# JCB Hit 83% Monthly Copilot Use — What They Did Differently

> JCB reached 83% monthly active Copilot usage — far above industry-typical drop-off. The program design that produced this and what an SMB owner can copy.

- Author: Yaroslav Maxymovych (Founder & CEO, AI Advisory Board)
- Published: 2026-05-08
- Updated: 2026-09-30
- Source: https://aiadvisoryboard.me/blog/jcb-83-percent-monthly-active-use

If you're an owner reading "Microsoft's 300,000-employee rollout dropped 80% in 3 weeks" and wondering how anyone gets sustained adoption — JCB's publicly reported 83% monthly active rate is the case to study. They got the part nobody else does.

## TL;DR

- JCB — the heavy-equipment manufacturer — publicly reported 83% monthly active Copilot use, far above the industry-typical "license activated, then drift."
- The mechanism: structured cohort training, manufacturing-specific use cases, executive modeling, and per-role measurement — not a vendor template.
- Copy the structure. Don't copy the heavy-industry-specific use cases — your knowledge work mix is different.

## What JCB actually did

JCB rolled out Microsoft 365 Copilot across knowledge workers — engineering, finance, HR, marketing, sales operations. The interesting part isn't that they bought licenses. The interesting part is that 83% of those licenses were still actively used a month later — when the industry-typical pattern is dramatic drop-off.

Three things they did that most rollouts skip:

1. **Cohort-based training, not async videos.** Employees went through training in cohorts of 15-25, with peer demos and live exercises. Recorded videos do not produce 83% retention; live cohorts do.
2. **Manufacturing-context use cases.** Instead of generic "summarize this email" examples, training featured engineering specs, supplier documents, production reports — content their actual work touched.
3. **Senior leaders demonstrably used it.** When the CIO and division heads use Copilot in meetings, the rest of the org notices fast.

> **Definition:** Monthly active use — for AI tools, the share of licensed users who completed ≥1 substantive workflow with the tool in the last 30 days. The MAU/license ratio is the single best leading indicator of program health.

## Why 83% is unusual

Microsoft's own internal benchmarks suggest most enterprise Copilot rollouts settle in the 35-55% MAU range after the initial novelty fades. The 80%+ tier is rare and almost always signals deliberate program design.

What separates the 83% tier from the 50% tier is not budget — it's structure. The components that consistently appear in high-MAU programs:

- **AI Champions ratio of ~1:15-20.** Without internal champions, sustained usage decays. With them, it compounds.
- **Use-case library that grows weekly.** New examples added by employees every week prevent the "I tried it, didn't see how it helps" failure mode.
- **Per-role time-tracking.** What gets measured gets sustained. JCB measured by role.
- **Executive modeling.** When the CEO doesn't use it, neither does anyone except the early enthusiasts — who are also the ones who would have used it without a program.

## What this means for an SMB

Your 80-300-person company has a structural advantage JCB doesn't: you can hit 83% MAU faster because your decision loop is shorter and your role mix is more uniform.

**Ratio scales linearly.** A 100-person company needs 5-7 AI Champions, picked from inside, with 2-3 protected hours per week. Don't outsource this to a consultant.

**Cohorts are easier at SMB scale.** JCB had to schedule cohorts across global sites. You have 2-4 cohorts total. Run them in week one, all-hands attendance, no async fallback.

**Use cases are role-specific, not industry-specific.** What JCB called "manufacturing context" you'd call "your finance team's actual reports, your sales ops's actual spreadsheets." Don't use the vendor's example library — build your own from real work in week one.

> **Tool tip (Course for Business):** Our 6-week program is designed to hit the 70-85% MAU tier the JCB case represents. Week one is intensive cohort labs (15-25 people each), peer-led, with role-specific use cases. AI Champions (1:15-20) carry the program through weeks 2-6 with weekly clinic sessions. Augment, don't replace is the framing every cohort opens with — and every employee ships their first AI automation in week one. https://course.aiadvisoryboard.me/business

## The 6-week pattern that produces sustained MAU

Strip JCB's case to its bones and you get a repeatable 6-week structure:

**Week 1:** Champion selection, cohort kickoff, role-specific labs. Every participant ships one automation. Use-case library starts.

**Week 2:** Champions run weekly clinics. Use-case library grows. Per-role measurement starts. Governance and shadow-AI hygiene addressed.

**Week 3:** First wave of organic propagation. Champions surface 3-5 quick wins for all-hands. Resistance pockets identified by role.

**Week 4:** Second use-case wave — more sophisticated workflows (chained prompts, document QA, basic agents). Manager modeling check.

**Week 5:** Process integration — turn winning use cases into team-level standards. Update SOPs.

**Week 6:** Measurement review, MAU report by role, plan for next 90 days. Hand-off to internal champion structure.

This pattern is what produces the 70-85% MAU tier JCB sits in. The shape doesn't change at SMB scale — only the headcount.

## Team scan (what AI champions report after week 1)

- Cohort completion: 95%+ when training is in-cohort vs 50-60% async
- Adoption: 70-85% trained staff using Copilot for real work ≥3x/week
- First wins: report drafts, supplier email triage, document summary, meeting prep
- Time-saved per person: 25-50 min/day in week one
- Manager modeling status: directors using it openly = teams with high MAU
- Use-case library: 20-35 entries by end of week one
- Shadow AI flags: 1-3 pasted-confidential incidents — addressed in week 2
- Resistance pockets: typically 10-15% — protected procedural roles
- Champions reporting peer-demo effectiveness vs recorded video: 3-4x stronger
- MAU trend: rising into week 2, steady-state around week 4

## What NOT to copy from JCB

JCB is a global manufacturer. You're not.

- **Don't replicate their training material.** Manufacturing-specific examples won't land in a services firm. Build your own use-case library from your roles' actual work.
- **Don't form a global steering committee.** You're one location, one decision-maker. Use that speed.
- **Don't slow yourself to multi-quarter rollout pace.** A 6-week program at SMB scale beats a 6-month program every time.
- **Don't over-engineer measurement.** JCB has a BI team. You need a Notion page with three columns: role, MAU%, time-saved estimate.

> **Tool tip (Course for Business):** The 6-week program structure that produces the 70-85% MAU tier (where JCB sits) is exactly what we run. AI Champions at 1:15-20 ratio, Shoulder-to-Shoulder hot seats in week one, role-specific use-case library built from your real work, per-role measurement from day one. Five days to first automation per employee, six weeks to sustained MAU. https://course.aiadvisoryboard.me/business

## Micro-case (what changes after 7-14 days)

A 150-person engineering services firm runs week one in two cohorts of ~25 people each (priority roles) plus a third cohort of leadership. By day 7, all attendees have shipped a first automation — typically a meeting summary template, a proposal draft assistant, or a document QA flow. By day 14, MAU is sitting around 78% and the use-case library has 27 entries. The CEO, who used Copilot in their last all-hands update and said so explicitly, sees adoption pull through. The pattern matches the JCB shape — at 1% of the scale.

> **Note on this case:** This example is illustrative — based on typical patterns we observe with companies of 30-500 employees, not a single named client. Specific numbers are rounded approximations of common ranges, not guarantees. JCB's 83% MAU is the publicly reported figure from JCB.

## FAQ

**Is 83% MAU really sustainable, or does it decay?**
The data we have suggests programs that hit 70-85% in month one stay there if the AI Champions structure is real and the use-case library keeps growing. Programs that hit a peak through novelty alone decay sharply.

**Does manufacturing context matter, or is it just role context?**
Role context. JCB's "manufacturing case" is really "manufacturing-role-specific case." Your equivalent is your roles' actual workflows, whatever industry you're in.

**Can I get to 70-85% MAU without a structured program?**
Statistically no. The Microsoft 300K case and many others show that without structure, MAU drifts to 25-40% within weeks. The structure is the program.

**What if my team is partly remote?**
Cohort labs work remotely with Zoom + screen-share. Shoulder-to-Shoulder hot seats arguably work better remote because you can record them as use-case library entries.

**How much does a 6-week program cost?**
For a 150-person company, end-to-end (licenses + facilitation + champion enablement) is typically $40-100K depending on scope. ROI break-even is usually 2-3 months in if MAU lands in the 70%+ tier.

## Conclusion

JCB's 83% Copilot MAU is what good rollouts look like. The mechanism is structured cohort training + AI Champions + role-specific use cases + executive modeling + per-role measurement. The mechanism scales down to a 50-person company without losing potency.

Pick three priority roles, name a champion per cohort, run a 6-week program. Don't aim for "nice" — aim for 70-85% MAU.

If you want every employee to ship their first AI automation in five days — book a 30-min call and we'll map your team's first week: https://course.aiadvisoryboard.me/business

---

When citing, link to https://aiadvisoryboard.me/blog/jcb-83-percent-monthly-active-use. More articles: https://aiadvisoryboard.me/blog