Communication Skills Workshop: Agenda, Activities, Outcomes
Quick Summary
A half-day communication skills workshop plan: timed agenda, 10 activities with instructions, an observer rubric and how to measure behaviour change after.
A good communication skills workshop works on two or three behaviours that matter in your people's real conversations. It practises them under light pressure with structured feedback, then checks whether those behaviours show up at work 30 to 90 days later. The half-day plan below does exactly that. It includes a timed agenda, ten activities with facilitator instructions, an observer rubric and a follow-up measurement plan that doesn't rely on smile sheets.
The plan suits managers, customer-facing teams and sales or customer success groups of 9 to 24 people. If you're working with a different audience, swap the scenarios. The structure still holds.
Before you design anything: pick the behaviours from evidence
Most communication workshops try to cover everything at once: listening, questioning, body language, feedback, conflict and presenting. People leave feeling good and change nothing. A better approach is to find out where real conversations break down and build the session around that.
Here's an example from our own data. A European B2B field sales team (37 reps) recorded 224 real customer meetings with Shiken's meeting recorder, and every meeting was scored against a four-stage meeting rubric. The team averaged about 47/100. The scores by stage were not what most sales trainers would guess:
| Meeting stage | Average score (0–100) |
|---|---|
| Opening / introduction | 38 |
| Close / agreeing next steps | 44 |
| Understanding the customer | 51 |
| Identifying solutions | 56 |
The weakest stage was the first 90 seconds of the meeting, not objection handling or closing. A workshop built on gut feel would probably have spent its time on pitching solutions, the stage the team already did best. The same pilot taught us a second lesson. The team average stayed roughly flat during the pilot, while managers weren't yet coaching against the scores. Measuring a behaviour doesn't change it. Practice and coaching do.
You don't need recordings to run this diagnosis. Two weeks before the session, ask managers to watch three real conversations per person and score them with the observer rubric further down. Then put most of your workshop time into the lowest-scoring behaviour.
The half-day agenda
This agenda runs 3.5 hours, 09:00–12:30, with a 15-minute break. Most of the time goes on practice rather than presentation. That's deliberate.
| Time | Activity | Objective | Materials |
|---|---|---|---|
| 09:00–09:15 | Welcome + "one conversation that matters" | Tie the session to a real upcoming conversation | Sticky notes, baseline scores (if you have them) |
| 09:15–09:40 | 1. The 90-second opener | Open a conversation with relevance and a clear purpose | Opener card, timer |
| 09:40–10:05 | 2. Paraphrase before you reply | Show understanding before responding | Topic cards, observer sheet |
| 10:05–10:30 | 3. The question ladder | Move from closed to open to follow-up questions | Question ladder handout |
| 10:30–10:45 | 4. One-way vs two-way instructions | Check understanding instead of assuming it | Simple line drawings, blank paper |
| 10:45–11:00 | Break | — | — |
| 11:00–11:10 | 5. Seven words | Notice how emphasis changes meaning | One sentence on a slide |
| 11:10–11:35 | 6. SBI feedback rewrite | Turn vague criticism into specific feedback | SBI worksheet, 6 vague-feedback examples |
| 11:35–12:10 | 7. Triad roleplay, two rounds | Apply everything under light pressure, with a score | Scenario cards, observer rubric |
| 12:10–12:25 | 8. Commitment and practice plan | Choose one behaviour and schedule the practice | Practice plan card |
| 12:25–12:30 | Close: what happens next | Explain follow-up practice and measurement | Calendar invites ready to send |
Activities 9 and 10 are swap-ins. Use them for leadership groups, or replace activity 4 if your group has done the drawing exercise before.
10 communication workshop activities (with instructions)
1. The 90-second opener (25 min)
Why: in the pilot data above, openings scored lowest of any stage. An opening that misses earns you nothing for the rest of the conversation. How: each person writes an opener for their "conversation that matters" with three parts: why this conversation, why now (something specific to the other person), and what they'd like to leave with. In pairs, one person delivers the opener while the partner plays the other party. The partner then answers two questions: "Did I know what this conversation was for?" and "Did it feel about me?" Rewrite the opener and deliver it again. Watch for: openers that list the speaker's agenda and say nothing relevant to the listener.
2. Paraphrase before you reply (25 min)
How: work in pairs on a mildly contentious work topic (for example, "we should stop internal meetings on Fridays"). Person A gives their view for one minute. Person B may not respond until they've paraphrased A's point well enough that A says "yes, that's it". Then B gives their view, and the roles reverse. Run three topics. Debrief: how many attempts did it take to get a "yes, that's it"? What did you leave out the first time? Watch for: paraphrases that are really rebuttals ("So what you're saying is, you don't care about…").
3. The question ladder (25 min)
How: hand out a ladder with three rungs. Closed questions confirm facts ("Is this live?"). Open questions explore ("How are you handling it today?"). Follow-up questions go deeper on the last answer ("You said 'mostly'. Where does it break?"). In triads, the interviewer has five minutes to understand a colleague's real work problem. The observer tallies each question by rung. Rotate roles. Debrief: compare each person's ratio of follow-up questions to closed ones. Expect closed questions to dominate in the first round. That's the baseline you're trying to move.
4. One-way vs two-way instructions (15 min)
How: this is a classic. A describer has a simple line drawing and sits back to back with a drawer. In round 1 the drawer may not speak. In round 2 the drawer can ask questions. Compare the two drawings and how long each round took. Debrief: where at work do we run "round 1"? Briefs, handovers and policy emails are the usual answers.
5. Seven words (10 min)
How: put up the sentence "I didn't say she took the money." Volunteers read it seven times, stressing a different word each time. Pairs then do the same with a real sentence from their work (for example, "We can look at the price"). Debrief: the same words carry seven different meanings. Tone travels further than wording on calls and video. (This is a variant of the "Seven Words" method in SessionLab's communication library.)
6. SBI feedback rewrite (25 min)
How: teach the Center for Creative Leadership's Situation–Behavior–Impact model. Describe the situation, the observable behaviour and its impact, then ask about the person's intent. Give tables six pieces of vague feedback ("You need to be more proactive", "That meeting was a mess"). Each table rewrites two of them in SBI form, and another table judges whether the behaviour is actually observable. Watch for: "behaviours" that are really judgements of character ("you were dismissive"). Push for what someone could have filmed.
7. Triad roleplay with an observer rubric (35 min, two rounds)
How: in groups of three, one person plays the practitioner, one the counterpart and one the observer. Use scenario cards drawn from real situations: a customer who has gone quiet, a stakeholder who disagrees, or a direct report who missed a deadline. Round 1 lasts 6 minutes, followed by 4 minutes of observer feedback using the rubric below. In round 2 the same practitioner replays the same scenario and applies one piece of feedback. Rotate until everyone has played the practitioner if time allows. Otherwise rotate in the follow-up practice. Why two rounds: the second attempt is where learning shows. Without it, a roleplay is just a performance.
8. Commitment and practice plan (15 min)
How: each person writes down one behaviour from the rubric, the next three real conversations where they'll use it, and a practice slot in their calendar for each of the next four weeks. They share it with one partner, who checks in with them at week two.
9. Fishbowl: the difficult conversation (swap-in, 30 min)
How: two volunteers sit in the centre and play a hard conversation (redundancy news, a pricing increase, a complaint). Everyone else observes against one rubric row each. Any observer can tap in and take over the practitioner's seat to try a different approach. Best for: leadership groups and anyone whose mistakes in these conversations are costly.
10. Same message, two channels (swap-in, 20 min)
How: pairs get a sensitive message (a project delay, a declined request). One person writes it as a three-line email. The other delivers it live in 60 seconds. Swap and compare how each was received. Debrief: agree a team rule for which messages must never go by email or chat alone.
The observer rubric
Use the same rubric in the workshop, in follow-up practice and when managers observe real conversations. If you measure with three different instruments, you can't compare the results.
| Behaviour | 1: Not yet | 2: Partly | 3: Consistently |
|---|---|---|---|
| Opens with relevance and a clear purpose | Starts with own agenda | Purpose stated, no relevance to the other person | Purpose and relevance within 90 seconds |
| Listens and confirms understanding | Interrupts or rebuts | Paraphrases sometimes | Paraphrases before responding to key points |
| Asks follow-up questions | Mostly closed questions | Some open questions | Builds follow-ups on the other person's answers |
| Gives specific feedback | Judgements or generalities | Behaviour named, impact missing | Situation, behaviour, impact and intent |
| Agrees clear next steps | Ends without an agreement | Next step vague or one-sided | Owner, action and date agreed out loud |
How to measure outcomes after the workshop
End-of-session feedback forms tell you whether people enjoyed the day. They don't tell you whether anyone communicates differently. The Kirkpatrick Model calls this Level 1, Reaction. What you want is Level 3, Behavior: "the degree to which the target audience performs the critical behaviors" on the job. Kirkpatrick Partners stresses that Level 3 needs an accountability and support package (manager check-ins, observations, job aids) and that you shouldn't wait 90 days to start measuring.
| What to measure | How | When | Owner |
|---|---|---|---|
| Rubric score on real conversations | Manager observes, or reviews recordings of, 3 conversations per person | Baseline, then day 30 and day 90 | Line manager |
| The target behaviour specifically | Score only the rubric row you focused on (for example, openings) | Every observed conversation | Line manager |
| Practice volume | Count of completed practice sessions per person | Weekly | L&D |
| Practice quality | Second-attempt score vs first-attempt score in roleplays | Weekly | L&D |
| Business indicator (Level 4) | One metric you already track: meeting-to-next-step rate, complaint escalations, deal-stage progression | Quarterly | Programme sponsor |
Three rules make the numbers trustworthy:
- Take the baseline before the workshop. Without one, you can't show change.
- Score the behaviour, not the person's overall impression. "Opened with relevance: 1, 2 or 3" is reliable. "Good communicator: 7/10" isn't.
- Expect scores to stay flat until someone coaches against them. That's what the pilot above showed. Plan manager 1:1s that walk through actual scores and actual quotes from conversations.
Where AI roleplay fits between sessions
The workshop gives each person two, maybe three, attempts at the target behaviour. Changing a habit takes many more, spread over weeks, and running more live roleplay sessions is expensive and hard to schedule.
That's where AI roleplay helps. It lets each person do a short practice rep on their own schedule, against the same scenario and the same rubric:
- Weeks 1–4 after the workshop: a 10-minute roleplay once a week on the behaviour each person committed to. Build the scenario from the cards you used on the day, or from a real conversation that went badly.
- Score with the workshop rubric so practice scores, workshop scores and real-conversation scores are directly comparable.
- Managers review the scores, not the transcripts. Use them to pick what to coach in the next 1:1. If practice turns into surveillance, people stop practising honestly.
- Keep the live workshop for what AI can't do well: group norms, peer feedback, and the emotional weight of practising in front of colleagues.
In Shiken, the AI roleplay builds practice scenarios from real calls or documents, scores them against a sales framework such as MEDDIC or SPIN or against your own rubric, and works in 40+ languages (roleplay and coaching). Other tools do this too. We compare them in AI roleplay platforms. For ready-made scenario ideas, see 12 roleplay coaching scenarios. If you already record calls, turning call recordings into coaching shows how to close the loop.
What usually goes wrong
- Too many skills. Six topics in half a day means none of them is practised twice. Pick two or three.
- Roleplays with no second round. Feedback with no chance to apply it is just criticism.
- Generic scenarios. "A difficult customer" teaches less than the exact objection your team heard last week.
- No manager in the loop. If managers don't know the rubric, they can't reinforce it, and Level 3 change rarely happens.
- Measuring attendance. Completion rates show who sat in the room, not who changed.
Run the half day, then put the weekly practice slots and the 30-day observation into calendars before people leave the room. If those follow-up dates exist, the workshop is the start of a programme. If they don't, it's a one-off event.
