The most important systems decision is what you remove

Ant Murphy wrote a piece called Remove. Don't Add. that's framed for product managers, but the underlying point applies to anyone running operations.

He cites a study from the University of Virginia. Eight experiments where participants were asked to improve something. Overwhelmingly, they added rather than subtracted. In one experiment, removing a Lego block was free and adding cost money, and 60% of participants didn't even consider the subtractive option.

We're wired to add. And in operations work, that wiring is expensive.

The problem this names

Think about your business. When something isn't working, what's the default move?

You add a meeting. You add a checklist. You add a tool. You add a step to the process. You add a dashboard. You add an automation. You add a hire.

Adding feels like progress. It feels like you're solving the problem. It looks like decisive action.

But Ant cites a stat that should stop most operators in their tracks: in software, an estimated 80% of features are rarely or never used. Just 12% drive 80% of usage. The vast majority of what teams build collects dust. And the dust isn't free. Every feature carries a maintenance tax. Every tool carries a subscription. Every step in a process carries cognitive load.

He calls this negative value. Not just waste. Things that actively make the business harder to run.

Where operators see this every day

Pull up your business right now and look at:

• The SaaS subscriptions you're paying for. How many are you actually using?
• The recurring meetings on the calendar. How many produced a decision in the last month?
• The dashboards or reports you've built. How many do you check more than once a week?
• The processes your team runs. How many actually catch problems vs. how many just generate paperwork?

If you're like most growing businesses, the honest answer is most of them aren't earning their keep. And every one of them is taking attention away from the few that are.

Why subtracting is hard

There are real reasons we don't subtract.

Adding a feature, tool, or process gets you credit. Removing one rarely does. When was the last time someone got promoted for shutting down a tool? Killing a meeting? Removing a step from a process?

There's also the fear that whatever you remove will turn out to have been important. Even if it's not used much, what if you need it later? This is real. Some things are rarely used but valuable when they're needed. (The reset password page on a website is the classic example.)

But the trap is using that fear as cover for never subtracting anything. The result is operational sprawl. A business that runs on dozens of tools, processes, and meetings, most of which aren't earning their place but none of which anyone has the courage to kill.

The operator's version of the fix

Two practices worth borrowing from Ant's piece, translated for operations:

1. Build a NOT doing list.

A deliberate, written list of things you've chosen not to pursue. Not a backlog. Not "we'll get to it later." An explicit we are not doing this, and here's why.

For an operator that might look like:

• Not running weekly all-hands until we hit 15 people
• Not using Slack channels per project until we have more than 5 active projects
• Not building automated reporting until the underlying data is clean
• Not adding another paid SaaS until something else gets cancelled

The list gets reviewed quarterly. Things get added. Things occasionally get removed. But the act of writing it down forces the decision out into the open.

2. Run a regular subtraction audit.

Once a quarter, walk through your operations and ask one question: what could we remove?

Tools. Meetings. Reports. Processes. Steps. Approvals. Tabs. Dashboards. Anything.

You're not committing to remove anything yet. You're just asking the question. Most operators almost never ask it. Even doing it for thirty minutes a quarter surfaces things that have been silently costing you for months.

The connection to AI

Ant's piece is also about how AI makes this worse. When delivery speeds up, the temptation to ship more increases. Everyone is shipping AI features nobody asked for. Throwing AI at your roadmap is a fast path to having a TV remote with 50 buttons nobody uses.

The operator version: just because AI makes it cheap to add a system doesn't mean every system you can add belongs in your business. Faster in the wrong direction doesn't help. The discipline of asking should we add this is more important now than it was two years ago, not less.

The takeaway

The companies you admire aren't the ones running the most tools. They're the ones running the fewest tools that actually work.

Strategy is as much about what you choose not to do as what you choose to do. That applies to operations too.

Read Ant's original piece →