Ask a manager how the AI rollout is going and you will often hear a usage number. Four hundred people logged in this month. Twelve thousand prompts. Ask how many hours it saved and the room goes quiet.

That silence is the real finding. Most companies count how much AI is used. Far fewer count what it gave back.

The two numbers can point in opposite directions, and when they do, the usage number usually wins the meeting.

Why usage numbers feel like progress

Usage is easy to count. The tool reports it, the chart goes up, and a rising line looks like momentum. Everyone feels the investment is working.

But usage measures activity. It tells you people are busy with the tool, not that they finish sooner. A gym card swipe shows who walked in. It says nothing about who got fitter.

This matters because decisions follow the numbers. If usage is the scoreboard, teams play for usage. They find reasons to open the tool, and they report that they did.

A tale of two teams

Here is an illustrative example. A company has two teams that each prepare 20 client reports a month. Each report takes about 4 hours: 3 hours to draft and 1 hour to check.

Team A has six people who use the AI tool every day. They rewrite emails, tidy meeting notes, and brainstorm subject lines. The client reports still take 4 hours each.

Team B has only two people using the tool, and they use it for one thing: the first draft of the monthly client report. Drafting drops to about 1.25 hours. Checking grows to about 1.25 hours, because the draft needs careful reading. Each report now takes about 2.5 hours.

Do the math for Team B: 1.5 hours saved per report × 20 reports = 30 hours a month. Team A saved nothing on the work that matters.

Now look at the usage dashboard. Team A shows six daily users and Team B shows two. By usage, Team A is the success story. By hours, Team B is the one worth copying.

Notice that Team B’s checking time went up, and I counted it. If you leave the checking out, you flatter the tool and mislead yourself.

Neither team is lazy or wrong. Team A found the tool useful for small things, and small things add up to a pleasant day. But a pleasant day is not a result the business can bank.

The number your customer notices

Hours are not the only measure. Suppose a report used to take five days from request to delivery, and with a first draft ready on day one, it now takes three. The team saved some hours. The client got answers two days sooner.

Clients rarely notice the hours inside your team. They notice how long they waited. That is why cycle time deserves a place next to hours saved, and why it often makes the stronger case to leadership. The five-day and three-day figures here are illustrative, so measure your own.

Four numbers worth tracking

You do not need a complicated dashboard. For any one workflow, four measures cover most of what matters.

• Time per task: how long one unit of work takes, including checking.

• Cycle time: how many days pass from the request to the finished result.

• Errors and rework: how often the work comes back for correction.

• Volume: how much work the same people can handle in the same period.

Pick the one or two that match the workflow. For client reports, time per task and cycle time may be enough. For invoice processing, errors may tell the story.

These are measures of work, not of tools. They stay useful when the tool changes, which it will.

How to measure one workflow

1. Choose one task that repeats often enough to measure, such as a monthly report, a weekly update, or a standard customer reply.

2. Time it before you change anything. Use a clock, not a guess, and time at least five examples.

3. Change one thing. Use AI for one step, not for everything at once.

4. Time it again for a month, including the checking.

5. Compare the two sets of numbers and write down the difference in plain hours.

If the difference is large, move on to the next workflow. If it is small, you have learned something just as valuable, quickly and cheaply.

Why one change at a time

It is tempting to roll AI into every step and call the result a win. The problem is that you cannot tell which step helped and which one made no difference, or made things worse.

Changing one step gives you a clean answer. It also gives the team a small, visible success to point to, which does more for confidence than a large announcement.

One caution: do not stop the clock after the first week. New tools often feel slower before they feel faster, so give the change a full month before you judge it.

Once one workflow has a trusted number, the next one is easier to start. People believe results from their own colleagues, measured on their own work.

What about benefits you cannot count?

Some gains are real but hard to count. A report may read better. A team may feel less stressed. A new hire may get up to speed faster.

Those matter. Still, try to attach one number to them: turnaround time for the new hire, or the number of revisions requested on the report. A benefit with a number behind it survives the budget meeting. A benefit that rests only on a feeling usually does not.

What to show your leadership

A usage dashboard answers a technical question: are we using what we paid for? It is a fine report for the person who manages licenses.

It does not answer the question a CEO is actually asking: is the business better off? For that, show one number per workflow each month. Hours saved. Days cut from the turnaround. Errors avoided.

Keep the labels plain, such as “Monthly client reports: 4 hours down to 2.5 hours each.” Anyone can read that in five seconds.

Two or three of those, with plain labels, tell a better story than twenty charts about logins. They also protect the AI effort. When budgets tighten, projects with a number survive. Projects with a login chart get a polite thank-you.

Try it this month

Ask one team lead for the recurring task their team would happily never do by hand again. Time it five times and write the average on a card.

Then run the new way for a month and time it again. You will have a before-and-after figure you built yourself, from your own work, and it will be worth more than any adoption rate.

If the number is disappointing, do not hide it. Look at which step did not improve, and why. The most useful finding is often that the old process had waste that AI cannot fix, and that waste should go first.

AI usage tells you who is busy with the tool. Hours saved tells you whether the business is better off.

If your CEO asked tomorrow how many hours AI saved your team last month, what number would you give?

If you’d like to talk through how to measure it in your team, I’d be glad to hear your thoughts.

ADDITIONAL READING

• Why Comparing Three Supplier Quotes Takes Two Days, and How to Fix It — https://jordanimutan.com/2026/10/15/why-comparing-three-supplier-quotes-takes-two-days-and-how-to-fix-it/

• Why Your Next AI Win Doesn’t Need a Developer — https://jordanimutan.com/2026/10/08/why-your-next-ai-win-doesnt-need-a-developer/

• You Trained Your Team on AI. Their Output Didn’t Change. — https://jordanimutan.com/2026/09/15/you-trained-your-team-on-ai-their-output-didnt-change/

• The Real Reason Most AI Projects Stall Before They Start — https://jordanimutan.com/2026/09/11/the-real-reason-most-ai-projects-stall-before-they-start/

• Your AI Rollout Won’t Fix What’s Actually Slowing Your Managers Down. — https://jordanimutan.com/2026/09/16/your-ai-rollout-wont-fix-whats-actually-slowing-your-managers-down/

A manager sends a routine decision up to their boss. The boss answers in two minutes. The decision waited two days.

If this happens often in your company, senior leaders are spending their week on decisions someone closer to the work could make.

Escalation means sending a decision or problem to a higher level for the answer. It is useful for real exceptions. It becomes a problem when it is the default.

Why managers escalate

The cause is usually not weak managers. It is one of three conditions:

• Unclear limits. Managers do not know where their authority ends, so they ask to be safe.
• Fear of blame. A past decision went wrong and someone was criticized. Asking first now feels safer than deciding.
• Bosses who always answer. Every time a senior leader gives the answer, the habit of asking gets a reward.

What it costs

Decisions wait. Customers wait. Senior leaders lose time they need for bigger problems. And managers never build judgment, because they never get to use it.

Four steps to fix it

1. Find the decisions that go up most often. Ask senior leaders to keep a simple tally for two weeks.
2. Sort each decision into one of three groups. Decide alone. Decide, then tell the boss. Ask first. Write the limits in numbers where you can. For example, a manager may approve refunds up to a set amount or overtime up to a set number of hours without asking.
3. Change what the boss says. When a manager brings a problem, the boss asks: “What do you recommend?” Over time, managers arrive with a problem and an answer.
4. Track two numbers. The number of escalations each week, and the days a decision takes. Review them monthly.

Where training fits

Managers do need practice in making and explaining decisions. But training alone will not help if the limits are unclear, because managers will go back to asking. Fix the limits first, then train.

Reducing slow or unnecessary escalation is one of the business priorities we build a 90-day pilot around at POC Management Execution. A pilot is a small trial with one group of managers, run before a change goes company-wide.

Managers escalate less when they know exactly what they can decide.

Which three decisions go up to your senior leaders most often, and could a manager make them with clear limits?

If you want to work this out for your company, message me and we can brainstorm it together.

Further reading on jordanimutan.com:

• How to Keep the Results of a Leadership Program After the Program Ends: https://jordanimutan.com/2026/10/21/how-to-keep-the-results-of-a-leadership-program-after-the-program-ends/

• How to Choose One Business Priority Before You Start a Manager Development Program: https://jordanimutan.com/2026/10/16/how-to-choose-one-business-priority-before-you-start-a-manager-development-program/

#DecisionMaking #ManagerEffectiveness #LeadershipDevelopment #HRLeadership #OperationalExcellence

Your meeting ended an hour ago, and nobody can say for sure what was decided.

Someone is writing the minutes. They are scrolling through notes taken in a hurry, trying to remember whether the budget was approved or only discussed. Two attendees have already sent messages asking who owns the follow-up. Nobody is sure, so nobody moves.

This is the moment many teams reach for AI. Let a tool listen, summarize, and send the minutes. It is a good idea. It also skips a more important question: did the meeting produce anything worth summarizing?

Where the time goes after the meeting

Take one weekly management meeting with eight people. These numbers are illustrative, so replace them with your own.

• Note-taker turns rough notes into clean minutes: 60 minutes

• Sending them out and fixing corrections: 20 minutes

• Eight attendees reading long minutes: 8 × 5 = 40 minutes

• Follow-up messages asking who owns what: 30 minutes

That is 150 minutes, or two and a half hours, every week. Over 50 weeks it adds up to 125 hours spent on the paperwork of one meeting.

Look at the last line. Thirty minutes of messages asking what was decided is not a writing problem. It is a sign the meeting ended without a clear answer.

The cost is also uneven. The note-taker is often a junior person or an assistant, and that hour lands on them every week. They carry a quiet tax for a problem they did not create and have little power to fix.

Would I automate this?

Meeting minutes are a good test case, because some parts of the job suit AI and some do not.

I would hand AI the mechanical parts: turning rough notes or a recording into a clean first draft, pulling out action items with names and dates, and sending a reminder two days before a deadline. These tasks repeat every week, and checking a draft is faster than writing one.

I would keep three things with a person: deciding what was actually decided, confirming who owns each action, and noting what was left open on purpose. These are judgments. A tool cannot tell the difference between “we agreed” and “we stopped arguing because it was five o’clock.”

That split is useful, and it also exposes the real problem. If a person has to decide what was decided, they should do it while everyone is still in the room.

What good minutes look like

Most minutes record the discussion. Readers need the outcome. A useful set of minutes fits on half a page and has three parts: decisions made, actions with an owner and a date, and questions still open.

The discussion itself rarely needs a record. If someone wants the reasoning behind a decision, one line is usually enough. Five paragraphs of who said what mostly give people something to disagree about later.

Shorter minutes also get read. Long ones get filed, which is a polite word for ignored.

Here is a simple test. Hand the minutes to someone who missed the meeting. If they can tell what was decided and what they must do next, the minutes work. If they need a phone call to understand them, they do not.

Fix the meeting before you fix the minutes

Here is a habit that costs five minutes and removes most of the confusion. In the last five minutes of the meeting, the chair reads back three things for each decision: what was decided, who owns it, and by when.

Say it out loud, then ask if anyone disagrees. If a decision cannot be stated in one sentence, the meeting did not make one, and it is better to find that out now than in next week’s meeting.

This is the part most teams miss. Good minutes are not written after the meeting. They are spoken at the end of it, and then written down.

When the read-back happens, an AI draft has something solid to work from. When it does not, the tool will produce a smooth, confident summary of a conversation that never reached a conclusion. That is worse than messy notes, because it looks finished.

Why the read-back feels awkward at first

People resist it. It feels like extra ceremony, and it can expose that the group never agreed on anything. That is exactly why it works.

Start by making it routine, not dramatic. Same five minutes, same three questions, every meeting. After three or four weeks, attendees expect it, and they speak more clearly during the meeting because they know a read-back is coming.

The chair carries this habit. If the chair skips it when the meeting runs late, the team learns it is optional. Protect the five minutes the way you protect the start time.

What the numbers look like after

Run the same meeting with the read-back and an AI-assisted draft.

• Read-back at the end: 5 minutes × 8 people = 40 minutes

• AI drafts the minutes from notes or a recording, and the note-taker checks it: 20 minutes

• Eight attendees reading a half-page summary: 8 × 2 = 16 minutes

• Follow-up messages: 10 minutes

The total is 86 minutes, down from 150. That is about an hour saved each week, and roughly 50 hours over a year.

Notice that the read-back adds time, and I counted it. A fair comparison includes every new step. The hours saved are modest. The larger gain is fewer lost decisions, and fewer weeks that start with “Wait, did we agree to that?”

Always check the draft

Even with a good read-back, read the draft before it goes out. Tools mishear names, mix up two people with similar names, and sometimes turn “let’s consider it” into “approved.”

Have someone who was in the meeting check every action item against their own memory. It takes a few minutes, and it prevents the kind of mistake that surfaces three weeks later in front of the boss.

Be careful what the tool hears

Some meetings should not be recorded or fed to a tool. Discussions about pay, hiring, discipline, or legal matters often carry confidentiality rules. Check your company policy, and tell attendees whenever a meeting is being recorded.

A simple rule works well. Use AI for ordinary operating meetings first. Keep sensitive conversations in the hands of a person until your policy says otherwise.

Try it for two weeks

1. Pick one recurring meeting that already has minutes.

2. Time the work around it: writing, sending, reading, and follow-up messages.

3. In week one, add the three-line read-back to the last five minutes.

4. In week two, let AI draft the minutes from your notes, and have the note-taker check every action item.

Keep the test fair. Use the same meeting, the same people, and roughly the same agenda in both weeks, so the difference you see comes from the change and not from a lighter week.

Compare the two weeks, and count the follow-up messages too. They tell you whether the meeting really ended.

If the read-back shows that your meetings often end without decisions, you have found a bigger saving than any tool can offer. Some meetings do not need better minutes. They need to be shorter, or to not happen at all.

A meeting that cannot name its decisions cannot be summarized well by anyone, human or machine.

In your last meeting, could every person in the room have stated the decision, the owner, and the date?

If you want to talk through how this could work in your team, I’d be glad to hear your thoughts.

ADDITIONAL READING

• Not Every Task Should Use AI. Here’s the Test We Use to Tell. — https://jordanimutan.com/2026/09/25/not-every-task-should-use-ai-heres-the-test-we-use-to-tell/

• The 20% of Any Job That’s Actually Worth Giving to AI — https://jordanimutan.com/2026/10/06/the-20-of-any-job-thats-actually-worth-giving-to-ai/

• Before You Automate That Weekly Report, Ask Who Reads It — https://jordanimutan.com/2026/10/13/before-you-automate-that-weekly-report-ask-who-reads-it/

• Your Company Isn’t Slow—Your Decisions Are Trapped in Manual Processes — https://jordanimutan.com/2025/12/19/your-company-isnt-slow/

• AI Was Supposed to Save Your Managers Time. It Didn’t. — https://jordanimutan.com/2026/09/03/ai-was-supposed-to-save-your-managers-time-it-didnt/

A supplier is late. A key customer is unhappy. A report has a mistake in it. Someone on the team knows, and a few days go by before the boss hears about it. By then, a small problem has become a costly one. This is one of the most common patterns I see in the managers I work with after more than 30 years of managing people, and it is rarely about honesty. It is about not knowing when raising a problem is the right move.

Why problems travel up so slowly

Most people do not hide problems to cause harm. They wait. They hope to fix it first. They don’t want to bother a busy boss. Or they are not sure whether it is serious enough to mention. Fear plays a part too. A 2026 survey of 2,000 UK workers by Mental Health First Aid England, as reported by Arinite, found that 45% did not feel safe raising mistakes or highlighting risks, and 15% said they had made preventable mistakes because they felt unsafe speaking up. It is a UK study, but the pattern is human, and managers in Philippine companies will recognize it.

The result is the same everywhere: the person closest to the problem knows first, and the person with the authority to fix it knows last.

Escalation is not a failure

Many managers treat escalating as admitting they cannot cope. So they carry the problem alone for too long. Good escalation is the opposite. It means sending a problem to the person who can act on it, early enough that acting still helps. Silent delay is the real failure.

Three triggers: when to raise it

Give your team a simple rule. Raise it the same day if any of these is true.

It will miss a deadline, a budget, or a promise to a customer.

Solving it needs authority or money that you do not have.

It touches another team or a customer who may be affected.

If none of the three is true, handle it and mention it in your next regular update. If any is true, do not wait to be sure. Being early and slightly wrong costs far less than being late and right.

How to raise it: What. So what. Need.

A good escalation fits in three short lines.

What happened. One or two facts, no story.

So what. What it will affect if nothing changes, and by when.

Need. What you want from the person you are telling: a decision, a resource, or simply to be aware.

The third line matters most. Without it, the boss does not know whether to act or just listen, and may react to the wrong thing.

What owners and senior managers must do

The rules above only work if the first reaction to bad news is calm. If the person who raises a problem is questioned harshly, blamed, or sighed at, they will wait longer next time. Thank them first. Ask your questions second. People learn how safe it is to speak up from how the last person was treated.

Think about the last problem that reached you later than it should have. What did that person expect would happen when they told you?

If you would like to talk through how to set escalation rules in your own team, send me a message. I’m happy to compare notes, no pitch attached.

Next reads from the blog:

• You Think People Aren’t Taking Ownership. They Think It’s Not Safe To. — https://jordanimutan.com/2026/03/21/you-think-people-arent-taking-ownership-they-think-its-not-safe-to/

• Your Managers Keep Avoiding Difficult Conversations — And It’s Quietly Killing Performance — https://jordanimutan.com/2026/05/05/your-managers-keep-avoiding-difficult-conversations-and-its-quietly-killing-performance/

• The Feedback That Arrives Six Months Too Late — https://jordanimutan.com/2026/09/30/the-feedback-that-arrives-six-months-too-late/

• How to Set Team Priorities: Why “Everything Is Urgent” Slows Managers Down — https://jordanimutan.com/2026/10/12/how-to-set-team-priorities-why-everything-is-urgent-slows-managers-down/

• How to Help Newly Promoted Managers Perform Faster in Their First 90 Days — https://jordanimutan.com/2026/10/22/how-to-help-newly-promoted-managers-perform-faster-in-their-first-90-days/

#ManagementSkills #Leadership #Accountability #PhilippineSMEs #TeamCommunication

Most companies promote their best performer and then wait to see what happens.

The skills that earned the promotion are rarely the skills the new job needs. A great salesperson, nurse, engineer or teller was rewarded for doing the work. A manager is judged on how well the team does it.

What changes on day one

A new manager has to do things they have never done before:

• Hand over tasks they used to do themselves.
• Run regular one-on-ones. A one-on-one is a private meeting between a manager and one team member.
• Set clear expectations.
• Give feedback, including hard feedback.
• Hold people accountable without taking the work back.

These are new habits. A new manager cannot learn them by guessing, and a two-day course before the promotion will not make them stick.

What a slow start costs

The team is unsure what is expected. Decisions wait. Good people start looking elsewhere. The time it takes a new manager to perform well in the role, often called time to productivity, stretches from weeks to months. Most companies never measure that cost.

A 90-day plan HR can set up

Days 1 to 30: Listen and learn. Meet each team member one-on-one. Ask what work they own, where it gets stuck and what the last manager did well. Learn the one business result the manager’s own boss cares about most.

Days 31 to 60: Build habits. Start a weekly one-on-one with each team member. Write down expectations for each role on one page. Hold one feedback conversation, and practice it first with the boss or a peer.

Days 61 to 90: Own one improvement. Pick one problem in the team’s work, fix it, and show the boss the before-and-after number.

Three supports that make the plan work

• A boss who checks. A 20-minute check-in at the same time each week. The question is simple: what did you do this week on the plan?
• Short sessions, not one long course. Two hours every two to three weeks works better than two days upfront. New managers have little free time, and each lesson lands best when they have just met the problem it solves.
• A peer to learn from. Pair the new manager with an experienced manager they can ask questions without feeling judged.

How to know it is working

At day 90, ask three questions. Are weekly one-on-ones happening? Are expectations written down? Did the team deliver one improvement? If the answer to all three is yes, the manager has a good start.

At POC Management Execution, we treat new-manager effectiveness as a business priority: one group of new managers, a few clear practices, a boss who reinforces them, and one number tracked.

A new manager’s first 90 days should be planned, not left to chance.

What happens in your company in the first 90 days after someone is promoted to manager?

If you would like to think through a plan for your own company, message me and we can brainstorm it together.

Further reading on jordanimutan.com:

• How to Choose One Business Priority Before You Start a Manager Development Program: https://jordanimutan.com/2026/10/16/how-to-choose-one-business-priority-before-you-start-a-manager-development-program/

• Turn Capability Into Consistent Execution: https://jordanimutan.com/turn-capability-into-consistent-execution/

#NewManagers #LeadershipDevelopment #ManagerOnboarding #HRStrategy #PeopleManagement