How to Run a Cold-Call Roleplay That Produces a Coachable Observation, Not a Vibe Check
Most roleplay sessions end in 'good energy, keep it up.' Here's the structure that turns 15 minutes into a specific, gradeable competency observation.
Fifteen minutes of roleplay, three sentences of feedback: "Good energy, I liked your confidence, keep it up." I've sat through hundreds of these sessions, run plenty myself in my first years managing, and none of them produced anything a rep could actually go and practise. "Good energy" isn't a skill. It's not attached to a competency, it doesn't tell the rep what to repeat versus what to change, and it will be forgotten by the next 1:1. The problem isn't that managers are bad at feedback. It's that the roleplay itself was never built to produce an observation — it was built to produce a feeling, and feelings don't compound into skill.
Here's the structure that fixes it. It's not complicated, but every step is load-bearing — skip one and you're back to vibes.
Step 1: Fix the scenario before you fix the rep
Open-ended roleplay — "pretend I'm a prospect, cold call me" — cannot produce a comparable observation because the scenario changes every time. You can't tell whether a rep improved at handling a gatekeeper if last week's gatekeeper caved after one line and this week's held the line for three minutes.
Write the scenario down, once, and reuse it. Fix the persona (title, company size, mood), fix the entry condition (cold dial, no prior contact), and fix at least one obstacle the persona must raise regardless of how the call goes (a specific objection, a gatekeeper who asks a specific screening question). Now every rep who runs this scenario is being measured against the same conditions, which is the only way a manager can honestly say someone got better.
Step 2: Name the competency before the call starts, not after
This is the step almost everyone skips. If you decide what you're grading after you've watched the roleplay, you'll grade whatever happened to stand out — usually tone, which is the least useful thing to grade because it's the hardest to turn into a repeatable behaviour.
Pick one named competency per roleplay. Not "cold calling" as a whole — that's a bucket of a dozen distinct skills. Pick one: Objection Reframe. Discovery Depth. Gatekeeper Navigation. Pattern Interrupt on the Open. Tell the rep which one before they pick up the phone. This does two things: it tells the rep exactly what "good" means for this session, and it stops you drifting into unrelated commentary about their voice or pacing when that's not what today's session is for.
Step 3: Constrain the roleplay so the target competency actually gets tested
A scenario can be fixed and still fail to test the competency you named, if the persona gives up the objection too easily or never raises it at all. Build the constraint into the persona's instructions, not the rep's. If today's competency is Objection Reframe, the persona must raise the same objection at the same point in the call, hold it for at least one rebuttal, and only concede on a genuinely strong reframe — not on repetition or persistence. That's the difference between testing whether a rep can reframe an objection and testing whether they can wear someone down, which is a different skill entirely and not one you want to be accidentally rewarding.
Step 4: Debrief against a rubric, not against a feeling
This is where "good energy" gets replaced with something gradeable. Before the session, write four short descriptions of what novice, developing, proficient, and expert look like for the named competency — specific enough that two different managers watching the same call would land on roughly the same grade.
| Level | Objection Reframe — what it sounds like |
|---|---|
| Novice | Repeats the same pitch louder after the objection, or agrees and moves on |
| Developing | Acknowledges the objection, then pivots to an unrelated benefit |
| Proficient | Reframes the objection into the actual concern underneath it, ties it to one relevant proof point |
| Expert | Reframes, ties to proof point, and uses the reframe to open a new discovery question rather than closing the topic |
Against that table, "good energy" isn't an option. You have to say which level the rep hit and, more usefully, what specifically was missing between where they landed and the level above it.
Step 5: Write the observation down in one sentence with a quote attached
The debrief conversation evaporates by the following week unless it's captured as a specific, referenceable sentence. Not "handled objections well" — that's the vibe-check version wearing a rubric's clothes. Something like: "On the 'we're happy with our current provider' objection, reframed to switching-cost concern and cited the migration timeline, but didn't open a follow-up question — proficient, one step short of expert." That sentence is coachable next time. It tells the rep exactly what to add, and it gives you a dated record to compare against three roleplays from now, which is the only way either of you will know whether the coaching actually worked.
Putting the five steps to work
None of this requires new software. It requires a written scenario, a named competency, a rubric, and the discipline to fill in the observation sentence before you move to the next rep. If you're building your first set of scenarios, a Sales Certification Rubric (Novice-to-Expert Levels) gives you a starting structure for the four-level grading rather than writing one from scratch, and running the actual debrief conversation against a GROW Model Coaching Conversation Script keeps you from sliding back into generic encouragement once the rep starts looking disappointed in themselves. Log the sentence somewhere durable — a Call Coaching Feedback Template works fine — so the next roleplay is measured against a written baseline instead of your memory of how last week felt.
The fifteen minutes were never the expensive part. What's expensive is running that fifteen minutes forty times a quarter and having nothing to show for it except reps who feel encouraged and managers who can't say, specifically, what got better.