Turn dormant seats into one shipped workflow (L2)

The steps

1. 01

   Pull the seat report and sort into four buckets

   Tool: Microsoft Entra ID

   Export 90 days of per-user activity. Sort every seat into: never logged in, logged in once and abandoned, occasional (1-3 sessions a week), and power user (daily). Report the four counts as percentages. Do not soften the never-logged-in number in the readout; that number is what buys you the mandate for the rest of this playbook. • Owner: Enablement, with IT for the export • Tool options: the assistant admin console (ChatGPT Team admin, Claude Console, Google Workspace admin, or Microsoft 365 admin center), plus Okta or Entra sign-in logs if SSO is enabled • Pitfall or what breaks: softening the never-logged-in number so leadership does not see the real scale of the problem. • Definition of done: a four-bucket table exists with a named list of power users and a named list of never-logged-in seats.

2. 02

   Interview the top five power users and steal their prompts

   Tool: Zoom

   Ask four questions each, in order: what do you use it for most, show me the exact prompt on your screen right now, what did you try that failed, what would break if we took it away. Get the literal prompt text, copy-pasted, not a description of it. The failure answers are more valuable than the successes because they tell you what to put in the rules section later. • Owner: Enablement • Tool options: 30-minute calls, recorded with Zoom, Granola, Fathom or Fireflies • Pitfall or what breaks: accepting a paraphrase of a prompt instead of the literal text, which loses the exact wording that made it work. • Definition of done: at least 15 verbatim prompts exist in a document, tagged by workflow and by who gave it to you.

3. 03

   Rank the extracted use cases and pick exactly three

   Tool: Google Sheets

   Cluster the 15 prompts into use cases. Score each on how many people could use it (1-5) and how much time it saves per use (1-5). Multiply. Take the top three. Three is the cap because each one needs a Project built, a walkthrough recorded and a named owner, and four is where quality starts slipping. • Owner: Enablement plus a revenue leader • Tool options: a scoring sheet • Pitfall or what breaks: building more than three use cases at once and letting quality slip on all of them. • Definition of done: three use cases are named with a projected user count for each.

4. 04

   Rebuild each one as a Project with context pre-loaded

   Tool: ChatGPT Projects

   For each of the three, create a Project containing: the power user's prompt rewritten as standing instructions rather than a one-off message; context files uploaded once so nobody has to paste them again, meaning your positioning one-pager, current pricing sheet, top three competitor battlecards, top 10 objections with approved answers, and your style rules; three good example outputs and two bad ones, each labeled with why; and a named owner for the Project plus a review date 90 days out, written into the description field. Then have a non-power-user run it cold, with no coaching, and watch. Every place they hesitate is a missing instruction. • Owner: Enablement, with the power user who contributed the prompt • Tool options: ChatGPT Projects, Claude Projects, Gemini Gems or Copilot Agents • Pitfall or what breaks: skipping the cold test with a non-power-user, so hidden gaps in the instructions never surface. • Definition of done: each Project has been run successfully by someone who did not help build it, without asking a question.

5. 05

   Ship a 3-minute walkthrough per Project, not a training session

   Tool: Loom

   Three minutes maximum, screen recorded, showing one real end-to-end use. Structure: here is the task, here is where you click, here is what you type, here is the output, here is what you should still check yourself. No slides, no theory, no scheduled training hour that half the team skips. Pin all three in the channel where the work happens, not in an LMS nobody opens. • Owner: Enablement • Tool options: Loom or Scribe, posted in Slack or Teams • Pitfall or what breaks: turning this into a 45-minute training webinar with a 30% attendance rate and no measurable change. • Definition of done: three videos are posted, each under 3:30, with view counts tracked.

6. 06

   Set a 30-day adoption target per Project and track weekly

   Tool: Google Sheets

   Target is 50% of the intended audience using the Project at least once a week by day 30. Publish the number every Friday by team, not by individual. Managers see their own team's line. If a team sits under 20% at day 14, the problem is the Project or the manager, and you go find out which by watching someone try to use it. • Owner: Enablement plus front-line managers • Tool options: the same sheet from step 1, refreshed weekly • Pitfall or what breaks: nobody owns each Project, so pricing changes in March and the Project confidently quotes February's numbers until August. • Definition of done: four weekly reports have been published and at least one Project has hit 50%.

7. 07

   Reclaim unused seats at day 60 and redeploy the money

   Tool: Microsoft Entra ID

   Any seat with zero logins in 60 days gets pulled. Send a one-line notice first with a reply-by date, so nobody is surprised. Then take the reclaimed spend and put it against the next playbook rather than back into general budget, and say so publicly. That is how the second project gets funded without a business case. • Owner: Enablement plus finance • Tool options: the admin console • Pitfall or what breaks: putting reclaimed spend back into general budget instead of naming the next playbook it funds. • Definition of done: seats are reclaimed, the dollar figure is documented, and it is assigned to a named next playbook.

8. 08

   Instrument the result

   Tool: Google Sheets

   Track weekly active users on the three named workflows, not total logins. Logins measure curiosity; repeat use on a named workflow measures adoption. • Where it breaks: you build eight Projects instead of three and none get maintained past month two; walkthroughs become a 45-minute training webinar with 30% attendance and no measurable change; or nobody owns each Project so stale pricing or facts persist for months unnoticed. • Visual guidance: a four-stage funnel: seats purchased, ever logged in, weekly active, weekly active on a named workflow. The drop between stages three and four is the entire argument for this playbook.

Tools in this playbook

• Microsoft Entra ID
• Zoom
• Google Sheets
• ChatGPT Projects
• Loom

Next playbooks

• Notion-Native Account Plan Agent, "Yes Chef" (L3)
• AI call highlight reels for coaching (L1)
• Onboarding ramp assistant built on your own content (L2)
• Certification with AI roleplay scoring (L4)