Start by asking where better work would matter most to the business. That might be sales follow-up, delivery that keeps falling behind or billing numbers that arrive too late to use. Then find the leader whose decisions shape that work. It may be an executive, a vice president or a department head.

Step 1 showed why one person is often the quickest and least expensive place to start. There is less to set up, and that person’s own work can improve by a larger percentage. When that person is a leader, the gain reaches further, because they set priorities, review the work and decide what gets followed up across many people.

An AI account on its own won’t do that. The leader needs the right tools, set up around their actual work, and needs to learn how to use them: what to ask for, which sources to rely on and how to check an answer before acting on it.

A leader doesn’t change a team just by becoming more productive personally. The change comes when better information, better questions and better use of their time reach the decisions that affect other people. So the setup has to give them:

• The records needed to see what is happening
• The company and team context needed to interpret it
• Prompts or reports that surface relevant signals
• Training on what to ask
• A way to verify the information
• A clear decision or coaching action

AI can prepare the information, organize patterns and suggest where to look. The leader still verifies the situation, decides what matters and coaches the team. Employees still do the work.

How one leader’s gain reaches many people →

Two scopes, two percentages

In the illustrative comparison, a 50% gain for one person adds half that person’s usual output. A 25% gain for a department of ten adds a quarter of the department’s usual output. With comparable starting workloads, the team can add more work overall at the lower rate. That wider rollout takes more time, money and shared infrastructure.

What travels from the leader to the team

Say a department head gets through their reporting and review preparation much faster. That speed belongs to them. Nobody else suddenly works at the same rate, so a leader’s percentage can’t be multiplied across the people they oversee.

What reaches the team is what the leader does with the extra time and the clearer view. A faster task changes one person’s output. A better leadership decision changes what several people work on, how they do it and what standard they are held to. It shows up in four places:

• Priorities. People spend their week on the work that matters most.
• Coaching. Feedback comes from a person’s real work, while it can still help.
• Accountability. Work has an owner and a next step, and someone checks on it.
• Process. What works gets written down and becomes how the team does it.

Which level of leadership to start with

The right level depends on which decision or work is holding the business back.

• Executive team. They decide where to invest next and which part of the business is running out of capacity. They need activity from sales and delivery next to the financials, so they can see their numbers and margins and act on them.
• Vice presidents. They oversee several teams and can be the people who fix a handoff between them. A vice president of operations might need to see whether work that was sold is being scheduled and completed, then coach the managers who run those teams.
• Department heads. They are closest to the daily work. A sales head reviews follow-up. A billing lead sees what is ready to invoice and what is missing.

As an illustration, if proposals keep going quiet in one sales team, the sales head is a natural place to start. If sales is doing well and orders are waiting on delivery, the leader who oversees both may be the better choice.

A practical first setup

The first setup can stay small. It needs four things:

1. Current priorities. What the company or team is trying to get done right now.
2. Approved records. The sales, job or billing records this leader is cleared to use. AI can only work from what it has been given, so someone has to connect or supply these sources and decide who may see what.
3. Company and team context. Who does what, how the work is defined and what good work looks like here.
4. One recurring review. A meeting the leader already holds, or should hold, where a real decision gets made.

Here is an illustrative version. Before a Monday sales review, AI pulls the week’s proposal records into a summary, lists open items with no next step and drafts questions for the meeting. The sales head checks the summary against the records, then decides what comes first, who needs coaching and what gets followed up.

A small trial for each level

These trials are illustrative. Each one tests the habit on real work before anything wider is built.

• Executive. Choose one company priority. Give the executive the relevant records, a short weekly summary and four fixed questions: What moved? What stalled? What is creating the constraint? Which decision belongs to me? The executive verifies the signal with the responsible leader and makes one documented decision.
• Vice president. Choose one recurring process in the function. Compare two teams using the same definitions and records. The vice president finds one difference, asks the managers what caused it and chooses one behavior or process to standardize.
• Department head. Choose one weekly workflow. Give the department head a simple view of activity and stalled work, then require one daily review, one coaching conversation and one process adjustment. Track what changed the following week.

What the leader has to learn

Training covers two skills: getting useful preparation from the tools, and checking it. That means opening the source record behind a claim, noticing what is missing and knowing when the context changes the answer.

After running the review for real, the leader also knows which steps to keep, which tools are worth paying for, what the team will need to be taught and who may need to be hired. That makes a later rollout much easier than trying to work all of this out during the rollout.

More numbers won’t help a leader on their own. They need the few that answer a question they have to act on, and with AI organizing the records they can spot sooner what is moving, what has stalled and who needs help.

Two kinds of measure belong side by side. Activity shows what people did, such as proposals sent, jobs scheduled or invoices issued. Outcomes show what came of it, such as sales closed, work delivered or payments collected. Activity alone can make a busy team look like a productive one. Outcomes alone can arrive too late to coach from, so read both together.

So which measures deserve a place on a leader’s screen?

Choosing what to track →

Start from the decision

For each measure, ask what the leader would do differently depending on what it says. If the answer is nothing, leave it out. These illustrative examples show how a measure connects to a decision:

• Sales. Proposals sent and follow-ups due show the activity. Opportunities and closed sales show the outcome. Together they help decide who needs coaching and where the team should focus.
• Operations. Jobs started and jobs waiting show the activity. Work delivered and what it cost show the outcome. Together they help decide whether capacity is running out.
• Billing. Completed work not yet invoiced and items missing from invoices show the activity. Invoices paid show the outcome. Together they help decide what to chase first and whether the numbers are ready for leadership to use.

What each level looks at

A department head needs the team’s numbers and the individual records beneath them, because that is where coaching happens. A vice president needs to compare teams and see where work is waiting between them. An executive team needs those measures alongside revenue, costs and margins, so it can decide where to invest next.

What has to be underneath

The order matters. First, structure the data. Next, put it into reports and visuals people can understand. Then build tools that let each department use it in daily work. Each record should show who owns the work, what was done and when, where it stands now and what happens next. Without that foundation, an AI summary is working from guesses and gaps.

Who keeps the view honest

AI can gather the records, build the view and flag gaps. The leader chooses the measures and decides what each signal means. Employees record their work accurately and correct what is wrong. A missing entry might mean the work wasn’t done, or that it was done and never written down, so the leader asks before drawing a conclusion.

How to check the view is earning its place

After a few reviews, look back at the decisions made from it. Keep the measures that changed what someone did. Replace the ones everyone glanced at and ignored.

A number tells a leader where to look. The improvement starts with the conversation that follows. The leader opens the actual proposal, job or invoice and goes through it with the person doing the work. That makes the feedback specific, and the person can use it today.

Accountability comes from the follow-through. Each decision ends with one owner and one next action, and the leader comes back to check on it. People come to expect that what was agreed will be looked at again.

One leader doing this across everyone they oversee is a large part of how a personal improvement starts to change a team’s work. What does that look like with a real piece of work?

Coaching from the actual work →

One billing problem, start to finish

This case is illustrative. While preparing the weekly review, AI flags an invoice for a completed job as incomplete. The billing lead opens the record and finds that the completion note from operations never arrived.

The billing lead now has a decision to make. The operations coordinator owns getting the note over by Thursday. The billing specialist issues the invoice once it arrives. The lead checks both at the next review.

The coaching happens on that record. The lead shows what a complete handoff looks like and asks what got in the way. If the same gap keeps appearing on other jobs, the fix moves from coaching one person to changing the process, perhaps by making the note a required step when a job closes.

Who does what

• AI can summarize records, surface gaps and draft questions or a possible next step from the sources it has been given.
• The leader verifies the facts, weighs the context, makes the decision, assigns the owner and has the conversation.
• The employee explains what is really happening, raises what is blocking them, completes the next action and records it.

How this changes at higher levels

The same habit carries up through the company, with different work on the table. A vice president might go through a stalled account with a sales manager, looking at the records and at how the manager coached the rep. An executive team might see a delivery backlog that crosses departments and name one leader to own it, with a date to report back.

Each level coaches the people who report to it. That is how better habits at the top can reach people several layers away.

How to tell it is working

At the next review, ask four questions. Was the next action done? Did the work move? Is the same problem still coming back? Is the person handling it better without being asked?

The team’s work can start improving before everyone has new tools. Clearer priorities, better coaching and steady follow-up all come from the leader and reach people right away.

Then the leader can turn what works into a repeatable process. They have used it themselves, so they know where it saves time, where a person has to check it and which steps matter. That makes it far easier to teach than a process being worked out during the rollout.

A wider rollout is a bigger decision. It needs more investment in shared data, connected tools and training, and the percentage gain may be lower than one person’s. Total output can still be higher, because many more people are producing more. That is why it pays to choose carefully where to go next.

What lets the improvement spread →

What can spread right away

Some improvements need no new software for the team. The leader can share what the priorities are, agree what each measure means, start a regular review and pass along a template or checklist that has proved useful.

A small way to test the spread

Before a wider rollout, try one shared practice with the team. In the illustrative sales example, that practice is Rep A’s follow-up template. The team uses it for a set period and the leader reviews how it fits real work. The leader changes it where it doesn’t fit, then compares the same team using the same definition of a proposal sent.

What a wider rollout needs

• A repeatable workflow. The steps are written down with real examples, along with the cases that need different handling.
• Shared data. Everyone works from the same structured records and the same definitions.
• Clear responsibilities. Each person knows who enters the information, who checks it, who acts on it and who coaches.
• Training on real work. People learn with their own proposals, jobs or invoices, and learn how to check what AI gives them.

This groundwork takes time and money, which is why the choice of team matters. It also helps to check whether the departments on either side can handle the extra work once this team speeds up.

An operations example

This case is illustrative. A vice president of operations has been using a weekly review to see which jobs are at risk. Once it proves useful, each supervisor runs the same review with the same definitions. Coordinators update job status by an agreed time. The supervisors handle most issues directly and bring the vice president only the ones that cross teams or need a bigger decision.

The vice president’s own routine has now become a way of working that many people share, and it no longer depends on one person looking at everything.

Who owns what

The leader decides what gets rolled out, explains why it matters and keeps reviewing it. Managers run it with their people and coach from the work. Employees do the work, record it accurately and say when the process doesn’t fit a situation.

How the result gets checked

Measure the team on its own numbers, compared with where the same team started. Then look at the work behind those numbers. Are people really using the workflow? Are the records accurate? Are the next actions clear? A finished checklist with no useful work behind it doesn’t count.

In this illustrative week, the sales leader sees that Rep A sent 6 proposals, Rep B sent 5 and Rep C sent 1. Separately, 4 of Rep C’s open proposals are flagged as stalled.

The two views do different jobs. The count of proposals sent shows the leader where to look. The stalled flag points to four specific proposals that someone can open and act on today.

Neither one explains why. The records can show the last contact, what the customer was promised and what is meant to happen next, and the leader needs that before coaching anyone.

With access to the proposal records, AI can build this count and list the stalled items before the review. Deciding what it means stays with the leader. Someone more senior, such as a vice president of sales, might look at the same kind of view team by team and ask each manager about what stands out.

Two decisions come out of this illustrative review. Rep C follows up on the 4 stalled proposals today. The whole team uses Rep A’s follow-up template.

Why the two decisions differ

The first is accountability for specific work. It has a named owner, a clear action and a deadline. The second is a process decision. It gives everyone the same starting point for follow-up, so each rep doesn’t have to write their own.

What the coaching conversation covers

The leader and Rep C go through the four proposals one at a time. What has happened so far? What does this customer need next? What help does Rep C need to move it today? The conversation stays on those records, which gives Rep C something specific to act on.

What gets followed up

At the next review, the leader checks that the four follow-ups happened and what came of each one. The example doesn’t tell us why Rep A sent more proposals, or whether the template had anything to do with it. So the leader also looks at how the template works across the team and changes it if it doesn’t fit.

It doesn’t say why activity changed, or how much of the change came from any of these. It doesn’t prove that revenue, quality, delivery, billing or cash improved.

In this illustrative example, the same team sent 12 proposals per week before rollout and 18 per week after.

Keeping the comparison fair

The two numbers can only be compared if they cover the same team and use the same definition of a proposal sent and the same way of counting.

What the number tells you

It tells the leader that recorded activity changed. It doesn’t say why, or how much of the change came from AI, coaching, the template or something else entirely.

The question it leads to

A proposal sent is activity. The business gets paid when a proposal is won, the work is delivered, the invoice goes out and the customer pays. So a leader who sees 18 proposals where there used to be 12 has a good next question: what did the extra activity produce? Step 3 follows what has to happen before activity becomes booked work, profit or cash.