From Training to Performance: How AI Is Improving Selling Skills

In the domain of learning there are many commonly held beliefs, often because they seem like common sense. Research results sometimes confirm these beliefs; other times they refute them. For instance, many (most?) people believe that people have preferred learning styles and will learn best when learning materials conform to their preferred style. But multiple research studies over many years disprove this belief.

On the other hand, we all know the expression practice makes perfect. Perhaps not surprisingly, this turn out to be correct. Ok, maybe not perfect, but certainly improved. It’s why we practice musical instruments or sports or do lots of math problems to study for an exam. In learning science this goes by the rather unimaginative name the Practice Effect. The magnitude of the practice effect can be substantial. In clinical cognitive studies, practice effects have explained 31% to 83% of the variance in post-test scores, demonstrating how strongly repeated exposure can influence performance.

Some skills, such as practicing the piano, don’t require a partner. But others, such as playing tennis, do. And it’s not just a matter of having the ball returned to you – for that you could hit against a wall – but often another player will return the ball in unexpected ways, requiring rapid decision making and reflexes. Like tennis, practicing selling skills greatly benefits from a practice partner. You could rehearse a sales call repeatedly on your own, but you won’t have the benefit of an unexpected objection or an unanticipated request for more information. In sales, practice works not simply because “repetition makes perfect,” but because repeated performance in realistic situations strengthens cognitive, behavioral, and perceptual capabilities that underlie effective selling.

Sales conversations require many micro-skills: asking targeted questions, active listening, framing value, handling objections, and guiding next steps. In the beginning these behaviors are declarative (“remember to ask open-ended questions”). With practice, they become procedural – automatic responses that require less cognitive effort. Research shows that repeated performance shifts knowledge from conscious recall to automatic execution, freeing cognitive resources for situational awareness and relationship building. Until recently sales representatives wanting to practice their skills had two choices: (1) find a human partner or (2) use a computer-based simulation.  A human partner (a trainer or a coach) is great, but not always available, and simulations were typically limited in robustness. (To make the programming manageable, simulations usually constrain the number of possible paths that the learner can take.)

But today, AI-native practice platforms allow a learner to practice selling skills repeatedly in realistic situations. To understand where this technology is today and where it is headed, we recently spoke with Benjamin Alouf MD, MBA Chief Operating Officer of the AI-native Learning and Intelligence Platform PraxisPro:

Question: What was the original vision behind your product?

Pulling account specialists out of territory for in-person practice is expensive. We built PraxisPro to enable account specialists to practice realistic selling scenarios on-demand, in their territory, with consistent coaching that actually sticks.

Question: What has been the general reaction from the learners? How quickly do they adapt?

The feedback has been overwhelmingly positive. After one or two initial roleplays, account specialists quickly adapt and move to customizing their own scenarios with different HCP types and objection levels; the realistic simulations help them recall and retain prior knowledge.

Question: Do they find the interactions realistic?

Yes, because we train our AI on the customer's proprietary materials and enforce strict compliance guardrails. Account specialists can practice with different HCP personalities and conversation styles, and we're building in memory so (the AI) HCPs remember prior conversations—making it truly dynamic.

Question: Do you score the interactions? If so, how?

We provide star ratings based on how closely responses match the knowledge base, plus color-coded feedback for quick consumption. Users get detailed session recaps showing their transcript alongside specific skill assessments like compliance, product knowledge, active listening, and tone.

Question: Do you have any data showing effectiveness?

We're deployed at top 50 pharma companies with 75%+ completion rates and a 71% average score increase over 9 months. We estimate customers will reclaim 5+ days of rep field time and cut new hire ramp time in half—a clear training and revenue win.

Question: What are the current limitations and how are you addressing them?

Many AI-native platforms are built on standard Large Language Models -- but in a highly regulated industry like pharma a single incorrect word or phrase can create potential compliance risk. We're building proprietary small language models that will take into account the nuances of the life sciences to drive trust and safety in this highly regulated industry.

AI-native roleplay platforms are a robust example of using AI to improve learning effectiveness. Unlike the pre-AI generation of decision-making simulations, they allow for dynamic, responsive conversations that approximate the cognitive and emotional complexity of real selling situations. Learners must retrieve knowledge, interpret cues, and adapt their responses in real time, mirroring the conditions of an actual sales call.

Authored by: Steven Just