When we launched our first AI training program, we went with self-paced online modules. The logic seemed sound: people could learn at their own pace, on their own schedule, without coordination overhead.

Completion rates were abysmal. And the people who did complete showed minimal behaviour change.

We rebuilt the program around cohorts—groups of learners progressing together through structured experiences. Completion jumped to over 80%. More importantly, participants actually started using AI in their work.

The difference wasn’t the content. It was the learning model.

Why Cohorts Outperform Self-Paced

The research on adult learning supports what we experienced. Cohort-based learning outperforms self-paced for several reasons:

Accountability

When you’re learning alone, no one notices if you skip a module or drop out entirely. When you’re learning with others, there’s social accountability:

• Peers expect you to participate
• Facilitators track attendance
• Group activities require contribution
• Falling behind is visible

This accountability isn’t punitive—it’s supportive. It helps people prioritise learning when competing demands pull at attention.

Peer Learning

Much learning happens through peer interaction:

• Seeing how others approach problems
• Hearing diverse perspectives
• Discussing applications to different contexts
• Normalising struggles everyone experiences

Self-paced learning lacks this peer dimension. Cohorts build it in.

Real-Time Support

When learners get stuck, immediate help matters:

• Questions answered when they arise
• Misconceptions corrected before they solidify
• Encouragement when frustration builds
• Troubleshooting in real time

Asynchronous support can’t match real-time responsiveness.

Motivation Through Community

Learning is more enjoyable with others:

• Shared experiences create connection
• Humour and energy sustain engagement
• Community makes learning feel worthwhile
• Relationships extend beyond the program

Self-paced often feels isolating. Cohorts feel communal.

Structured Pace

Self-paced means self-scheduled, which often means never:

• Other priorities crowd out learning
• “I’ll do it later” becomes “I never did it”
• Momentum is hard to generate alone

Cohorts create external structure that ensures progress.

Designing Effective Cohort Programs

Cohort programs require thoughtful design:

Cohort Size

Balance intimacy with efficiency:

• 8-15 participants: Ideal for most learning. Small enough for everyone to contribute, large enough for diverse perspectives.
• 16-25 participants: Works for programs with breakout groups. Requires more facilitation skill.
• 25+ participants: Moves toward broadcast. Can work with strong breakout structures but loses intimacy.

Smaller is usually better for behaviour change.

Cohort Composition

Who learns together matters:

Homogeneous cohorts (same function/level):

• Easier to tailor content
• More directly relevant examples
• Less intimidating for less-senior participants
• May lack diverse perspectives

Heterogeneous cohorts (mixed function/level):

• Cross-functional learning and networking
• Diverse perspectives and applications
• Can be intimidating for some
• Harder to make content universally relevant

Choose based on learning objectives and organisational culture.

Program Duration

How long should cohorts run?

• Sprint (1-2 weeks): Intensive immersion, high energy, limited depth
• Standard (4-8 weeks): Balance of depth and sustainability, time for practice between sessions
• Extended (3-6 months): Deep development, significant practice time, community building

AI literacy programs often work well in 4-6 week formats with weekly sessions.

Session Cadence

How frequently should cohorts meet?

• Daily: Only for intensive sprints. Exhausting to sustain.
• 2-3 times weekly: Good for intensive programs with sufficient time.
• Weekly: Most common. Allows time for application between sessions.
• Bi-weekly: Works for lighter programs or advanced topics.

Weekly sessions of 60-90 minutes work well for most AI training cohorts.

Content Structure

Each cohort session needs structure:

Opening (10-15%): Connect to previous session, preview current session, warm-up activity.

New content (30-40%): Introduce new concepts, demonstrate tools, share examples.

Practice (30-40%): Hands-on application, exercises, experimentation with support.

Reflection (10-15%): Discuss learnings, plan application, preview next session.

This structure balances information transfer with active learning.

Between-Session Activities

Learning happens between sessions too:

• Practice exercises to apply session content
• Reflection prompts to consolidate learning
• Peer check-ins to maintain connection
• Reading or video content to prepare for next session

Between-session work should be realistic—30-60 minutes is achievable for most working professionals.

Facilitation Excellence

Cohort programs depend on strong facilitation:

Facilitator Capabilities

Effective cohort facilitators need:

• Subject matter competence (AI tools and applications)
• Facilitation skills (managing group dynamics, timing, energy)
• Adaptive capacity (adjusting to learner needs in real time)
• Supportive presence (creating psychological safety)

AI expertise isn’t enough. Facilitation skill matters as much.

Live vs. Recorded Content

Maximise live time for what requires it:

Live: Discussion, Q&A, practice with support, troubleshooting, peer interaction.

Pre-recorded: Foundational concepts, demonstrations that don’t change, content people can consume at their pace.

Don’t use expensive live time for content better delivered asynchronously.

Managing Energy and Engagement

Keep cohorts energised:

• Vary activities every 10-15 minutes
• Include movement and breaks
• Balance individual work with group interaction
• Use humour appropriately
• Address energy dips directly

Engagement doesn’t happen automatically. Facilitators create it.

Handling Diverse Levels

Cohorts often include learners at different levels:

• Provide extension activities for advanced learners
• Offer additional support for struggling learners
• Create peer mentoring within the cohort
• Differentiate expectations where appropriate

Mixed levels can be strength if managed well.

Building Cohort Community

Strong cohorts become communities:

Starting Strong

First impressions matter:

• Meaningful introductions that create connection
• Clear norms and expectations
• Early collaborative activities
• Quick wins to build confidence

Invest in community building upfront.

Ongoing Connection

Maintain community throughout:

• Shared channels (Slack, Teams) for ongoing conversation
• Peer pairing for mutual support
• Celebration of progress and wins
• Addressing struggles collectively

Community sustains motivation when individual effort flags.

Continuing Beyond the Program

The best cohorts continue after formal programs end:

• Alumni communities for ongoing connection
• Periodic reunions or follow-up sessions
• Peer support for continued learning
• Shared resources and experiences

Community value extends beyond program duration.

Logistics and Coordination

Cohort programs require coordination:

Scheduling

Finding times that work:

• Survey participants for availability
• Consider time zones for distributed cohorts
• Build schedule before recruitment
• Communicate expectations clearly

Scheduling is often the biggest logistical challenge.

Technology

Platform requirements:

• Video conferencing with breakout capability
• Chat or messaging for ongoing communication
• File sharing for resources
• Practice environments for hands-on learning

Ensure participants can access required technology.

Communication

Keep cohorts informed:

• Clear pre-program communication about expectations
• Reminders before each session
• Follow-up after sessions with resources and assignments
• Ongoing channel management

Communication keeps cohorts organised and engaged.

Measuring Cohort Effectiveness

Track what matters:

Engagement Metrics

• Attendance rates
• Participation levels (chat, discussion, activities)
• Between-session activity completion
• Community channel activity

Learning Metrics

• Assessment scores (if applicable)
• Demonstrated skill improvement
• Confidence ratings
• Knowledge retention

Application Metrics

• AI tool adoption post-program
• Behaviour change in actual work
• Manager observations
• Self-reported application

ROI Metrics

• Productivity improvements
• Quality improvements
• Business outcomes
• Comparison to self-paced alternatives

Measure across levels to understand full impact.

When Self-Paced Works Better

Cohorts aren’t always the answer. Self-paced may work better for:

• Highly diverse schedules that can’t be coordinated
• Foundational content everyone needs regardless of role
• Reference material people access on demand
• Very large populations where cohorts aren’t feasible
• Learners with strong self-direction

Consider blended approaches: self-paced foundations plus cohort application.

Getting Started

To launch cohort-based AI training:

1. Define learning objectives and target population
2. Design cohort structure (size, duration, cadence)
3. Develop or adapt content for cohort delivery
4. Recruit and prepare facilitators
5. Coordinate logistics and scheduling
6. Recruit and enrol first cohorts
7. Deliver, learn, and improve

Start with a pilot cohort. Learn from experience. Scale what works.

Cohort-based learning requires more coordination than self-paced alternatives. But the difference in outcomes—completion, learning, behaviour change—makes the investment worthwhile.

People learn better together. Design your programs accordingly.