This lesson is being recorded

Every video of this course is being uploaded to my YouTube channel. Subscribe so you do not miss it.

Bruno AshfordBruno Ashford @brunoashfordSubscribe

Module 13 · Lesson 13.5 · 10 min

Measuring Adoption

Track usage, results and resistance with a simple weekly scorecard.

Module 13
Rolling Out to Your Team

From one owner's laptop to a company system

Bruno Ashford
Bruno Ashford · Module 1301 / 12
Why rollouts fail
Sending your team a link is not a rollout. It is a lottery.
Bruno Ashford
Bruno Ashford · Module 1302 / 12
The rollout

Three waves in 30 days

  1. 1Pilot groupTwo or three credible people
  2. 2First departmentProven skills only
  3. 3Whole teamWhen no safety issue is open
Bruno Ashford · Module 1303 / 12
Lesson 13.1

From owner to team

Share what is proven, not everything you built.

  • Company knowledge in shared CLAUDE.md
  • Personal preferences stay personal
  • Policy before tools
  • Credibility beats technical skill
Bruno Ashford · Module 1304 / 12
Lesson 13.2

What a plugin bundles

One folder, one install, one version for everyone.

  • Skills in skills/
  • Subagents in agents/
  • Hooks in hooks/hooks.json
  • Connectors in .mcp.json
Bruno Ashford · Module 1305 / 12
Lesson 13.2

The plugin commands

StepCommand
Check itclaude plugin validate ./acme-ops
Test itclaude --plugin-dir ./acme-ops
Add marketplaceclaude plugin marketplace add org/repo
Installclaude plugin install acme-ops@acme-tools
Manage/plugin
Bruno Ashford · Module 1306 / 12
Lesson 13.3

Four roles

  1. BBuilderCreates and edits skills
  2. AApproverSigns off changes and sensitive output
  3. UUserRuns skills daily
  4. MMaintainerKeeps versions and CLAUDE.md current
Bruno Ashford · Module 1307 / 12
The RACI rule
Two people accountable means nobody is accountable.
Bruno Ashford
Bruno Ashford · Module 1308 / 12
Lesson 13.4

30 days, 15 minutes a day

WeekFocus
1Install and first real win
2Two tasks from their own week
3Verification habits and pairing
4Contribute one improvement back
Bruno Ashford · Module 1309 / 12
Lesson 13.5: starting targets, adjust to your business

Adoption by day 30

80%use a skill 3+ times a week
3.5 / 5average confidence
< 15%rework rate
Bruno Ashford · Module 1310 / 12
Lesson 13.6

Change management

Fear goes underground if you ignore it.

  • Say clearly what AI is for
  • Link tools to real frustrations
  • Make champions visible
  • Ask resisters what worries them
Bruno Ashford · Module 1311 / 12
Recap
Package what works, name who owns it, train on real work and measure every week.
Bruno Ashford
Bruno Ashford · Module 1312 / 12
01 / 12

Use ← → to move. The marker highlights text and the pencil draws on the slide; navigation locks while you annotate. Undo removes the last mark, Esc exits.

1Track usage per person

Record how many times per week each person runs a company skill or routine. Self-reporting in a two-minute Friday form is fine for a small team. If you use a Team or Enterprise plan, check what usage information your admin console provides.

2Track results per person

Record tasks completed with AI and estimated hours saved, using the same method as Module 12. Usage without results means the skill does not fit the work, which is a Builder problem, not a people problem.

3Track friction and sentiment

Ask two questions every Friday: what slowed you down this week, and how confident do you feel with the tools from 1 to 5. Low confidence with high usage means someone needs pairing. Low usage with high confidence means the tools do not match their tasks.

Prompt for Claude
Here is this week's adoption scorecard (CSV below). Summarize it for me in five lines: overall adoption rate, total hours saved, the two people who need support and why, the most common friction point, and one change to the plugin or training that would help most next week.
4Set simple targets

By day 30: at least 80 percent of the rollout group using a skill three or more times a week, average confidence of 3.5 or higher, and rework under 15 percent. These are starting targets, adjust them to your business after the first month.

5Act on the data weekly

Every Monday, pick one action from the scorecard: pair someone, fix a skill, or remove a friction point. Adoption improves through small weekly fixes, not through reminders to "please use the AI".

You cannot manage what you only feel. "People seem to like it" is not a metric. Some of your team will use the tools daily, some weekly, and some will quietly never open them. You need to know which, kindly and early.

Adoption has three layers: are people using it, is it producing results, and what is blocking the rest. A one-page weekly scorecard covers all three.

Key points

  • Measure usage, results and friction every week.
  • Low usage is usually a fit problem, not a people problem.
  • Confidence scores tell you who needs pairing.
  • One concrete action per week beats constant reminders.

Avoid

  • Using the scorecard to shame people instead of to help them.
  • Tracking usage without results.
  • Setting targets so high the team games the numbers.
  • Collecting data and never acting on it.
Real-world example: an e-commerce store

An online store rolled its customer service skills out to six agents. The scorecard showed two agents with near zero usage but high confidence. A quick chat revealed they handled wholesale orders, which the skills did not cover. The Builder added a wholesale reply skill, and both agents became the heaviest users within two weeks.

Bruno Ashford Subscribe