3 hard truths about AI role play (and what to do about them)

Over the last two and a half years, we've done more than 150 deployments of AI role play technology across some of the largest enterprise companies in the world. Workday, Intel, IBM. Teams of hundreds, sometimes thousands of reps.

And I'll be honest with you: we've learned some things the hard way.

When we started HyperBound, the pitch felt almost self-evident. Practice makes perfect. Everyone agreed. You'd walk into a room, make the case for AI role play, and heads would nod. Then you'd deploy it, and... nothing. Tumbleweeds.

So we stopped trying to come up with clever workarounds and started asking a more honest question: why?

What we found, distilled from all those deployments and conversations, came down to three hard truths. Three things that, once you understand them, completely change how you think about rolling out enablement technology and driving what we now call revenue activation.

Hard truth one: reps don't want to practice

There. I said it.

This was a difficult thing for us to reckon with, especially as a role play company. We believed deeply in the value of practice. We still do. But belief doesn't create behavior.

When we started digging into why adoption was so low, we kept coming back to the same simple reality. Reps are measured on two things: activity and performance. That's it. Those are the metrics that determine whether they make money, whether they hit quota, whether they keep their jobs.

So unless practice is directly connected to one of those two things, it's always going to feel like extra work. And people don't voluntarily take on extra work when they're already stretched thin.

We had one customer where adoption sat at around 20%, no matter what we tried. We dedicated a full-time person to cracking it. We tried competitions, gamification, all kinds of creative nudges. Nothing moved the needle in any meaningful way.

That's when we realized the problem wasn't the tactics. The problem was that we were trying to manufacture motivation artificially, and that just doesn't hold. A rep has to want to open HyperBound. You can't engineer that from the outside in.

The only way to genuinely create that pull is to make the connection between practice and performance undeniable. And to do that, you need data.

Specifically, you need to show a rep that the objection they keep stumbling on in real deals is exactly what they're about to practice. That changes the framing entirely. It's no longer "here's a generic skill-building exercise." It's "here's what's costing you deals right now, and here's how you fix it."

When practice is tied to a rep's actual pipeline, their actual calls, their actual struggles, the motivation becomes intrinsic. They do it because it clearly helps them. And that's the only kind of adoption that actually sticks.

For expert advice like this straight to your inbox, sign up for Pro+ membership.

You'll also get access to 100+ hours of exclusive video content, a complimentary Summit ticket, and so much more.

So, what are you waiting for?

Get Pro+

Hard truth two: frontline managers are set up to fail

Frontline managers carry more responsibility in a sales org than almost anyone else. They spend the most time with reps week over week. They have the highest potential to drive real behavior change. And yet, in most organizations, they get the least support.

Think about how most frontline managers end up in that role. They were great reps. They hit their numbers. They got promoted. And then, suddenly, they're expected to coach a team, run pipeline reviews, forecast accurately, and hold people accountable, often without any real training on how to do any of it.

They know how to be accountable for themselves. They've never had to figure out how to be accountable for eight, ten, fifteen, sometimes twenty people at once.

At a one-to-eight manager-to-rep ratio, it's hard but manageable. We work with companies where that ratio is one-to-fifteen or even one-to-twenty. At that point, you're asking someone to do something that's genuinely impossible. And then we pile on dashboards, data requests, coaching expectations, and forecasting calls, and wonder why nothing gets done.

What ends up happening is decision paralysis. There's so much information coming at these managers that they can't prioritize any of it. So they prioritize nothing, or they default to the most urgent fires rather than the most important work.

Getting frontline managers bought into your enablement strategy requires making their job easier, not harder. It means giving them specific, actionable information rather than asking them to swim through thirty-five dashboards to find a signal.

It means surfacing what their reps are struggling with without requiring them to listen to every call. It means making the path from insight to action short enough that they'll actually walk it.

Hard truth three: CROs want an easy button, and that's actually reasonable

I've spoken with more than 150 CROs in the last six months. And almost every single one of them wants the same thing: a fast, clear path from initiative to revenue impact.

When I first started encountering this, I'll admit I found it a little frustrating. These are experienced leaders. They know business doesn't work that simply. But then I started thinking about it from their perspective, and it makes complete sense.

The average CRO tenure is around 18 months. That's it. So if you're a CRO and you want to run a real initiative, say you've changed your pricing, and you need all your reps to start delivering that new message consistently, that alone can take six months to actually shift behavior.

Then you need another three to six months to see whether it's actually moving closed-won revenue. If everything goes perfectly, you get two or three real at-bats during your entire tenure.

Two or three. That's not a lot of room for error.

So when a CRO looks at a coaching and practice program and thinks "that's not the fastest path to revenue," they're not being dismissive. They're doing the math. And the math feels brutal.

The challenge for us as enablement professionals is that we know coaching and practice do drive revenue. That's why we have the jobs we have. But knowing it and proving it in the numbers are two very different things. And if we can't demonstrate the connection clearly and quickly, we're going to keep losing that conversation.

The answer, again, comes back to data. And to something we need to be honest about.

Your insights are dying in dashboards

Here's a stat that should give everyone pause: 95% of conversational intelligence insights, call recordings, transcripts, and emails never get reviewed.

Managers spend about 5% of their time coaching, partly because they're only listening to around 1% of actual calls. Not because they don't care, but because reviewing all of it is simply not humanly possible.

So you have this enormous amount of data sitting in your systems, and almost none of it is being turned into action. It's not that the data isn't valuable. It's that no human can process it at the scale at which it exists.

And this is where AI is genuinely changing things, though maybe not in the way people expect.

The problem with most dashboards is that they show you what happened without telling you what to do about it. You can see that talk-listen ratios went up last month.

You can see that deal velocity slowed in Q2. But when you ask a follow-up question, most people can't explain why, and they certainly can't tell you what to change.

By the time you've jumped between ten dashboards, gathered some anecdotal feedback, and formed a hypothesis, two or three months have passed, and the competitive landscape has already shifted.

Dashboards, as they currently exist, are where insights go to die. They feel like progress because they're full of information. But information without action is just noise.

Revenue activation: applying the scientific method to enablement

What we've been building toward at HyperBound is a framework we call revenue activation. And at its core, it's just the scientific method applied to sales enablement: observe, hypothesize, experiment.

Observe. You have data everywhere. In your CRM, in your call recorder, in your email platform. The first step is actually using it. Not skimming it, not waiting for a quarterly review, but systematically analyzing what's happening in the field right now.

Hypothesize. Once you've observed, you start forming theories. Which objections are reps struggling with most often? Which competitors are they consistently losing to, and why? Which personas tend to trigger which deal stalls? These aren't hunches anymore. They're patterns pulled from real conversations and real opportunities.

Experiment. Now you act on those theories. You build targeted practice scenarios around the specific objections that are costing you deals. You create role plays tied to actual personas from an actual pipeline. And you measure whether it moves the needle.

The shift here is subtle but important. When practice is connected to observed data from the field, it stops feeling arbitrary. A rep who knows they're about to face a CFO who always pushes back on pricing, and who has just practiced that exact conversation three times before the call, is in a fundamentally different position than a rep who completed a generic MEDDPICC certification six months ago.

What this looks like in practice

Let me walk you through how this actually works inside HyperBound, because the theory only means something if you can see it running.

Using our AI super agent, Coda, an enablement leader can ask something like: "What were the top reasons deals stalled in the last 30 days?" Coda pulls from your CRM, your call data, and your email history, and it surfaces the actual patterns.

Legal and security concerns. Budget constraints. Stakeholder bandwidth. Integration uncertainty. And it doesn't just name them. It tells you how many deals were affected and how much pipeline is at risk.

From there, you can go deeper. Ask it to identify the top five objections coming up across your calls. It'll show you specific quotes from real prospect conversations, how frequently each objection appeared, and which personas tend to raise them most often.

And then, here's where it gets genuinely useful: it recommends what to do next. Build a competitive battle card for these specific pricing objections. Create a one-pager on Salesforce and Teams integrations because those keep tripping reps up. Develop a clearer implementation talk track because timeline uncertainty is frustrating buyers.

Once you have that analysis, you can ask Coda to build role play scenarios for each persona, built from your own data. These aren't generic simulations. They're constructed from what's actually happening in your pipeline.

Then you can assign those role plays to specific rep segments and set up automatic notifications to their managers when they complete them, along with a summary of where each rep struggled. Suddenly, managers don't need to track everything manually. They get a clear, timely signal about where their people need support, without having to dig for it.

Just-in-time enablement

The rep-facing side of this is what we call just-in-time enablement, and it's probably the piece I'm most excited about.

When a rep has a call coming up, HyperBound already knows who they're meeting with, what's happened in that deal so far, and what objections are most likely to surface in that next conversation. Before the call, it automatically builds a short, targeted role play, typically three to five minutes, focused on the specific objection that's most likely to come up.

These aren't hour-long training sessions. They're bite-sized, deal-specific, and delivered right when the rep needs them, whether that's through a Slack message, inside their CRM, or wherever they're already working. No login required. No friction.

The manager gets notified when the rep completes it. They know their rep has a high-stakes call tomorrow. They know what the rep practiced. And if the rep didn't practice, that's information too.

This is how you make the connection between practice and performance concrete enough that reps actually feel it. And it's how you give managers the visibility they need without adding to their already impossible workload.

The common thread

When you look at all three hard truths together, reps who won't practice, managers who are overwhelmed, CROs who need to see results fast, the thread running through all of them is the same. Data, and specifically, the gap between having it and using it.

We have more data than ever before. Call recordings, email threads, CRM opportunity data, conversation intelligence. It's all there. The problem is that 95% of it sits untouched because no human team can realistically process it at scale.

AI closes that gap. Not by replacing human judgment, but by doing the analysis work that humans genuinely can't do at volume, and then surfacing clear, actionable recommendations rather than another dashboard to scroll through.

The organizations that are going to win at this are the ones that move from insight collection to insight activation. From dashboards that describe the past to systems that shape what happens next.

That's what revenue activation is about. And it's what we've seen work, again and again, across more than 150 deployments.

If you're in the middle of rolling out AI role play right now, or thinking about it, the single most important thing I'd leave you with is this: don't assume that building the role plays and assigning them is enough.

There's more to the story. Connect the practice to real data from the field, make it easy for managers to see what's happening, and give reps a reason to show up that's tied to their actual pipeline.

Do that, and adoption stops being something you have to chase.

Sriharsha Guduguntla, Co-Founder & CEO at Hyperbound, gave this talk at our Sales Enablement Summit, Seattle, 2026.

Why 95% of your call insights die in dashboards