Operations Note 01 · how to find what’s actually limiting your business
Find the Constraint
A business rarely has a shortage of problems. It has a shortage of clarity about which problem is actually worth fixing. Here’s a method for using AI to find that one thing, instead of guessing.
The idea in one line: a business is a connected system, and at any given point, one constraint is usually doing more than any other single thing to limit it. Fix something else, however inefficient it is, and the result often doesn’t move. We call that limiting part the constraint.
Friction isn’t the same as a constraint
Every business has a list like this: the team keeps asking the same questions, work sits waiting on one person, customers wait too long for a reply, mistakes get made and then re-made. All of it feels like the problem.
Some of it is. Most of it isn’t — not in the sense that matters. Fixing an annoying task saves the person doing it some time. Fixing a constraint changes what the business can actually produce, sell, or deliver. Those are different outcomes, and the method below is built to tell them apart before you spend money closing the wrong gap.
Pick one process
Don’t start with “analyze my business.” Start with one process that has a clear beginning and a clear end.
Then feed AI the evidence, not your summary of it. Emails, CRM records, project or task data, invoices, proposals, support tickets, meeting notes, spreadsheets, anything that shows what actually happened, not what people remember happening. Ask it to reconstruct the process from that evidence, and to keep facts and guesses clearly separate:
- “Based only on what I’ve given you, describe this process from beginning to end: the major steps, who’s involved, where a step depends on another person or system, and where work seems to wait.”
- “Don’t recommend anything yet.”
- “Separate what the evidence shows from what you’re inferring, and say plainly where there isn’t enough evidence to know.”
Look for six kinds of friction
Have AI point out where the process behaves differently than you’d expect. These six patterns are a good place to start.
Waiting
- Work is ready to move but sits until someone approves, replies, or gets to it.
Handoffs
- Work changes hands between people or systems. Not automatically bad, but every handoff is a place things get lost or misread.
Rework
- Something gets corrected, re-sent, or re-done. A proposal rewritten, an order fixed, information asked for twice.
Exceptions
- The normal process works until something unusual happens, then everyone improvises. Frequent exceptions usually point to a process or decision problem underneath.
Bottlenecks
- Work piles up at one point: one approval, one inbox, one person’s calendar. This is often the strongest signal of all.
Dependency
- One person is the only one who can do something. Invisible when they’re available, a hard limit the moment they’re not.
Test each one before you trust it
You’ll end up with a list of things that feel broken. Don’t act on the longest list or the loudest complaint. For each item, finish this sentence:
“If we fixed this, the business would be able to ______.”
Specific answer
- “Handle 30% more orders without hiring.”
- “Get qualified proposals in front of customers within a day instead of four.”
- “Stop the owner from having to review every deliverable personally.”
Vague answer
- “Be more efficient.”
- “Save some time.”
- “Have a cleaner process.”
If the honest answer is vague, you haven’t found the constraint yet, even if the problem is real. Vague answers are usually a sign you’ve found friction, not the thing limiting the system.
A worked example
Take a business that assumes its growth problem is lead volume: not enough people in the pipeline. Reconstruct the sales process from the CRM and email evidence, and a different picture shows up. Leads are arriving faster than the sales team can follow up with them. Follow-up capacity, not lead volume, is the bottleneck.
Look one layer further and it can move again. Every proposal needs the owner to review pricing before it goes out, and the owner only reviews proposals twice a week. Now the constraint isn’t follow-up capacity either. It’s a decision that only one person is allowed to make, on a fixed schedule that has nothing to do with how fast the business could otherwise move.
That’s the pattern worth internalizing: the first thing that looks like the problem is rarely the constraint. The constraint is usually one or two layers underneath it, and you only find it by following the evidence instead of the first plausible explanation.
Once you’ve got a specific, testable answer to “what would change if we fixed this,” you have a constraint worth taking seriously. Note 02 covers what to do next: finding out why it’s actually happening, before you touch it.
Catalyst Studio · Operations Notes