Copilot Adoption in Professional Services

Behavioral economist Dan Ariely put it perfectly: "Organizations think they experiment. They don't."

This is particularly true in professional services firms, where structural barriers make true experimentation difficult:

• Failure aversion: Psychological safety—the ability to take risks without fear of punishment—is often lowest exactly where innovation is most needed. (Amy Edmondson's research)
• Process rigidity: Service delivery systems are built for predictability and standardization, not exploration.
• Implementation theater: Many "pilot programs" are demonstrations designed to validate decisions already made, not genuine learning opportunities.

When you roll out a tool like Copilot, your teams aren’t equipped to try, learn, and adapt. They default to old workflows with new tools layered on top—gaining only a fraction of the potential value.

Good news: effective GenAI experimentation doesn’t require massive resources or disruption.
Think of it more like a science fair project:

• A clear question you're curious about
• A simple hypothesis to test
• A few hours of focused attention spread over a few weeks
• A straightforward way to document what you learn

The key is making experiments structured rather than hoping "trying things out" magically leads to improvement.

Research on organizational learning shows that effective Gen AI adoption depends on four neglected types of learning:

1. Experimental Learning: Hands-on testing (e.g., trying different usecases, once a week)
2. Reflexive Learning: Quick check-ins on what’s working (15-minute weekly debriefs)
3. Vicarious Learning: Learning from others (30-minute share sessions)
4. Contextual Learning: Staying current on GenAI advances (10 minutes weekly)

Even minimal commitment to these learning habits can dramatically accelerate your Copilot adoption results.

Even simple experiments fail if people aren’t committed.
That’s where our RISE framework comes in:

• Results: Clarify success ("We’ll know it worked if..."), Show career benefits ("This skill will help you...")
• Invest: Investments made in tools and time ("We spent $400-1000 per person  per year on licenses ...")
• Support: Provide minimal but real resources ("Training resources, office hours...")
• Expectations: Set clear parameters ("We need weekly notes on what you have done and learned...")

Without these conditions, even your most talented professionals will hesitate to invest time.

Here are a few ideas to start experimenting:

1.Start Small and Specific

• Pick a targeted workflow with:
  • A clear pain point (e.g., long email trails)
  • Copilot capabilities that seem well-suited
  • Success that would be obvious to everyone

2. Design a Simple 4-Week Experiment

• Week 1: Define the current process, document friction points, and create a simple hypothesis.
• Weeks 2–3: Test 2–3 approaches, track what works/doesn’t, and share findings.
• Week 4: Compare results, document findings, and make a recommendation: adopt, adapt, or abandon.

3. Create Simple Learning Habits

• Experimental Learning: Try two different approaches
• Reflexive Learning: End each session by noting surprises
• Vicarious Learning: Mid-experiment coffee chat
• Contextual Learning: Weekly 10-minute Youtube scan

4. Measure Just Enough

• Before: How long does the task take? How satisfied are people?
• After: Compare time and satisfaction again.
• Keep it simple.

5. Share What You Learn

• Reflect on findings weekly
• 30-minute demo to colleagues
• Store notes in a simple shared folder

A tax advisory practice wanted to explore how Copilot could help juniors get up to speed faster on new clients and matters. Instead of launching a huge initiative, they ran a simple experiment:

Team: Tech-savvy junior + Senior associate

Hypothesis: "Using Copilot could help juniors identify key tax issues faster and focus their research."

Results:

• Reduced research time from 5 hours to 90 minutes per client
• Developed a prompt structure that consistently surfaced relevant tax concerns
• ROI: projected 280 hours savings annually from a 10-hour experiment
• Bonus: The junior became proficient faster, improving overall team capacity

Here’s how you can make this happen:

• Week 1:
  • Identify 2–3 annoying tasks
  • Find willing participants
• Weeks 2–5:
  • Let teams run experiments with minimal interference
  • Request brief weekly updates
• Week 6:
  • Collect one-page summaries
  • Hold a 60-minute findings session
  • Identify approaches to scale
• Weeks 7–8:
  • Document winning approaches
  • Select next tasks to experiment with
  • Begin integrating successful changes into operations

Total organizational overhead: Approximately 10 hours per participant.

Don't overcomplicate it. Start simple. Start now.

Unlock practical results - one small experiment at a time.