A while back we noticed our sellers were coaching themselves. They pasted their pitches into chatbots and asked for a critique. They rehearsed objections on their own time, out of their managers' view.
It would have been easy to call that a governance problem and shut it down. We saw it as demand. Sellers clearly wanted private, self-directed practice. It just had no structure, no quality bar, and no link to what winning conversations actually sound like.
That became our change management starting point: find the workaround people have already built, and give it a better home.
Next week in Carlsbad, Beth Potter and I are telling that story at Seismic Shift 2026, in our session "AI role play turns enablement into repeatable sales practice." This post is the preview. The full write-up comes after the talk.
The gap we were closing
Our commercial teams already trusted one place to find what they needed. But being able to find the right deck doesn't make you confident when a client pushes back. What sellers were missing was a sense of what "good" sounds like, a chance to practice before the real conversation, and coaching that didn't depend on a manager's calendar.
The tech was the easy part
Role play inside Seismic gave sellers a realistic client conversation they could repeat as often as they liked. It sounded convincing from day one, and one of the personas got so popular that sellers made him his own Slack emoji.
The real work was everything around it:
- Building scenarios from real call signals so practice feels like the actual job
- Keeping the scenario library fresh as objections change
- Using scorecards so "good" is visible and consistent
- Spreading practice across channels sellers already use, so it becomes a habit
- Treating trust and privacy as requirements from the start
Did it work?
We tested it with a product launch this summer. The sellers who practiced performed differently from the ones who didn't. We'll share the detail in the session.