The Sales Manager’s First 90 Days With AI Coaching Data
Your company just deployed AI coaching. Every call is now recorded, transcribed, and scored against your sales methodology. A dashboard shows coaching scores for each rep on every conversation. You have more data about your team’s selling behavior than you have ever had. And you have no idea what to do with it first.
The temptation is to open every scorecard, identify every gap, and start coaching everything immediately. That is the fastest way to overwhelm your reps, erode their trust in the new system, and burn yourself out in the first month. The managers who succeed with AI coaching data follow a deliberate 90-day sequence: observe first, then coach selectively, then run the full cadence once the foundation is built.
This guide is for the frontline sales manager who just received AI coaching data for the first time. Not the RevOps team deploying the platform and not the enablement leader measuring the program. You. The manager who has a dashboard full of scores and a team that is watching to see how you use them.
Weeks 1-4: Observe and Build Trust
The Goal: Understand the Data Before Acting on It
Your first instinct will be wrong. You will see a rep with a 38% coaching score and want to pull them aside immediately. Do not. That 38% is a week-one baseline, not a diagnosis. You do not know yet whether the scoring criteria are calibrated correctly for your team. You do not know whether 38% is genuinely low or whether the entire team averages 42%. You do not know whether that rep’s low score reflects a skill gap or a call type that the scorecard was not designed for (a brief follow-up call scored against full discovery criteria).
Spend the first four weeks watching the data rather than acting on it.
What to Do
Learn the scoring criteria. Open the scorecard configuration and understand what each criterion measures, what “met” looks like versus “not met,” and how the criteria are weighted by call type and deal stage. If you do not understand the scoring, you cannot coach from it credibly. Your reps will test you. “Why did I score low on decision process when I asked about their timeline?” If you cannot explain the difference between asking about timeline (not the same as decision process) and asking about how they evaluate and approve purchases (decision process), you lose credibility on day one.
Establish your team’s baseline. After 2 to 3 weeks of scored calls, pull the team average by criterion. You will see patterns: the team as a whole might score 70% on pain identification and 35% on economic buyer engagement. That team-level pattern tells you more than any individual score because it reveals the systemic gap that training addressed poorly or never addressed at all.
Listen to 3 to 5 scored calls per rep. Not to coach. To calibrate your own understanding of how the scores map to actual conversations. Listen to a call that scored 85% and one that scored 40% from the same rep. Hear the difference. Understand what the scorecard is measuring by experiencing it alongside the audio. This calibration is essential because you will be coaching from scores soon and you need to trust that the scores reflect reality.
Flag scoring issues early. If a criterion is scoring every call as “not met” across the entire team, the criterion may be poorly defined rather than universally missed. If brief follow-up calls are being scored against full discovery criteria and dragging down averages, the call type filters may need adjustment. Surface these issues to your RevOps or enablement team now so the scoring is clean before you start coaching from it.
What NOT to Do
Do not coach from the data yet. Coaching from unvalidated scores in week one produces feedback the rep can dispute (“that score is wrong, I did cover the decision process”). Wait until you trust the scores and can defend them with specific call examples.
Do not share individual scores publicly. No leaderboards, no team-wide score comparisons, no “Sarah scored 82% and Alex scored 41%.” Public scores in the first month create fear rather than development. Reps will associate the new system with surveillance rather than coaching. You can introduce transparency later once trust is established.
Do not use scores for performance management. This is the most important rule in the first 90 days. If a low coaching score triggers a PIP conversation or appears in a performance review during the first quarter, every rep will learn to game the system or resist recording rather than engaging authentically with the coaching process. Scores are for development, not discipline. State this clearly and mean it.
Weeks 5-8: Start Coaching One Behavior at a Time
The Goal: Build the Coaching Rhythm With Low Stakes
By week 5, you have a validated baseline, you trust the scoring, and you understand what each criterion measures. Now you start coaching, but selectively.
What to Do
Pick one criterion per rep. Review each rep’s trailing 4-week average by criterion. Identify the single criterion where they score lowest consistently. Not the three lowest. Not all the gaps. One. A rep who needs to improve on economic buyer, decision process, and champion development works on economic buyer this week and the others in subsequent weeks. Coaching one behavior at a time produces faster improvement than coaching five simultaneously because the rep can focus their attention on one change.
Pull a specific call moment for each coaching conversation. Do not say “your economic buyer scores are low.” Say “on Tuesday’s call with Acme at the 8-minute mark, the prospect mentioned their CFO needs to approve this. You acknowledged it and moved on. Listen to the segment. Next time, that is the moment to ask: ‘Would it make sense to include your CFO in our next conversation so we can address any budget questions directly?'” Conversation intelligence that makes every call searchable lets you find this moment in minutes rather than listening to the full recording.
Run 20-minute coaching sessions, not 60-minute reviews. The coaching conversation has three steps: play the specific call segment (2 minutes), discuss what happened and what could improve (8 minutes), roleplay the improved approach (5 minutes), and agree on one thing to practice this week (5 minutes). That is 20 minutes. If the session runs longer, you are coaching too many things at once.
Start with 3 reps per week. You probably have 8 to 12 reps. Coaching all of them in week 5 is not sustainable. Start with the 3 reps who have the largest gap between their score and the team average on their weakest criterion. Coach those 3 this week. Coach 3 different reps next week. Within two weeks, every rep has had one data-driven coaching session. That is the cadence you are building toward.
What NOT to Do
Do not coach every call. A rep who makes 30 calls per week and receives coaching feedback on 10 of them feels surveilled, not coached. Pick the one or two calls per rep that illustrate the criterion you are coaching. Ignore the rest this week.
Do not lead with the score. “You scored 38% this week” puts the rep on defense before the conversation starts. “I listened to your Tuesday call and heard a moment I want to work through with you” opens the conversation without judgment. The score is evidence you use to identify the coaching target. It is not the opening line of the coaching conversation.
Do not skip the roleplay. The most common shortcut managers take is discussing the improvement without practicing it. Discussion produces understanding. Practice produces behavior change. Five minutes of roleplay after every coaching conversation is what turns insight into habit.
Weeks 9-12: Full Coaching Cadence and First Results
The Goal: Run the System at Full Speed and Measure Impact
By week 9, you have coached every rep at least twice. You have a rhythm. The reps know what to expect. The scores are calibrated and trusted. Now you run the full coaching cadence.
The Weekly Cadence
Monday: Team scan (15 minutes). Review team-level coaching scores from the previous week. Identify the 3 reps with the lowest scores or the biggest score declines. These are your coaching priorities for the week. Also note any reps whose scores improved significantly so you can recognize them.
Tuesday through Thursday: Individual sessions (20 minutes each, 3 per week). One criterion per rep per session. One specific call moment. One roleplay practice. One commitment to try the improved approach on this week’s calls. Three reps coached per week means every rep on a 12-person team receives a data-driven coaching session every four weeks at minimum, which is 3x to 4x more than most managers deliver without the data.
Friday: Progress check (10 minutes). Pull this week’s scores for the 3 reps you coached. Did the coached criterion improve? If yes, recognize it Monday. If flat, continue coaching the same criterion next cycle. If declined, ask whether the rep practiced and adjust the coaching approach.
Measuring Your First Results
At the 90-day mark, pull these comparisons:
Team coaching score trajectory. Plot the team average by week from week 1 to week 12. The line should trend upward, especially on the criteria you coached most actively. If it is flat, the coaching is not translating to behavior change and you need to adjust your approach (more roleplay, different criteria, more frequent sessions).
Individual rep improvement. For each rep, compare their week 1-2 baseline to their week 10-12 average. Which reps improved most? Which improved least? The reps who improved most validate your coaching approach. The reps who improved least need a different approach, not more of the same one.
Coached criterion vs. uncoached criterion. Compare score improvement on the criteria you actively coached versus criteria you did not. If coached criteria improved and uncoached criteria stayed flat, your coaching is producing targeted improvement. If everything improved equally, the improvement may be driven by general awareness of being scored rather than your specific coaching. Both are positive outcomes but they tell you different things about what is working.
Early deal outcome signals. If 90 days spans enough deal cycles, compare win rates on deals where the rep scored above 70% on methodology versus deals where they scored below 50%. This correlation is the strongest evidence that the coaching produces revenue outcomes, not just score improvement. If the correlation exists, present it to leadership as proof the program is working.
The Mistakes Every New Manager Makes (and How to Recover)
“I tried to coach everything at once and my reps shut down.” This happens in week 2 or 3 when the data reveals gaps you feel urgent about. The recovery is simple: pick one criterion per rep per session going forward and tell the rep you are narrowing the focus. “I threw too much at you last week. This week we are working on one thing: how you ask about the decision process.”
“I used the scores in a performance conversation and now reps do not trust the system.” This is the hardest mistake to recover from. The repair requires a direct, honest statement to the team: “I made an error using coaching scores in a performance context. Coaching scores are for development. I will not use them for performance evaluation.” Then follow through consistently for 60+ days. Trust rebuilds slowly.
“My top performer scored low and pushed back hard.” Top performers sometimes score lower than expected because they have developed instinctive techniques that do not align perfectly with the methodology as written. The coaching conversation is different for top performers: “Your approach clearly works. The score highlights where your approach diverges from the methodology. Let us look at whether those divergences are intentional strengths or blind spots.” Sometimes the top performer teaches you something the methodology missed. Sometimes the score reveals a gap even the top performer did not know they had.
“I do not have time for 3 coaching sessions per week.” You do. Three 20-minute sessions is one hour per week. If you are spending more than one hour per week in status meetings that could be replaced by a signal-first pipeline review, reclaim that time for coaching. Coaching is the highest-ROI activity a sales manager can do. Status meetings are the lowest. Trade the low-value meeting for the high-value one.
Frequently Asked Questions
How long before I am comfortable with the coaching data?
Most managers feel confident interpreting scores and coaching from specific call moments by week 4 to 5. The first two weeks feel overwhelming because the data volume is high and the scoring is unfamiliar. By week 3, you recognize patterns. By week 5, you can pull a coaching moment and run a session without preparation stress. The comfort comes from repetition, not from studying the dashboard.
Should I share coaching scores with my reps?
Not in weeks 1-4. Start sharing individual scores in the context of 1:1 coaching sessions during weeks 5-8, framed as development data. “Your economic buyer score has been averaging 35%. Let us work on that together.” Team-level averages can be shared in team meetings once trust is established (usually by week 8-10). Individual leaderboards should wait until the team has experienced coaching scores as developmental rather than punitive.
What if a rep’s scores do not improve after coaching?
Diagnose in order: Are they practicing the coached behavior on calls (listen to post-coaching calls to verify)? Is the coaching specific enough (one behavior with a call example, not general feedback)? Is the criterion well-defined (could the scoring itself be the issue)? Are they engaged in coaching sessions (asking questions, roleplaying actively) or passively nodding? Flat scores after 6 weeks of active coaching usually indicate the coaching approach needs to change, not that the rep cannot improve.
How do I handle reps who say “the scores are wrong”?
Take it seriously. Listen to the call they are questioning alongside the scorecard. If the score is genuinely inaccurate (the rep did cover the criterion but the AI missed it), flag the calibration issue with your RevOps team. If the score is accurate and the rep is defending their approach, play the specific moment: “Listen to the segment at 7:20. The prospect asked about approval steps and you moved to pricing. The score flagged decision process as not covered because the follow-up question was not asked.” Specificity resolves most disputes.
What is the single most important thing to get right in the first 90 days?
Trust. If your reps trust that coaching scores are for development and not discipline, they will engage with the process, practice the coached behaviors, and improve. If they believe the scores will be used against them, they will game the system, resist recording, and disengage from coaching. Every decision you make in the first 90 days should be evaluated against one question: does this build trust or erode it?
Conclusion
The first 90 days with AI coaching data are not about coaching as much as possible. They are about building the foundation that makes coaching sustainable: trusting the data, establishing the cadence, earning your team’s confidence that the scores are for their development, and learning how to translate a number on a dashboard into a 20-minute conversation that changes how a rep sells.
Observe for four weeks. Coach one behavior at a time for four weeks. Run the full cadence for four weeks. By day 90, you will have a validated baseline, a proven coaching rhythm, measurable score improvement across your team, and the early evidence that coaching from data produces results that coaching from memory never could. That is the foundation. Everything after day 90 builds on it.