Most growing companies discover the same uncomfortable truth about sales training. The content is usually fine. The playbooks exist, the messaging deck is current and the onboarding curriculum has been rewritten twice this year.
What is missing is practice. Reps sit through the training, nod along, then walk into a live call and improvise. The first time they attempt a difficult objection out loud, it happens in front of a real buyer with real revenue attached.
That gap between knowing and doing is where most sales enablement budgets quietly leak value. It is also the part of the problem that has changed most in the past twelve months.
For corporate leaders watching headcount costs and ramp times, it is worth understanding what has shifted and what has not.
The practice gap nobody budgets for
Every sales organisation has a handful of conversations that decide whether a deal advances. The cold call opener. The discovery question that uncovers real pain rather than a stated preference.
Then there is the moment procurement asks for a discount and the rep has to hold the line without souring the relationship. Four or five exchanges, spread across a deal cycle, carry most of the commercial weight.
These are motor skills, not knowledge. They are learned through repetition under mild pressure, then corrected, then repeated again. Nobody becomes fluent in objection handling by watching a video about objection handling.
Yet the standard enablement model is almost entirely informational. Companies invest heavily in creating the material and then leave the rehearsal to chance, or to whatever a line manager can squeeze into a Thursday afternoon.
Why the old model breaks at scale
Manual roleplay works beautifully with six reps and one attentive manager. It stops working somewhere around thirty.
The arithmetic is unforgiving. A manager with eight direct reports who runs one meaningful practice session per person per month is already spending a full day a month on rehearsal alone, before any deal review or pipeline work. Multiply that across a distributed team and the practice simply does not happen.
Consistency suffers too. One manager grades hard, another grades kindly and a third has never been trained to give structured feedback at all. Two reps can deliver identical performances and receive contradictory advice, which teaches them very little.
There is also a dignity problem that enablement leaders rarely mention. Peer roleplay in a group setting is uncomfortable, so people either avoid it or perform politely for each other. Politeness is the enemy of useful practice.
What actually changed
The category that emerged to solve this now looks materially different from the text based simulators of a few years ago. Reps can hold a live, multi turn conversation with an on screen avatar that listens, pushes back, raises objections and behaves like a buyer who is busy and unconvinced.
Synthesia, the London based AI video company behind Roleplay Sessions, released its version of this in July 2026, extending its existing enterprise training platform into interactive practice and coaching. The company reports that its wider platform is used by more than 50,000 organisations and over ninety per cent of the Fortune 100.
For teams evaluating the market, the company has published a hands on review of the best AI sales roleplay tools that assesses each option against realism, feedback quality, analytics depth, customisation, ease of use, integrations and sales fit.
The structural shift is that repetition became cheap. A rep can run the same discovery call eleven times in an evening, get scored on each attempt and never burn a live opportunity to do it.
What separates a serious platform from a demo toy
The market has filled quickly, and not every product in it survives contact with a real enablement programme. A few criteria tend to separate the two.
Realism is not optional. If the simulated buyer agrees with everything, reps learn that selling is easy and arrive at real calls unprepared. The conversation needs to handle interruptions, resist a weak pitch and occasionally end badly.
Scenarios must reflect your business. Generic templates are useful for a first week. After that, practice needs to run on your products, your buyer personas, your objections and your sales methodology, ideally built from documents and call transcripts the team already has.
Feedback has to be specific. A numerical score with no explanation changes nothing. Useful coaching names the moment, explains what went wrong and describes what a stronger response would have sounded like.
Analytics should measure skill, not attendance. Completion rates tell a manager who clicked through. Skill breakdowns, score trends across attempts and drill down into individual transcripts tell them who is actually getting better.
Coverage should match your sales motion. A team running high volume outbound has different rehearsal needs from an enterprise team managing eighteen month cycles. Synthesia supports cold calls, discovery calls, objection handling, negotiation, renewals and expansion conversations alongside field sales scenarios where no call recording exists to review.
Language coverage matters for distributed teams. Roleplay Sessions currently runs in English, German, Spanish and French, with the surrounding video training platform supporting more than 160 languages. For a company selling across several European markets, that difference decides whether practice reaches the whole organisation or only headquarters.
Security has to clear procurement. These systems ingest pricing, positioning and competitive strategy. Synthesia lists SOC 2 Type II, ISO 27001, ISO 27701 and ISO 42001 certification, GDPR compliance with EU data residency and SSO with SCIM for access control.
Where practice fits in the wider revenue engine
Rehearsal is one component of a system, not a replacement for one. It sits downstream of positioning and targeting, and it only pays off if the conversations reps are practising are the ones that matter.
That means practice design should follow demand generation design. If a company has already rebuilt its B2B lead generation strategy around account based targeting and multi stakeholder buying committees, the roleplay library needs to reflect that reality rather than a single decision maker on a cold call.
The same discipline applies to who practises. Enablement teams often default to new hires, but the largest measurable gains frequently come from mid tenure reps who have plateaued and stopped receiving structured feedback altogether.
The integration question follows the same logic. Platforms built on existing training infrastructure tend to export cleanly to a learning management system through SCORM and xAPI, which matters if learner analytics already live there. Teams that want practice generated from live CRM opportunities are solving a different problem and should weight that requirement accordingly.
A realistic rollout sequence
Start narrow. Pick the single conversation where the team loses the most deals, usually discovery or pricing pushback, and build one scenario properly rather than twelve badly.
Define the rubric before anyone practises. Write down what a strong discovery call actually contains, in the language your team uses, and score against that. A rubric assembled after the fact tends to reward whatever the tool happens to measure.
Make the first cohort voluntary. Early adopters who enjoy the format will do the internal selling far more effectively than a mandate, and their scores give you a baseline worth comparing against.
Keep managers in the loop without making them the bottleneck. The most effective pattern is a short weekly review where a manager reads two or three transcripts and adds one comment each, rather than sitting in on sessions live.
Then measure something that matters. Ramp time to first closed deal, win rate on the practised motion or objection handling scores at ninety days are all defensible. Session counts are not.
Frequently asked questions
Does AI roleplay replace manager coaching? No. It removes the mechanical repetition from a manager’s workload so their time goes to the judgement calls that genuinely need a human, such as account strategy and career development.
How long before this shows up in revenue? Most teams see behavioural change within a quarter and commercial signal in two. Ramp time for new hires tends to move first because the effect is largest where the baseline is lowest.
What if reps refuse to use it? Adoption usually fails for one of two reasons. Either the simulation is unconvincing, or the feedback is too generic to feel worth the effort. Both are product selection problems rather than culture problems.
The uncomfortable conclusion
Most companies do not have a sales training problem. They have a sales practice problem, and they have been solving it with more content.
The organisations pulling ahead are the ones that treat rehearsal as infrastructure rather than an event. They have decided that the first time a rep handles a hard objection should not be the time it costs them a deal.
