The Power of AI Practice and the Might of Human Coaching
AI role-plays are transforming how sellers practice, but technology alone is not enough. Explore how AI-powered repetition, reflective human coaching, and structured feedback can work together to improve sales performance.
The Power of AI Practice and the Might of Human Coaching
Why the Future of Sales Development Requires Both Human Behavior and Artificial Intelligence
What improves your chances of learning to play the piano?
Practicing alone?
Or practicing on your own and working with an instructor?
We know the answer almost instinctively.
Imagine taking one piano lesson every few weeks but never touching the keys between lessons. Your instructor could be exceptional, but your development would still be limited by a lack of repetition.
Now imagine the opposite.
You practice alone every day, but no experienced instructor ever observes your technique. No one notices that your posture is wrong, your timing is inconsistent, or that you’ve developed habits that may eventually limit your performance.
You could become very consistent at practicing something incorrectly.
Developing sales skills works much the same way.
After 28 years of selling, leading sellers, and supporting sales organizations, I’ve become convinced that one of the biggest opportunities in sales development is not choosing between technology and human interaction.
It is learning how to combine them.
We need the Power of AI Practice and the Might of Human Coaching.
AI gives us something sales organizations have historically struggled to provide at scale:
Unlimited opportunities to practice before performing.
Human coaching provides something technology should not be expected to fully replace:
Context, judgment, empathy, reflection, and accountability.
The future of sales development should bring these strengths together.
Selling to a Webcam Isn’t Selling
For years, one common form of sales certification has involved asking sellers to record themselves delivering a pitch into a webcam.
The seller prepares.
They press Record.
They deliver the messaging.
Someone reviews the recording against a scorecard.
Did they deliver the value proposition?
Did they mention the required differentiators?
Did they ask for the next step?
Check.
Check.
Check.
There is certainly value in confirming that someone understands a message and can articulate it clearly.
But we should also acknowledge the limitation.
Selling to a webcam isn’t selling.
A webcam doesn’t interrupt you.
It doesn’t challenge your assumptions.
It doesn’t give you an unexpected answer.
It doesn’t say:
“That’s not really our problem.”
“We already have a solution for that.”
“Your competitor told me something completely different.”
“We don’t have budget.”
“Send me some information.”
And a webcam certainly doesn’t make an offhand comment that should cause you to completely abandon your next planned question because you just uncovered something far more important.
Real selling requires interpretation.
Listening.
Thinking.
Adapting.
Responding.
Curiosity.
Judgment.
A recorded pitch can test whether a seller knows what to say.
An interactive conversation tests whether the seller knows what to do with what they hear.
That distinction is enormous.
A recorded pitch tests whether you can recite.
A conversation tests whether you can think.
Salespeople Need a Place to Practice Before They Perform
Think about almost any high-performance profession.
Musicians rehearse before concerts.
Athletes practice before games.
Actors rehearse before opening night.
Pilots spend time in simulators before certain situations occur in the air.
Yet sales professionals are often introduced to a new skill, methodology, product, competitive message, or objection-handling technique and then sent directly into conversations with real prospects.
In other words:
We sometimes practice on the customer.
That creates an obvious problem.
Every customer conversation has consequences.
A poorly handled objection can damage credibility.
A weak discovery conversation can leave an important business problem unexplored.
A premature product pitch can cause the seller to lose the customer’s attention.
And once that moment has passed, the seller cannot simply say:
“Can we start over? I’d like to try that again.”
This is where AI-powered role-play becomes incredibly interesting.
Platforms such as Yoodli, Brevity, and Second Nature represent a growing category of technology designed to create interactive practice experiences rather than relying solely on static pitch recordings.
Yoodli, for example, currently positions its platform around AI-powered role-play for scenarios including discovery, objection handling, demos, and executive conversations, while Second Nature similarly offers conversational AI practice for sales scenarios.
The larger opportunity is bigger than any one platform.
AI can create a sales practice environment.
And that changes the economics of practice.
Human Role-Plays Are Valuable, but They Are Difficult to Scale
Traditional role-playing has always had value.
I’ve used it.
I’ve facilitated it.
I believe in it.
Having one person play the seller while another plays the customer can create valuable learning moments.
But traditional role-play also has limitations.
First, it consumes a tremendous amount of time.
Imagine a sales leader managing ten sellers.
The leader wants each seller to practice a difficult discovery conversation.
If each meaningful role-play, debrief, retry, and coaching conversation takes only 30 minutes, that is already five hours.
For one scenario.
One round.
Now add:
New products.
New messaging.
Competitive situations.
Objection handling.
Negotiation.
Prospecting.
Executive conversations.
Renewals.
Different buyer personas.
Different industries.
Different skill levels.
The math becomes difficult very quickly.
Managers should coach.
Enablement teams should facilitate practice.
But neither group has unlimited hours.
AI can absorb a significant portion of the repetitive practice burden while allowing humans to spend more of their limited time where human judgment creates the most value.
Second, human role-plays can be inconsistent.
Playing a realistic customer is harder than it looks.
The person playing the buyer may accidentally help the seller.
They may give away too much information.
They may forget the customer profile.
They may make the objection unrealistically difficult.
Or unrealistically easy.
They may behave differently for Rep A than they did for Rep B.
Even experienced facilitators can introduce variation.
Well-designed AI scenarios can help organizations create greater consistency around the persona, business situation, objections, objectives, and information available to the seller.
That doesn’t make AI automatically better than a human.
It makes AI useful for a different part of the learning process.
Third, people behave differently when they feel judged.
Anyone who has facilitated role-plays has probably seen it.
The rep suddenly becomes self-conscious.
Their peers are watching.
Their manager is evaluating.
Enablement is taking notes.
The seller isn’t only thinking about the conversation anymore.
They’re thinking:
“How do I sound?”
“Did I mess that up?”
“What is my manager thinking?”
“Is everyone judging me?”
Sometimes that pressure is useful. Eventually, people need to perform under pressure.
But it isn’t always the best environment for early practice.
Learning requires room for imperfection.
AI creates something psychologically important:
Permission to fail privately.
Try something.
Get it wrong.
Start over.
Take a different approach.
Experiment with a question you’ve never used.
Run the objection again.
And again.
And again.
Nobody rolls their eyes.
Nobody becomes impatient.
No prospect gets burned.
No manager has to play the same customer persona 20 times.
Just practice.
That matters.
Research on expertise associated with psychologist K. Anders Ericsson emphasized the importance of purposeful, structured practice rather than simply accumulating hours. Deliberate practice focuses attention on specific aspects of performance and uses feedback to drive improvement.
That distinction is critical.
Because practice alone does not guarantee improvement.
Practice Does Not Make Perfect. Practice Makes Patterns.
We’ve all heard the phrase:
“Practice makes perfect.”
I prefer a slightly different version:
Practice makes patterns.
Practice something well repeatedly, and good behaviors can become easier to retrieve.
Practice something poorly repeatedly, and poor behaviors can become habits too.
That is why simply giving sellers unlimited access to AI role-play isn’t enough.
The goal is not:
Practice more.
The goal is:
Practice better.
An effective AI role-play program should intentionally focus practice around defined behaviors and realistic situations.
For example:
A seller should not simply “practice discovery.”
What specifically are we developing?
The ability to recognize business impact?
Follow-up questioning?
Listening?
Quantification?
Identifying decision criteria?
Executive-level curiosity?
Avoiding premature pitching?
AI becomes more useful when the practice goal becomes more specific.
The same principle applies to coaching.
Trying to improve everything at once often means improving nothing particularly well.
AI Can Give You Feedback. That Doesn’t Mean the Score Is the Truth.
One of the most useful features of modern learning technology is the ability to evaluate performance quickly.
But we need to be careful.
Because the easier something becomes to measure, the easier it becomes to overvalue the measurement.
This is where I believe sales organizations need a healthy perspective on scorecards.
Let me be clear:
I am not against scorecards.
I’ve used them for decades.
They can provide structure.
Consistency.
Expectations.
Data.
And a common language for performance.
But scorecards are tools.
They are not the truth.
Over my career, I’ve seen outstanding sellers score surprisingly low on scorecards.
I’ve also seen lower-performing sellers check every box beautifully.
Why?
Because selling is contextual.
Imagine a discovery scorecard requiring eight questions.
Seller A asks all eight.
Perfect score.
But the customer gives a fascinating answer to Question #2 that exposes the most important business issue in the entire conversation.
Seller A ignores it.
Why?
Because Question #3 is next.
Seller B asks only five of the eight questions.
But after Question #2, they hear something important.
They stop.
Explore.
Clarify.
Ask another question.
Quantify the impact.
Challenge an assumption.
And uncover an opportunity neither party had fully considered at the start of the conversation.
Who conducted the better discovery call?
The scorecard may say Seller A.
The customer may say Seller B.
That is why:
A scorecard can measure adherence.
It cannot fully measure judgment.
A scorecard can tell you that something was said.
A great coach asks whether it needed to be said at that moment.
AI Should Help Inform Human Judgment, Not Replace It
This is where I believe organizations can make a mistake when adopting AI.
We sometimes take a process that wasn’t perfect before AI and simply automate it.
Old model:
Seller performs → Human completes scorecard → Seller gets score.
New model:
Seller performs → AI completes scorecard → Seller gets score faster.
That is more efficient.
But is it more developmental?
Not necessarily.
Efficiency and effectiveness are not the same thing.
The greatest value of AI may not be replacing the human coach.
It may be allowing the human coach to spend less time on repetitive evaluation and more time on higher-order coaching.
Instead of:
“You scored 72.”
We can ask:
“Why do you think the conversation changed here?”
Instead of:
“You missed a discovery question.”
We can ask:
“What did you hear when the customer said that?”
Instead of:
“You should have asked about business impact.”
We can ask:
“What question could you have asked next to understand the impact more deeply?”
Those conversations develop something more valuable than compliance.
They develop thinking.
Reflective Coaching: Help Sellers See What They Cannot Yet See
Many top-performing sellers naturally reflect on their own performance.
They leave a conversation thinking:
“I started pitching too early.”
“I should have stayed with that topic longer.”
“That objection wasn’t really about price.”
“I talked too much.”
“I missed what she was trying to tell me.”
“That question worked. I need to use it again.”
That internal conversation is incredibly valuable.
The seller isn’t simply performing.
They are studying their own performance.
But not every seller has developed that level of self-awareness.
And even highly self-aware sellers have blind spots.
That is where reflective coaching becomes critical.
Donald Schön’s influential work on reflective practice distinguished between reflection occurring while professionals are acting and reflection after an experience. The broader idea is highly relevant to selling: expertise involves more than executing a technique; it includes learning to think about one’s own actions and adapt.
A great sales coach helps sellers develop that ability.
Not by providing every answer.
But by helping sellers discover more answers themselves.
Ask Before You Tell
This may be one of the simplest and most difficult coaching disciplines.
When you know the answer, you want to give it.
The seller says:
“I don’t know what I could have done differently.”
The coach immediately thinks:
“I do!”
And then we tell them.
Sometimes that is appropriate.
Coaches should absolutely teach.
But if every coaching conversation begins and ends with the manager providing the answers, we may unintentionally create dependence.
The better long-term question is:
Can I help this seller become better at recognizing the answer without me?
Consider the difference.
Telling:
“You should have asked about the financial impact.”
Reflective coaching:
“When the customer said this project was delaying their expansion, what opportunities did that create for you to explore?”
Or:
“What might you have asked to understand the impact of that delay?”
The first approach gives the seller a fish.
The second helps teach the seller how to fish.
And eventually, we want sellers recognizing these moments during the conversation, not only when their manager points them out afterward.
A Practical Framework: Factor 8’s COACHN Model
One framework I’ve found valuable for structuring coaching conversations is Factor 8’s COACHN℠ Model.
Factor 8 defines the framework as:
C — Clarify Expectations
O — Observed Behavior
A — Ask Questions
C — Commit to Actions
H — How Can I Help
N — Next Steps
Here is how I think about applying those principles to sales coaching.
Clarify Expectations
What are we actually coaching today?
Discovery?
Listening?
Objection handling?
Executive presence?
Question quality?
Negotiation?
Do not try to coach the entire call.
Be intentional.
If today’s focus is discovery, the seller’s imperfect closing technique may not be today’s priority.
Coaching becomes more useful when both people know what they are looking for.
Observed Behavior
Describe what happened.
Not what you assume happened.
For example:
“You asked three questions in the first ten minutes.”
Is an observation.
“You weren’t curious enough.”
Is an interpretation.
Observations create better coaching conversations because they give the seller something concrete to examine.
Also observe patterns.
Does the seller:
Interrupt when excited?
Answer objections too quickly?
Ask multiple questions at once?
Move to product too early?
Avoid discussing money?
Fail to follow up when customers reveal emotion?
Talk more when uncertain?
Repeated behaviors tell us far more than isolated mistakes.
Ask Questions
This is where the reflection happens.
Ask questions that make the seller think.
“What did you notice?”
“What were you trying to accomplish there?”
“What did the customer’s response tell you?”
“What else could that statement have meant?”
“Where did the conversation change?”
“If you could replay that moment, what would you try?”
“What question might have helped you go deeper?”
Patience matters.
Silence matters.
You do not have to rescue the seller from every pause.
Sometimes the pause is where learning begins.
Commit to Actions
“Improve discovery” is not an action.
It is an aspiration.
A stronger commitment might be:
“On my next three discovery calls, when a customer identifies a business challenge, I will ask at least one follow-up question before moving to another topic.”
Now we have something observable.
Specific.
Practiceable.
Coach-able.
How Can I Help?
Development should not end with:
“Go fix it.”
Ask what support is needed.
Another AI scenario?
A strong-call example from Gong?
A job aid?
A coaching session?
A demonstration?
Peer practice?
Additional product knowledge?
Sometimes the behavior we see is not the actual problem.
A seller who avoids certain questions may lack confidence because they do not know what to do with the answer.
That requires a different intervention.
Next Steps
Every coaching conversation should go somewhere.
Who owns what?
What will happen next?
When will we review it?
What will success look like?
One of my favorite practices has been asking sellers to send me a few bullets after a coaching conversation:
What did you learn?
What will you do differently?
What outcome do you hope to achieve?
It is simple.
But it moves the seller from passive recipient to active participant.
Confidence Matters — But Confidence Is Not Competence
There is another psychological dimension worth considering.
Practice can build confidence.
And confidence matters tremendously in sales.
A seller who believes they can handle a difficult conversation is more likely to engage rather than avoid it.
But confidence alone is not competence.
Someone can feel very confident and still perform poorly.
Someone else can be highly capable and still underestimate themselves.
Albert Bandura’s work on self-efficacy helps explain part of this relationship. Self-efficacy concerns people’s beliefs about their capabilities in specific situations, and mastery experiences—actually succeeding at challenging tasks—are an important source of those beliefs.
This is another reason repeated practice matters.
AI can give sellers opportunities to experience:
“I handled that objection.”
“I figured out what to ask next.”
“I recovered when the conversation changed.”
“I can do this.”
But we should not confuse an AI score saying “92%” with genuine mastery.
The real goal is not confidence created by a number.
It is confidence earned through increasingly competent performance.
The Opportunity: Create a Continuous Development Loop
This is where I believe AI and human coaching can become much more powerful together.
Rather than treating:
Training,
AI role-play,
Coaching,
Conversation intelligence,
And performance measurement
as separate activities, connect them.
I think of the development cycle this way:
LEARN → PRACTICE → REFLECT → COACH → RE-PRACTICE → PERFORM → REVIEW
1. LEARN
Introduce the concept.
Maybe it is discovery.
Objection handling.
Negotiation.
A new product.
A methodology.
A competitive message.
Give sellers enough knowledge to understand what “good” should look like.
But do not confuse knowing with doing.
2. PRACTICE
Move quickly from passive learning into application.
Use AI role-play to create a realistic situation.
Give sellers a customer.
A problem.
A persona.
An objective.
An obstacle.
Let them experience the conversation.
Not just describe what they would do.
Actually do it.
3. REFLECT
Before giving the seller a score or an answer, ask them:
How did that feel?
What worked?
Where did you struggle?
What surprised you?
What would you do differently?
This is where we begin developing the seller’s ability to observe themselves.
4. COACH
Now bring in human judgment.
Review patterns.
Ask questions.
Use AI-generated observations and scorecards as inputs—not absolute truth.
Help the seller identify one or two high-value behaviors to improve.
5. RE-PRACTICE
This may be the step organizations skip most often.
Do not simply tell someone what to improve and send them back into the field.
Practice the improvement.
Immediately, if possible.
Run the scenario again.
Change the objection.
Change the persona.
Increase the difficulty.
This is where AI becomes extraordinarily valuable.
The coach does not have to personally conduct every repetition.
6. PERFORM
Now take the behavior into the real world.
The customer conversation.
The sales meeting.
The discovery call.
The negotiation.
The demo.
The recital.
7. REVIEW
Use real-world conversation data where available.
Platforms such as Gong can help organizations review actual customer conversations at scale.
What transferred?
What didn’t?
Did the seller demonstrate the behavior?
What happened when reality differed from the simulation?
And then:
Return to practice.
The process is not linear.
It is a loop.
Learn → Practice → Reflect → Coach → Re-Practice → Perform → Review
Then repeat.
AI Should Make Human Coaching Better, Not Make Humans Disappear
There is a tendency whenever new technology emerges to frame the conversation as replacement.
Will AI replace trainers?
Will AI replace sales coaches?
Will AI replace managers?
Will AI replace sellers?
I believe those questions often miss the more interesting opportunity.
The better question is:
What becomes possible when AI does what AI is good at and humans do what humans are good at?
AI is exceptionally useful for:
Repetition.
Availability.
Consistency.
Simulation.
Scalability.
Rapid feedback.
Pattern identification.
Safe experimentation.
Humans remain exceptionally valuable for:
Empathy.
Context.
Nuance.
Judgment.
Motivation.
Trust.
Reflection.
Accountability.
Understanding the person behind the performance.
We should not force either side to do everything.
Use technology to extend human capability.
Not remove humanity from development.
Don’t Practice on Prospects
There may be no simpler way to summarize the opportunity.
Organizations spend enormous amounts of money generating demand.
Marketing campaigns.
Events.
Advertising.
SDRs.
Technology.
Data.
Content.
Partnerships.
Branding.
Then, after all that investment, a lead finally agrees to talk with a seller.
That should not be the seller’s first meaningful attempt at handling the conversation.
Practice before the prospect.
Practice the difficult objection before hearing it from the CFO.
Practice discovery before meeting the customer.
Practice explaining value before the executive presentation.
Practice the negotiation before the contract is at risk.
Practice.
Reflect.
Adjust.
Then perform.
Back to the Piano
Let’s return to where we started.
I cannot teach someone to become a skilled pianist in one coaching session.
The finest instructor in the world cannot do that.
But the opposite is also true.
If someone’s only practice occurs while the instructor is sitting beside them, their progress will be painfully slow.
They need both.
They practice alone.
They practice with an instructor.
They make mistakes.
They receive feedback.
They reflect.
They correct.
They practice again.
Eventually, their fingers begin finding the keys without conscious effort.
They hear mistakes sooner.
They adjust faster.
They become more confident because they have become more competent.
Then comes the recital.
Salespeople deserve the same developmental environment.
Let AI provide the repetitions.
Let humans provide the reflection, judgment, empathy, and coaching.
Let AI create safe places to experiment.
Let coaches help sellers understand what those experiences mean.
Let scorecards provide data without allowing the score to become the whole story.
Let sellers practice the difficult conversations before those conversations matter.
And let technology help us build something sales organizations have struggled to create for decades:
A scalable culture of practice.
Because the goal should never be simply to complete training.
Or pass a certification.
Or earn a score.
The goal is behavior change.
Better judgment.
Better conversations.
And better outcomes.
The customer call should not be rehearsal.
It should be the recital.
A Question for Sales and Enablement Leaders
How does your organization currently develop conversational sales skills?
Are your sellers primarily watching training?
Recording static pitches?
Practicing with managers?
Using AI role-play?
Reviewing real customer calls?
Or have you created a connected system where AI-powered practice and human coaching reinforce one another?
The opportunity, in my view, is not choosing between artificial intelligence and human intelligence.
It is understanding how to combine them.
The Power of AI Practice.
The Might of Human Coaching.
That may be where the next generation of seller development begins.
About the Author
Eric Matthews is a sales and enablement leader with 28 years of experience across sales, sales leadership, learning, and enablement. Through Successful Talent, he focuses on helping organizations improve seller performance by bringing together sales psychology, adult learning, coaching, technology, AI, and practical enablement strategies.
About Successful Talent
Successful Talent helps organizations turn knowledge into performance through sales and revenue enablement, learning design, AI-enabled practice, coaching, onboarding, and performance development.
Learn. Practice. Reflect. Improve. Perform.