How do I decide which problem to fix first as a new ops manager?

Rank every problem on two axes: annual cost to the business, and how confident you are the root cause is what you think it is. Fix the highest-cost problem you're most confident about. Don't go after the biggest number if you can't name the root cause yet. Go after the biggest number you can describe in one sentence.

What does 'pick your battles' actually mean in lean?

It means accepting that a site has fifty things wrong with it and you have capacity for one major fix and three small ones in your first ninety days. The rest go on a list with dates against them. The list is not a backlog. It's a queue, and queues need a rhythm.

How long should the first improvement project take?

Long enough to land properly, short enough that everyone remembers what changed. On most sites that's six to eight weeks for the analysis and another four to twelve weeks for the cadence to set. If you can't see the win by week twelve, the scope was too big.

What if my MD wants me to fix three things at once?

Show them the queue. Show them what each fix costs in your time and the team's attention. Ask which one matters most this quarter. Most MDs back off when they see the actual capacity maths. The ones who don't are telling you something about how the company is run.

When should I escalate vs. quietly absorb a problem?

Escalate the second a problem becomes a safety, compliance, or customer-facing risk. Absorb the politics. The job is not to win every argument. It's to ship the improvements. Pick the battles that change a number on the board, not the ones that change someone's mind about you.

← Notes · 24 May 2026

Pick your battles: the ops manager rule that fixes the right thing first

Most ops managers try to fix everything in the first quarter. The ones who keep their jobs fix the right one thing and let the rest wait. Here's how to choose.

The first three months in any ops manager job are the same. You walk the floor on day one and write down forty things that need fixing. By the end of week two you’ve added sixty more. By the end of month one you’ve got a list of problems longer than your arm and a meeting calendar that doesn’t have room to fix any of them.

This is where most ops managers lose the job before they realise they’re losing it. They try to fix everything in the first quarter. They pick three big projects, run them in parallel, and by month four all three are stuck because the team they need is the same team for all three.

The rule that took me years to learn: pick your battles, and fix the right one first.

What “the right one” actually means

It doesn’t mean the biggest. It means the biggest one you can describe in a single sentence with a number on the end.

If you can’t name the root cause yet, it’s not a battle. It’s a survey. Surveys come before battles. Don’t pick a fight with a problem you haven’t measured.

Here’s the test I use. Write down the problem. Then write down the annual cost. Then write down what the root cause is, in one sentence. If any of those three is fuzzy, the problem isn’t ready to fight yet. Park it. Go and measure.

The problems that are ready usually look something like this:

“The Tuesday lunch changeover takes 95 minutes when the work-study says 22. Costs us £39k a week. Root cause: parts trolley arrives after the line stops, not before.”
“Bottle line OEE is 62%. Should be 77% on the capacity model. Root cause: micro-stops on the labeller that nobody logs.”
“Picker rate is 78/hour. The standard is 110. Root cause: pickers walk to the printer twice per order because labels and pick notes print in different bays.”

Those are battles. Pick one. Win it. Then pick another.

The £40k Tuesday lunch fight

I’ll give you the worked example.

Site I was running, the Tuesday lunch changeover took 95 minutes against a 22-minute work-study. Three changeovers a shift, two shifts, six days a week. At the line’s contribution per hour, that’s a £39,000 weekly hole nobody had named.

Forty things needed fixing on that site. The asset register was a mess. The label printer kept jamming. The morning huddle had no agenda. The KPI board hadn’t been updated in a fortnight. All of those things mattered.

But none of them were £39,000 a week.

So we picked the changeover. Two of us spent a Tuesday with stopwatches. Eight minutes of actual internal work. Fourteen minutes of fetching the parts trolley because nobody had thought to bring it up before the line stopped. Twenty-two minutes of two people stepping on each other at the inspect station. Thirty-one minutes of waiting for the lab to release.

Three rules went up:

Parts trolley arrives ten minutes before the line stops. Team leader’s job, not the operator’s.
QC release happens during the previous batch’s final thirty minutes.
One person does internal work, one person does external. They swap when the line restarts.

Six weeks later, mean changeover was thirty-four minutes. £25k a week recovered. £1.3M annualised. From a Tuesday with a stopwatch.

The asset register was still a mess. The label printer still jammed. The huddle still had no agenda. We knew about all of them. We had them on the queue.

By month four we’d fixed the huddle. By month six we’d cracked the label printer. By month eight the asset register was clean. None of them moved the £39k number. They were the battles we picked second, third, and fourth, once we’d won the one that paid for the rest.

How to actually queue the rest

The list of forty things doesn’t disappear when you pick your battle. It sits in the way, in your head, in every meeting. You need a queue.

The queue I use is one A4 sheet on the noticeboard with three columns: Now, Next, Later.

Now has one thing in it. Whatever I’m fighting this quarter. Nobody puts anything else in this column. Not the MD, not the FD, not the customer.
Next has three things. The fights queued for the following quarter. Names against them. Rough dates.
Later is everything else. It’s allowed to be long. It just isn’t allowed to jump.

When a new problem lands (and one will land every week) it goes in Later by default. Promotion to Next requires a number against it. Promotion to Now requires the current battle to be won.

That’s the rule. People will hate the rule for the first month and quietly love it by month three, because for the first time in their working life they know what they’re being asked to do this week.

When to break the rule

Three things break the rule. Safety, compliance, customer escalation. If any of those land, they go straight to Now and whatever was in Now drops to Next.

Everything else waits. The CEO’s pet project waits. The new bit of kit your engineer’s excited about waits. The supplier audit waits. The list is allowed to be long. The Now column is not.

The job of an ops manager is not to be available to fight every fire. The job is to pick the one fire that’s burning down the building and put that one out first. The rest are background noise until they aren’t.

What this looks like at day 90

If you’ve done this right, here’s what your first quarter looks like:

One major win on the board with a number against it
Three things queued in Next, scoped, dated, owned
A noticeboard that everyone on the floor can read in thirty seconds
A team that knows what they’re being asked to do this week

If you’ve done this wrong (if you’ve tried to fight on three fronts at once) here’s what your first quarter looks like:

Three projects, all at 60% complete, none of them moving a number
A list of forty things in your head
A team that thinks you flit between priorities
An MD who’s started asking what you’ve actually fixed

Pick your battles. Fix the right one first. Queue the rest.

The 90-day cadence behind this (what gets fought in week one, what gets standardised at day thirty, what handover artefact is due at day ninety) is laid out in the Playbook. £97 for the first 100 buyers, £197 evergreen after. Same operator, same approach, written by someone who’s done it three times in three businesses.

Tags: lean prioritisation ops-manager