Sample data. No real people or teams on this page.
AI change management side quest · telemetry prototype

Every signal in the framework, on one page.

Every signal in the Success Metrics Framework, computed from a usage log, with the triggers wired to the playbook and a view per manager. All sample data until the Greenlight feed lands.

0%active usage rate, trailing week
0%users at sustained use
0interventions open
0%manager alerts answered
StatusPrototype, sample data. Waiting on the Greenlight feed.
ReadsSuccess Metrics Framework v1.0, Current State Assessment v1.0 (Kathryn, Aug 18)
Window
ByDallas Andrews
In one screen
  • What this is. The charter's "telemetry tracking built": all three levels of the framework from one usage log, four triggers wired to the playbook.
  • What it runs on. 40 synthetic users, 4 teams, 8 weeks. Swap in a Greenlight export with the same columns and it recomputes.
  • What it needs. Thresholds (what counts as sustained use), the Greenlight answers on the contract sheet, and a pilot team.
Level 2 · adoption signals

What counts as adoption is a threshold, and those are yours to set.

Every [X] and [Y] in the framework lives in one config file. Change them and the page recomputes.

uses in days
uses in days
% over days
no lift after days
no use after days
Showing the values in config/signals.json.

Uses per week, whole cohort

By team

TeamUsersActiveSustainedNever usedMedian days to first useMedian days to sustainedAgents usedSpreadWeeks

Spread is Team Adoption Variance: standard deviation of per-user uses over four weeks inside one team. Agents used is Feature Utilization Depth: share of available agents touched in four weeks.

Level 2 · risk signals to playbook

A trigger is a condition in the data; the playbook says what happens next.

Three of the four triggers fire from usage telemetry. Sustained workaround needs a source it doesn't have yet. Each firing carries the action, owner and response window, then whether a response was logged and whether usage lifted inside 14 days.

Owners and response windows are placeholders. The playbook sets the real ones.

UserTeamTriggerFiredPlaybook actionOwnerDueStatusLift in 14 d

What the engine never does

Send anything. It lists what fired, who owns it, and when it was due. The playbook decides what goes out and where.

Level 2 · manager engagement

Manager behavior is the largest controllable variable, so it gets its own view.

One card per manager: adoption, spread, alerts waiting on them, whether they've opened this page. In the pilot each manager sees only their own.

Levels 1 and 3 · process integrity and strategic impact

Process metrics say the pilot ran as designed; impact metrics wait for a baseline.

Level 1 · process integrity

From project records, not telemetry. Sample values.

Level 3 · strategic impact

Directional in the pilot, quantified at close. Empty rows wait for a baseline.

Data contract

Each signal traces to one field the feed has to carry.

Section B2 of the Current State Assessment from the engine's side: what each signal needs, and whether this page computes it, proxies it, or has no source. The contract sheet turns it into the questions for Jason Dowden.

Framework signalOn this feedWhat the feed needs to carry
Open the data contract sheet →
Open with the team

Seven decisions turn the prototype into the pilot's dashboard.

  • Sustained-use threshold. Try values in the panel above; the one that sticks goes into config.
  • Greenlight telemetry. Which signals the reporting can feed, at what latency, export or API. Joey has the contract sheet for Jason Dowden.
  • The pilot team. One or two groups with an agent rollout ahead of them. The AI champions know who.
  • Productivity baseline. Defined with the pilot team and Finance before launch, or the row stays empty.
  • Qualitative methods. Survey, interviews, or both at midpoint and close.
  • Where it lives. Hosted here for now. The Zuar path Matt Graves's trackers took is a question for Jason Dowden.
  • Vocabulary. No change-stage labels on this page, on purpose. The productized version needs its own words; Kathryn's call.
Next: Greenlight answers back, export in, first weekly review on this page.
AI change management side quest · telemetry

Every signal in the framework, on one page.

Usage log in, three levels of metrics out, four triggers wired to the playbook. Sample data until Greenlight lands.

What it measures

Process, behavior, impact.

Level 1 · process integrityReadiness assessments done, baselines documented, managers briefed, triggers monitored, milestones held. From project records.
Level 2 · adoption outcomesActive use, time to first and sustained use, agents used, spread inside a team, manager engagement. From the usage log.
Level 3 · strategic impactAdoption at close, time to sustained adoption, intervention effectiveness. Productivity and cost wait for a baseline.
Risk signals to playbook

A condition in the data, an owner and a clock.

Each firing records the response and whether usage lifted inside 14 days.

Manager view

One card per manager, their team only.

What turns it into the pilot dashboard

Thresholds, a feed, and one team.

Intradiem · adoption telemetry
1 / 5