Why is cutting sales managers for AI coaching risky?
Managers move a team's output, so cutting the wrong ones is expensive. Economist Edward Lazear and two colleagues studied technology-based services workers at one large company. Moving a team from a boss in the bottom 10% for quality to one in the top 10% lifted its output more than one extra worker would on a team of nine. Workers with better bosses were also less likely to leave. The study was published in 2015 and draws on one firm's data.
The pull toward cutting is strong. In October 2024, Gartner predicted that 20% of organizations will use AI to flatten their structure by the end of 2026, "eliminating more than half of current middle management positions." That is an analyst forecast. The release also warned that "mentoring and learning pathways may become broken." On a sales team, many of those pathways run through the front-line manager.
What questions should you ask before AI replaces sales manager coaching?
Ask these before anyone redraws the manager role. Each one tests what the tool leaves for a person to do.
-
What does the AI coach cover, and what does it miss?
An AI coach reviews what gets recorded. In commercial services, much of the selling happens on a site walk or in a facilities director's office, and the tool hears none of it. In a 2021 field experiment with 265 employees in a loan-collection call center, Siliang Tong and colleagues found that AI feedback flagged more mistakes than managers did, and employees who received it averaged 12.9% higher daily collections over 30 days. The authors limit that result to jobs with clear goals and known good and bad scripts. Listen for a plain map of the conversations the tool sees and the ones it never will.
-
Who acts on what it finds?
A finding with no owner is a report. A tool can flag a missed renewal date or a rushed price conversation, and someone still has to decide whether it matters and talk it through with the rep. Listen for a named person behind each kind of finding, and for what that person does when it lands.
-
What does a rep do differently after AI feedback, and who checks?
Feedback has done its work when a rep handles the next customer differently. A tool can show that a rep received a note. It has no view of the next renewal if that renewal happens on site. Listen for someone who watched the change happen with a customer present and can describe it in detail.
-
Which manager conversations would no tool hold?
Some conversations decide whether a seller stays and grows. Tong and colleagues name the managerial work their finding does not reach, including unstructured conversations about promotion and performance improvement, and the personal connection that motivates people. In a study of a large high-tech firm published in 2021, Mitchell Hoffman and Steven Tadelis found that managers' people skills had a strong negative relation to employee turnover. Those skills did not consistently improve the other outcomes the authors measured. Listen for the conversations your managers name first, such as a seller weighing a competitor's offer or a renewal concession that needs judgment.
-
What will we ask managers to do with the time?
A tool brings new work along with the hours it saves. In a working paper revised in March 2026, Anders Humlum and Emilie Vestergaard linked adoption surveys to Danish labor records and found employers absorbing AI chatbots through task reorganization, including new tasks in AI oversight. Gartner's 2024 forecast pictures the remaining managers with a wider span of control, each leading more people. A deeper conversation with each rep is the other option. Listen for which one the plan chose and who chose it.
"If AI handled every conversation review, what would my managers be for?"
It sits under the other five because each of them asks what the tool leaves for a person. A manager whose week goes to finding problems has less to do once a tool finds them. A manager whose week goes to the conversations that fix them, and to judgment about people and the team's number, keeps a full job.
What are the signs AI coaching still needs a manager beside it?
Good signs
- Managers can name the customer conversations the tool never sees.
- Each kind of finding has a named person who decides what happens next.
- Reps describe a change they made and the manager who talked it through with them.
- Someone has watched a rep use the change with a customer.
- The plan for managers' returned hours was written down before the tool went live.
Warning signs
- Feedback goes straight from the tool to reps, and nobody follows up.
- Managers point to a dashboard when asked what a rep now does differently.
- The headcount plan cut managers before anyone mapped what the tool covers.
- Reps hear from a colleague that a tool reviews their meetings.
- Retention and promotion conversations have no owner.
Case study Fictitious numbers
A PE-backed commercial grounds maintenance company has 20 sellers under three managers. The sellers renew annual maintenance agreements and sell enhancement work such as irrigation upgrades and tree care. The company adds an AI coach that reviews recorded customer meetings and sends each seller written feedback.
By the end of the quarter, the tool has flagged the same gap for 12 sellers. In renewal meetings, they confirm the new price and move on before the property manager can raise what went wrong last season. Managers sit down with 5 of the 12 to talk it through. The other 7 receive only the written feedback.
The renewal meetings that follow happen on site, unrecorded, so the VP of sales joins them to see what changed. Of the 5 sellers who talked it through with a manager, 4 now walk the property with the customer and ask about last season before they discuss price. Of the 7 who received only the written feedback, 1 does. That is 5 changes among 12 sellers.
- Changed how the meeting ran
- No change
Twelve sellers is a small group, and the managers may have picked the five they expected to change. The VP treats it as a question.
Twelve sellers is a small group, and the managers may have picked the five they expected to change. The VP treats the result as a question. The next conversation with the three managers is about who owns the other 7, and what each manager will stop doing to make room.
FAQ
Will AI reduce the number of sales manager jobs?
The projections disagree. The U.S. Bureau of Labor Statistics projects employment of sales managers to grow 4% from 2025 to 2035, about as fast as the 3% average for all occupations. Gartner's 2024 prediction points the other way for middle managers in general. Both are forecasts. A working paper using Danish labor records found precise null effects of AI chatbots on earnings and recorded hours two years after ChatGPT launched, even as tasks inside jobs changed.
Has any company tried running without managers?
Google did, briefly. Harvard Business Review reported in 2013 that in 2002 the founders removed engineering managers to create a flat organization. The experiment lasted only a few months and ended when too many people took expense questions and interpersonal conflicts straight to Larry Page. Google's later Project Oxygen research set out to show that managers did not matter and concluded they did. That was an engineering organization, years before today's AI tools.
Should reps know when an AI reviews their conversations?
Yes, and how you tell them matters. In the 2021 loan-collection experiment, employees told their feedback came from AI performed 5.4% worse than employees told it came from a manager, whatever the true source. They reported lower trust in the feedback and more worry about being replaced. The authors recommend telling employees up front what the tool is for and how far it reaches. Do that before it goes live, in words a seller could repeat.
What is the difference between conversation intelligence and an AI coach?
Conversation intelligence records and transcribes customer conversations, then surfaces patterns across them. An AI coach goes a step further and turns those patterns into feedback for each rep. Both depend on what gets recorded. Deciding which change matters most for a given rep, and holding the conversation that makes it stick, stays with the front-line manager. That is where sales coaching happens.
Sources
- The Value of Bosses, Lazear, Shaw and Stanton, NBER Working Paper 18317 (Journal of Labor Economics, 2015) checked 10 Oct 2026
- Gartner Unveils Top Predictions for IT Organizations and Users in 2025 and Beyond (Gartner, analyst predictions, 22 Oct 2024) checked 10 Oct 2026
- The Janus face of artificial intelligence feedback, Tong, Jia, Luo and Fang, Strategic Management Journal, 2021 (field experiment) checked 10 Oct 2026
- People Management Skills, Employee Attrition, and Manager Rewards, Hoffman and Tadelis, NBER Working Paper 24360 (Journal of Political Economy, 2021) checked 10 Oct 2026
- Still Waters, Rapid Currents: Early Labor Market Transformation under Generative AI, Humlum and Vestergaard, NBER Working Paper 33777 (revised March 2026) checked 10 Oct 2026
- Occupational Outlook Handbook, Sales Managers, U.S. Bureau of Labor Statistics (2025 to 2035 projections) checked 10 Oct 2026
- How Google Sold Its Engineers on Management, Garvin, Harvard Business Review, December 2013 checked 10 Oct 2026