The RevOps Checklist for Deploying AI Sales Coaching in Salesforce

The budget is approved. The Salesforce admin has configured the permissions, tested the integrations, and deployed the platform in a sandbox. Now the project lands back on your desk as the RevOps leader who has to make sure the deployment actually changes how your team coaches, sells, and operates. Because the tool working technically and the tool working operationally are two very different outcomes.

Most AI coaching deployments that fail do not fail because of technology. They fail because nobody designed the coaching process the tool is supposed to support. Managers do not know what to do with the data. Reps experience the tool as surveillance rather than support. Success metrics are defined retroactively rather than upfront. And the RevOps team spends months answering “is this thing working?” without having established what “working” looks like before launch.

This checklist covers the operational deployment that RevOps owns: designing the coaching cadence, enabling managers to use the data, managing the change for reps, defining success metrics at 30, 60, and 90 days, and building the optimization loop that makes the tool more effective over time. The technical Salesforce preparation is a separate workstream. This is the process and people workstream that determines whether the investment produces results or becomes expensive shelfware.

Phase 1: Pre-Launch Process Design (Weeks 1-2)

Define the Coaching Cadence Before the Tool Goes Live

The number one mistake in AI coaching deployment is turning on the tool and assuming managers will figure out how to use the data. They will not. Most frontline sales managers have never coached from systematic call data. They have coached from the handful of calls they overheard, their gut feeling about each rep, and whatever came up in the last pipeline review.

Before launch, design the specific coaching cadence managers will follow. The coaching with conversation intelligence guide provides the full framework. At minimum, define these four elements:

Weekly team scan (15 minutes, Monday). The manager reviews team-level coaching scores from the previous week, identifies the three reps with the lowest scores or the biggest score declines, and sets their coaching priorities for the week.

Individual coaching sessions (20 minutes each, Tuesday through Thursday). The manager coaches one specific behavior per rep per session using a call segment from the previous week. One behavior. One call. One roleplay practice. One tracking commitment.

Friday progress check (10 minutes). The manager pulls coaching scores for the three reps they coached and compares this week to last week. Score improved? Acknowledge it. Score flat? Continue coaching the same behavior next week. Score declined? Adjust the approach.

Monthly coaching review (30 minutes). The manager reviews 30-day trends, identifies which reps have improved sustainably, which need continued focus, and which new skill gaps have emerged. This feeds into the next month’s coaching plan.

Document this cadence. Train managers on it. Make it the standard operating procedure before the tool goes live. Without it, the tool generates data that nobody acts on systematically.

Choose Your Scoring Methodology

AI coaching scores are only valuable if they measure criteria that matter for your sales motion. Before launch, define what the AI scorecards will evaluate.

If your team runs MEDDIC, configure the scorecards to evaluate the six MEDDIC criteria with stage-appropriate weighting. If your team uses BANT, Challenger, or a custom methodology, configure the scorecards accordingly. If your team does not have a defined methodology, this is the moment to choose one. Deploying AI scoring without a defined methodology produces generic scores that managers cannot translate into specific coaching actions.

Also define which call types get scored. Discovery calls, demos, proposal presentations, and negotiation calls should all be scored. Administrative calls, brief follow-ups, voicemails, and internal calls should be excluded. Scoring the wrong call types creates noise that dilutes the coaching signal.

Establish Baseline Metrics

You cannot prove the deployment worked if you do not know where you started. Before the tool goes live, pull and document these baseline metrics:

Win rate by rep and by team. This is the ultimate outcome metric. Snapshot it by rep for the trailing 90 days.

Average deal cycle length. Snapshot by rep and by deal segment. AI coaching should compress cycles over time as reps execute methodology more consistently.

Ramp time for recent hires. Weeks from start date to first closed deal for hires in the last 12 months. This is the baseline against which you will measure coached new hires.

Forecast accuracy. Trailing four-quarter variance between forecast and actual. AI coaching that improves deal visibility through conversation intelligence should narrow this variance as deal health assessments become based on engagement data rather than rep self-reporting.

Manager coaching frequency. How many coaching conversations does each manager have per week currently? Most managers report 1 to 2 per week, often unstructured. This baseline shows how much the coaching cadence changes post-deployment.

Phase 2: Manager Enablement (Week 2-3)

Managers are the deployment’s success or failure point. If managers adopt the coaching cadence and use the data, the tool produces results. If managers ignore the data and coach the same way they always have, the tool is shelfware. Manager enablement is not a training session. It is a multi-week process that builds comfort and competence with the new workflow.

Session 1: The Data Walk-Through

Walk each manager through the dashboards, scorecards, and reports they will use weekly. Show them where to find team-level scores, individual rep scores, score trends, and specific call segments. Do not show them every feature. Show them the five views they will use every Monday morning and nothing else. Overloading managers with capabilities they will not use in the first month creates confusion rather than confidence.

Session 2: The Coached Coaching Session

Sit with each manager and run the first coaching conversation together. The RevOps leader (or enablement partner) observes while the manager pulls a rep’s data, identifies the coaching target, selects a call segment, and runs the 20-minute session. Provide feedback after: “You identified the right criterion but spent too long on the call segment. Next time, play 60 seconds, not 4 minutes.” This coached practice builds the manager’s skill with the new cadence before they are on their own.

Session 3: The Independent Run

The manager runs a coaching session independently and reports back on how it went. Review the session structure, the coaching target they chose, and whether the rep responded positively. Course-correct any issues. After this session, the manager should be comfortable running the weekly cadence without support.

Ongoing Manager Support

Schedule a 30-minute weekly check-in with all frontline managers for the first 60 days. Review adoption metrics (are they running the coaching cadence?), share best practices across managers, and troubleshoot any issues. This standing meeting can be dropped after 60 days when the cadence is established, but skipping it during the first two months is the most common cause of manager disengagement.

Phase 3: Rep Communication and Change Management (Week 3)

Reps will have three concerns about AI coaching. Address all three proactively before launch.

Concern 1: “Is This Surveillance?”

This is the most important concern to address and the one most deployments handle poorly. If reps believe coaching scores will be used for performance management, PIPs, or termination decisions, they will game the system, avoid being recorded, or disengage entirely.

What to communicate: “Coaching scores are for development, not discipline. Low scores trigger coaching conversations, not performance reviews. Your manager will use scores to identify specific skills to work on together, not to build a case against you.” If leadership is genuinely committed to this principle, state it clearly from the CRO. If leadership is not committed, the deployment will face adoption resistance that no communication can overcome.

Concern 2: “This Is More Work for Me”

Frame the deployment as work removed, not work added. Automatic activity capture eliminates 1 to 2 hours of daily manual logging. AI-generated call summaries eliminate note-taking. Guided selling recommendations eliminate the planning time reps spend deciding who to call and what to do next. The net effect is reps gaining time, not losing it. Lead with what is removed before describing what is added.

Concern 3: “My Numbers Are Fine, I Don’t Need Coaching”

Top performers often resist coaching tools because they believe (sometimes correctly) that their performance speaks for itself. Address this by framing the tool as serving them too: their best calls become teaching examples for the team, their coaching scores validate what they do well, and real-time coaching prompts give them competitive battlecards and methodology reminders that even experienced reps benefit from during complex conversations.

Phase 4: Launch and 30-Day Assessment

Week 1 Post-Launch

Monitor adoption daily during the first week. Key metrics to watch: percentage of calls being recorded and scored (target 90%+), manager logins to coaching dashboards (every manager should access the dashboard at least twice in week one), and rep feedback (surface any technical issues or user experience complaints immediately).

Expect a calibration period. The first week’s coaching scores should not be used for coaching conversations. They are the baseline. Managers and reps should observe the scores, note any that seem inaccurate, and flag calibration issues (criteria that are scored too harshly or too leniently). RevOps adjusts scoring criteria based on calibration feedback before week three.

Weeks 2-4

Managers begin the coaching cadence. RevOps monitors whether each manager is running the Monday scan, the three coaching sessions, and the Friday progress check. Track this through dashboard login data, coaching conversation notes in Salesforce, and the weekly manager check-in.

30-Day assessment metrics:

Call scoring coverage (percentage of eligible calls scored, target 90%+). Manager coaching cadence adherence (are all managers running 3+ coaching sessions per week?). Rep coaching score trends (are bottom-half reps showing score improvement?). Activity capture rate (is automatic logging producing complete data?). Qualitative feedback from managers and reps (what is working, what is frustrating?).

If scoring coverage is below 80%, there is a technical or configuration issue. If manager cadence adherence is below 70%, there is an enablement issue. If rep scores are flat after 30 days with active coaching, there is a scoring calibration issue. Diagnose and fix before proceeding.

Phase 5: 60-Day Optimization

By day 60, the initial deployment issues should be resolved and the coaching cadence should be running consistently. Now the focus shifts from adoption to optimization.

Refine scoring criteria. Review the criteria with managers. Are any criteria consistently scored as “not met” because the criteria are poorly defined rather than because reps are underperforming? Are any criteria consistently scored as “met” on every call, providing no coaching signal? Adjust criteria to ensure they produce meaningful variance that managers can coach against.

Run the first correlation analysis. Pull 60 days of coaching scores alongside deal outcomes. Do deals where reps scored above 70% on methodology adherence close at higher rates than deals where reps scored below 50%? If yes, share this data with the team. It is the most powerful adoption reinforcement available because it proves the methodology (and the coaching based on it) actually drives revenue. If the correlation is weak, the scoring criteria may need adjustment or the methodology itself may need reinforcement through training.

Identify team-wide gaps. If 70% of the team scores below 50% on a specific criterion (like quantifying business impact), that is not a coaching problem. It is a training problem. Schedule a team-level skill session on that specific topic. Use the highest-scored call in the dataset as the teaching example.

Phase 6: 90-Day Business Review

At 90 days, you have enough data to evaluate whether the deployment is producing the outcomes that justified the business case. Pull these metrics and compare to the baselines you established before launch.

Win rate change. Compare pilot group (or full team) win rate for the 90-day post-launch period against the 90-day pre-launch baseline. Even a 2 to 3 percentage point improvement on meaningful pipeline is significant.

Coaching score improvement. Show the average score trajectory from week 1 to week 12. The trend should be upward, especially for bottom-half performers. Flat scores after 90 days of active coaching indicate a process problem (managers not coaching effectively) or a calibration problem (criteria not producing actionable scores).

Ramp time comparison. If any new hires joined during the 90-day period, compare their time to first coaching score above 70% and time to first deal against the pre-launch baseline cohort.

Manager coaching frequency. Compare post-launch coaching session frequency to the pre-launch baseline. The cadence should show at least 3x increase in structured coaching conversations per manager per week.

Forecast accuracy. If 90 days spans at least one full quarter, compare forecast variance to the trailing baseline. Improvement may be modest at 90 days but should be directionally positive.

Present these results to the CRO and CFO alongside the original business case projections. Show which scenarios (conservative, moderate, optimistic) the actual results are tracking against. If results are tracking at or above the conservative scenario, the deployment is on track. If below, diagnose whether the issue is adoption, calibration, or timeline (some impacts take 2 to 3 quarters to fully materialize).

Frequently Asked Questions

What is the biggest risk in an AI coaching deployment?

Manager non-adoption. If frontline managers do not use the coaching data systematically, the tool produces scores that nobody acts on. The data is generated. The behavior change is not. Manager enablement (Sessions 1-3 plus 60 days of weekly support) is the highest-leverage investment in the deployment because manager behavior determines whether coaching scores become skill development or digital noise.

How do I handle rep resistance to call recording and scoring?

Address the surveillance concern directly and early. State clearly that coaching scores are for development, not discipline. Lead with what the tool removes (manual logging, note-taking) before what it adds (recording, scoring). And show top performers that their best calls become team assets rather than lost conversations. Resistance almost always decreases after 2 to 3 weeks once reps experience the time savings and see that coaching conversations are genuinely developmental.

How long until we see revenue impact?

Coaching score improvements: 2 to 4 weeks. Behavioral changes visible on calls: 4 to 8 weeks. Deal outcome impact (win rate, cycle length): 8 to 12 weeks. Full impact at scale: 2 to 3 quarters. Set these expectations with leadership before launch so the deployment is not judged against unrealistic timelines.

Should we deploy to the full team or start with a pilot?

Start with a pilot of 10 to 15 reps if leadership needs proof before committing fully, or if your RevOps team cannot support full-team enablement simultaneously. Deploy to the full team if the business case is already approved, manager enablement capacity exists, and the organization is committed. Pilots produce proof. Full deployments produce impact faster. Choose based on your organization’s risk tolerance and readiness.

What if coaching scores do not improve after 60 days?

Diagnose in this order: Are managers running the coaching cadence consistently (adoption issue)? Are the scoring criteria well-defined and producing meaningful variance (calibration issue)? Are managers coaching one specific behavior per session or giving generic feedback (coaching quality issue)? Are reps practicing the coached behavior or reverting to habits (reinforcement issue)? The answer is almost never “the tool does not work.” It is almost always one of these four process issues.

Conclusion

An AI coaching tool that is deployed technically but not operationally produces dashboards full of data that nobody uses. The technology works. The results do not appear. The difference is the operational deployment that RevOps owns: designing the coaching cadence before launch, enabling managers through coached practice sessions, communicating change to reps in terms of value rather than monitoring, measuring success at 30, 60, and 90 days against pre-defined baselines, and optimizing scoring criteria and coaching processes based on what the data reveals.

Follow this checklist in sequence. Do not skip the manager enablement. Do not launch without baselines. Do not wait until month three to define what success looks like. The RevOps team that designs the operational deployment with the same rigor as the technical deployment is the one that delivers the ROI the business case promised.