watches a handful of signals and weights them. No single signal flips a customer at-risk on its own; Pi combines them.

SignalWhat it measures
Phase stallDays in current phase versus the workspace median for that phase
Task slipOpen customer-facing tasks past their due date
Reply gapDays since the primary contact last replied
Forecast slipDays the Pi forecast has drifted past the target launch date
Health dropDrop in the customer’s health score over the last 14 days
Sentiment shiftA run of negative comments in the recent thread

A customer at the top of the lane usually has two or three of these stacked together. Click any entry to see the contributing signals listed in order of weight.

How the list orders

Pi re-ranks the list when:

  • A phase moves
  • A customer-facing task is marked done or slips past due
  • A new comment lands on the customer thread
  • The nightly forecast job runs

The Workbench widget shows the top 10 by default. The full list lives at Workbench > Risk lane > See all.

Acting on an entry

Open an entry and you have three choices.

  1. Address it. Open the customer, take the action Pi recommends (reach out, reassign, move the phase), and the entry drops on the next re-rank.
  2. Dismiss it. Click the dismiss icon on the entry. Pi removes it from the lane and notes the dismissal. The same combination of signals won’t re-flag this customer for 7 days unless a new signal joins.
  3. Mute the customer. From the entry menu, choose Mute risk flags for 30 days. Use this when you know a customer is going through a quiet phase by design (a Q4 holiday freeze, for example).

At-risk customer status vs the risk lane

These are two different things. A customer’s status field (one of active, at_risk, churned, archived) is a manual classification you set on the customer record. The risk lane is Pi’s automatic ranking. A customer can be on the risk lane while still flagged active, and vice versa. See At-risk and churned for the status field.

Gotcha: dismissed flags that come back

A dismissed flag returns when a new signal joins the stack. If you dismissed an entry that was flagged for phase stall plus reply gap, and a forecast slip appears the next day, Pi re-flags it with the updated reason list.

Email help@pivotal.app with a screenshot of where you got stuck and the customer or onboarding id from the URL.