The Manager Multiplier

Train ten reps on a new AI tool and you have made ten interventions. Equip their manager once and you have changed what all ten people see every week, for as long as that manager runs the team. Most AI rollouts fund the first and skip the second, then answer the stall with more training.

In Two Rollouts, One Variable, a vertical-market technology provider and an industrial manufacturer ran substantially the same rollout playbook and got different results, and the variable was the direct manager. That piece showed the pattern. This one is about why that layer carries so much weight, and what it takes to equip one.

The multiplier math

Executive sponsorship is episodic by design. The sponsor appears at the kickoff, signs the mandate, sends the follow-up emails. Each appearance is real, and each one is an event. Between events, nothing.

The manager is in the room every day. The Monday 1:1. The Thursday pipeline review. The hallway conversation about why a deal stalled. Those are the moments where a rep decides what actually counts on this team, and the manager is present for every one of them.

Nobody decides what is mandatory by rereading the announcement. They watch the person who reviews their work. If that person asks to see the AI-assisted call plan, the call plan gets done with the tool. If that person never mentions it, the tool is optional by the second week, whatever the memo said. I care about what my manager cares about. That is the whole mechanism.

Gallup put a number on how much of a team's engagement comes down to the manager back in 2015, and still stands behind it: managers account for at least 70% of the variance in team engagement. That is a finding about engagement, not about AI, and the step from one to the other is ours. If the manager decides most of how engaged a team is, and taking up a new way of working is an engagement problem before it is a skills problem, then the manager is where adoption is decided.

This is not a story about a bad sponsor

The view most rollout budgets are built on goes like this: employee appetite is the barrier, so the fix is tooling, training and an executive sponsor. Fund all three and the rollout is resourced.

That view is wrong on the first clause and incomplete on the rest. Most teams will use a tool that makes their work easier; appetite is rarely the problem. And all three line items can be fully funded while the rollout still stops short, because none of the three touches the work daily.

It is tempting to read the two-rollouts story as one manager who did not try hard enough. It was not hostility toward the initiative. It was daily behaviour that just never showed up. The same pattern recurs across AI rollouts that treat the manager as the last person to hear about the initiative and the first person expected to enforce it. The manager gets a finished thing to approve and a mandate to police. Sequencing went wrong, not effort.

The layer nobody equips

Ask the people whose job it is to know. In a 2025 Gartner survey of 114 HR leaders, 8% believed their managers have the skills to use AI effectively. That is a belief about managers, not a measurement of them, and it comes from the function that owns manager development. The people responsible for building that capability, by an overwhelming margin, do not think it is there yet.

Here is what the standard rollout does with that layer. The manager gets a link to the documentation and a recorded webinar. Then they are expected to lead a behaviour change they have not themselves made, in front of a team that is watching to see whether they made it.

That is not a knock on managers. It is a description of what they were handed, and of how little room they had to do anything with it. A Sales Management Association study across 99 firms, with data from 2015, put the average at 36 minutes of one-on-one coaching per direct report per week for sales managers. Sales management's own view in the same study is that the sales force gets too little coaching: 77% say so, against 15% who say the amount is right. Read those two figures together and the manager who did not carry the rollout looks less like a skeptic and more like someone out of hours.

The training budget usually goes one level down, to the reps, and the case for that is real. London School of Economics research published by Protiviti finds workers with AI training save 11 hours a week against 5 for the untrained, and that 68% of employees have received no AI training in the past 12 months. Training the individual is the barrier most within a leader's control. It is also the ten interventions again, unless something makes it stick. What makes training stick is the week after: whether anyone asks to see the work, whether a weak first output gets diagnosed or quietly abandoned, whether the tool shows up in the pipeline review. The person in the room for that week is the manager. Train the reps and skip the manager, and the 11 hours last exactly as long as the enthusiasm does.

What it takes to equip one

Three things have to be true of a manager before they can carry a rollout, and each one can be checked from outside. Marie Rodgers, contributing author on this piece, helped lead a 30-interview companion benchmark study of B2B revenue leaders. The quotes below highlight three principles for what it takes for managers to be successful change agents.

1. Desire. It looks like showing up in person rather than delegating the kickoff, asking about the tool inside existing routines without being prompted, and treating early friction as a problem to solve rather than a reason to let adoption slide. The founder and CEO of a professional-services consulting firm described the reflex: "Someone will say, oh, that's a pain in the butt, in passing, and then I'll just start up a chat to figure out how to solve it. And then 24, 48 hours later, it's like it's solved forever."

2. Knowledge. It looks like having used the tool on their own real work, and being able to say why a specific output was weak rather than re-forwarding the training deck. A sales leader at a mid-market technology reseller described one of his managers building a small assistant for a single recurring task, the kind of purpose-built helper the industry calls an agent: "One of my managers a few months back created an AI agent specifically to help with renewals analysis. What used to be about an hour long manual analysis, it's now two minutes." Knowledge includes knowing where the tool fails. The founder of an events and conference business: "I was leaning really hard on AI for my last agenda. My timing was screwed up so I didn't check the timing it used, so I didn't edit... I had a lot of mess ups from AI on this last one and I'm a lot more careful now. Much more careful." A manager who has been burned once and checks now is further along than one who has only watched the webinar.

3. Coaching know-how. It looks like coaching tool use in the flow of work rather than in a separate session, turning a rep's "this saved me an hour" into a conversation about what the hour is for, treating adoption as a coachable skill with real feedback, and connecting one person's tool use to the team's numbers. A B2B SaaS sales leader described the shift in their own prep: "I'm going to read a rep's transcript, but I'm going to go to Claude, I'm going to throw in three, four of her past calls at that exact same stage and then this one as well. I'd be like, what are themes?" The coaching conversation that follows is about a pattern across four calls, not an impression from one.

The ADKAR model for change reinforces that the order matters. A manager who is willing but stops at the first of these stalls almost as reliably as one who never engaged. Desire without knowledge produces a manager who endorses the tool at the kickoff and has nothing to say when a rep brings back a weak output. The team reads that silence correctly.

The honest objection is that this bar is high, and the study's own field notes agree. Most of the thirty leaders interviewed are early. A general manager in manufacturing: "So the answer is that I'm only using Copilot currently and I do use it sporadically. I haven't integrated it into my daily routine yet." If the people running revenue teams are there, the objection goes, asking frontline managers to coach the tools is asking too much. It is asking a lot. It is not asking for talent. Every behaviour above is specific enough to put on a checklist, teach in a small group of peers, and check for in a pipeline review. That is the difference between a bar that is high and a bar that is innate, and the case here rests on it being the first.

What ownership looks like across a rollout

Ownership is a set of behaviours placed on the timeline of the rollout. None of it requires charisma.

1. Before the build. The manager maps the team's workflows and supplies the examples the assistants are built from. Gartner finds organizations are 14 times more likely to succeed at a change when employees help build it rather than being told what is coming. The manager inside the build is the shortest route to that.

2. At launch. The manager delivers the "why" in their own words, on camera, to their own team. The reason this team is doing this, in the vocabulary this team already uses for its own numbers, rather than the sponsor's slide forwarded with a note.

3. Every week. The manager asks to see AI-assisted work in 1:1s and pipeline reviews. Unblocks friction the same week it appears. Uses the tools visibly on their own work, so the team sees the behaviour rather than hears about it.

Programs that get this right do the unglamorous version. They map each manager's own workflows before writing a line of training, translate the initiative into that manager's own team numbers, and put managers in small peer groups so the first person to figure out a weak output is not the only one who knows.

Carry the renewals manager from the study through that arc, because the study records the before and after and the rest follows from it. The task was a recurring analysis that took an hour by hand. The manager built an assistant for it and it now takes two minutes. Before any team rollout, that manager can already name the workflow worth doing next, because they have done one. At launch, the "why" is a demo of their own analysis, not a slide. Every week, when a rep's output is weak, the manager can say what went wrong because they have seen their own go wrong. And when the sponsor asks what the reclaimed hour is for, the manager has an answer, because the question was theirs first. What the study does not record is what that manager did with the hour, and that is the question that matters most.

The payoff is the manager's own

So far this reads like a longer to-do list for a layer that is already out of hours. The reason it is not is that AI reclaims manager time too. Nothing in the evidence set measures how much, so treat this as an argument rather than a finding, and treat the renewals manager as its shape: an hour of work became two minutes, and the person who got that hour back was the manager.

Asked what they would do with 20% more time, the most common answer among the thirty revenue leaders in the study was to coach their people more. The appetite for the highest-return activity already exists. Capacity is the constraint, and capacity is exactly what the tool gives back.

Coaching is the highest-return place to put that capacity, and two independent findings point the same way. A Gartner survey of 1,026 B2B sellers found that sellers who effectively partner with AI tools are 3.7 times more likely to meet quota than those who do not. That is a finding about sellers and it is a correlation. The step to the manager is ours: a seller who partners with AI effectively learned to somewhere, and the somewhere is coaching.

Salesforce's own research finds high performers are 1.4x more likely than underperformers to use agents for coaching. That does not prove coaching agents caused the performance. It does show where the teams that are winning chose to point AI.

Put those together and the initiative changes shape for the manager. It stops being an enforcement burden handed down from the kickoff and becomes their own performance lever: the tool gives back hours, the hours go into coaching, and coached reps are the ones who hit quota. That is what makes a manager care, and what a manager cares about is what the team does.

Most leaders in the study are still at the first stop, saving time on single tasks. The ones further along have turned those savings into routines the team runs every week, and further still, direct work where the tool produces the first draft and the team checks it. Where the reclaimed time goes, and how a leader knows whether it turned into anything, is the next question in this series.

Three questions to ask this week

Run this before the next rollout is funded, not after. If the answers send you back to scoping, that is what the three weeks of an AI Jumpstart are for: knowing which workflow is worth it before the spend.

Ask about your own managers right now:

1. Have they used the tool themselves, on their own real work?

2. Do they already ask about it in 1:1s and pipeline reviews, without being reminded?

3. Can they say in one sentence what the freed-up time is for?

A VP who can't answer yes to all three has just found the actual project. It isn't deploying the tool. It's equipping the manager.