Skip to content
Back to blog
Plain English·4 min read

What Does "Best" Even Mean?

Ask an AI to make you the best restaurant in Boone and it has to guess what you meant. The guess is where the disappointment comes from, and ten minutes with a pen is the fix.

Say you want to be the best restaurant in Boone.

Think about how that actually lands. A good AI should come back and ask what "best" means to you, because if it quietly decides best means the top review score, that may not be what you meant at all. Maybe you meant the best outdoor dining in town. Maybe the best spot for people who live here, rather than people passing through for a weekend. Or the best breakfast, which is a different question than the best dinner, with different food, different hours, and different people coming through the door.

Most tools won't ask. They'll pick one and start helping.

Where the disappointment comes from

Best. Better. More. Improve. Those words feel specific when you say them out loud, and they go blurry the second something else has to act on them. The machine fills the blur with the most common version of the answer, which is another way of saying the most generic one. Then it does a genuinely good job on a question you never asked.

A law office that asks for help getting more clients gets the same twelve suggestions as a salon asking for more walk-ins, because neither request carried anything about the office or the salon. Both owners walk away thinking the tool is overrated. The tool answered the question in front of it.

These things are extraordinarily capable. Clear problems are the rare part.

The sidewalks nobody planned

There's a thing you see on old college campuses, and App State is no exception. Brick walkways wander across the grounds at angles nobody would draw on purpose. The story goes that those weren't the original design. The university laid sidewalks where the blueprint said people should walk. Then people walked where they actually wanted to walk, across the grass, on the shortest line to wherever they were going, until the grass wore down and died in exactly those lines. Eventually the university stopped fighting it and paved the routes people had already made.

Planners call those desire paths. The official sidewalks were the theory of how people should move. The dead grass was the truth of how they did.

I've sat through more than one argument about where a sidewalk ought to go, and the ones that go badly are the ones where everybody is looking at the map. The map is easier to argue about. It's also the thing that's wrong.

Defining a problem works the same way. Watch what's really happening and build the path there, instead of pouring concrete on an assumption.

Sketch it before you automate it

Before you open any of these tools, write the problem out yourself. By hand, in plain language, and with no AI. What am I actually trying to solve? What does a good outcome look like when it shows up?

That sounds too simple to be the answer, which is exactly why almost nobody does it. It's also the most reusable habit here, because it works the same on a menu change, a hiring decision, and a piece of software somebody is trying to sell you. The sketch is where you find out that "best restaurant in Boone" was three different businesses wearing one sentence. It's where a hotel owner works out that the occupancy number they've stared at all year was hiding the real question, which is why they keep discounting the same three nights out of habit. Ten minutes with a pen routinely kills an idea that would have eaten a month, and that's the ten minutes doing its job.

Then hand the sketch to somebody else

Take what you wrote and see whether it survives another person. Run it past a partner, an employee, or a friend who runs their own shop. Paste it into a chatbot and ask it to poke holes in it.

This is the step that separates the people who get real value from these tools from the people who get disappointed, and it costs one conversation. If you can't explain the problem clearly to another person, you aren't ready to hand it to a machine. You're about to give a robot a blurry instruction and act surprised when it flows downhill.

The whole method is a pen, ten minutes, and the willingness to find out that the question you've been asking for a year has three answers inside it. Nothing to buy, nothing to install, and no new skill to go learn first.

What you're actually buying

The tool is only as good as the problem you hand it. That's easy to agree with, and easy to skip on a Tuesday when you just want the thing done.

Every owner has a version of that dead grass somewhere. The route people actually take through your business, worn into the ground, in plain view, while the official sidewalk sits there in the plan you wrote three years ago and never looked at again.

So before the next thing you automate, go find yours.

Common questions

Isn't this just prompt engineering?

The habit comes before any tool touches it. Defining the problem in your own words, on paper, works the same on a menu change, a hiring decision, and a piece of software somebody wants to sell you. Better prompts are a side effect of a clearer problem, and the clearer problem pays even when no AI is involved.

How long should the problem sketch take?

About ten minutes with a pen. If it takes much longer, you have usually discovered that the question has several answers living inside it, which is the sketch doing its job early instead of a project doing it expensively later.

What does a well-defined problem look like?

It names what you are actually trying to solve, what a good outcome looks like when it shows up, and what you have already tried. The test is specificity, and the bar is that the closest generic answer will not pass for what you meant.

Can I use AI to sharpen the problem itself?

Yes, and it is one of the better uses. Paste the sketch in and ask it to poke holes, the same way you would run it past a partner or an employee. The step worth refusing to skip is the sketch, because handing over a blur and letting the machine guess is where the disappointment comes from.

Want to see this in action for your business?

The assessment is free. We'll pull your public data and show you what we see.