They call about revenue being flat. About an ops process that keeps breaking. About marketing that isn't producing. About a sales team that won't close. Those are the words. That's the presenting complaint, and it's always specific and always urgent.

Then we get into it, and it's people. Every time. I have been doing this for twenty years and I cannot name an exception.

Not people as in "people matter." I mean the problem you called me about is not the problem. It's a symptom wearing a costume, and the costume is usually a department.

The five surfaces

Here's where it actually lives. Every time, it is one of these, and usually more than one.

You're not hiring the right people. Not because you're bad at interviewing. Because the role you posted is five jobs taped together, and the only person who ever did all five is you. You wrote a job description for a unicorn, because that's what it takes to run the business without you standing in the middle of it. Then you're surprised nobody fits.

You're not onboarding them in a way that works. You found somebody good, and ninety days later they're underwater and you're privately wondering if you misjudged them. You didn't. You handed a competent adult a job that only exists in your head and expected them to reconstruct it from context clues.

You don't have an organization the right people can assimilate into. This is the one nobody wants to hear. You can hire well and onboard well and still lose them, because there is nothing to join. No clear seats. No standard. No way for work to move from one person to the next without routing through you. Good people do not leave companies over money nearly as often as they leave over having no idea what winning looks like.

You're not building relationships with the people who would refer you. Your best channel has always been somebody saying your name in a room you're not in. Most owners treat that as weather. It isn't weather. It's a system, and it's buildable, and almost nobody builds it on purpose.

And you're not doing good enough work for the people you serve. The uncomfortable one. Sometimes the referrals aren't coming because you haven't earned them. No tactic fixes that, and every tactic sold to you as a fix for that is a tactic that expires.

It's a loop, not a list

Look at the shape of those five. They're not parallel items. They're a circuit.

Who you let in determines what gets built. What gets built determines whether good people can do good work inside it. Whether they can do good work determines whether the work is worth talking about. Whether the work is worth talking about determines who says your name. Who says your name determines who shows up at your door. And who shows up at your door is your next hire and your next client both.

Every one of those handoffs is a person handing something to another person.

Break the circuit anywhere and it degrades everywhere else, which is why this feels like five problems instead of one. You fix the ops process and six months later it's back, because you fixed the artifact and not the circuit. You hire a great salesperson and they leave, and you conclude sales talent is hard to find. You buy the tool and the tool does not stick.

The reason your fixes don't hold is that you keep treating symptoms as systems.

Which brings me to AI

Here's the part that makes this urgent instead of philosophical.

AI is leverage. That's the whole pitch, and the pitch is true. It's the most accessible leverage that has ever existed for a business your size. You don't need a data team. You don't need a million dollars. You need a login and about an afternoon.

Leverage multiplies what is underneath it.

If what's underneath is a clear operation with defined roles and a real standard, AI compounds it. You get speed you could not buy at any price a decade ago.

If what's underneath is a business that lives in your head, AI multiplies that too. It just makes the wrong thing happen faster, at scale, with more confidence, and now with a monthly subscription attached.

I've described AI as an employee who never sleeps and never stops to ask questions. That's the actual danger, and it's not a joke. Every good employee you have ever had has silently corrected for your bad instructions. They saw the process didn't make sense and quietly fixed it. They knew what you meant instead of what you said. That correction layer is invisible, and it is the only reason a lot of undefined businesses function at all.

AI doesn't do that. Point it at an undefined job and it will run the wrong direction faster than any human ever could, and it won't tell you, because it doesn't know.

So: human-first, AI-native. In that order, and the order is the entire argument.

Not because people are precious. Because sequence determines outcome. Fix the circuit, then apply the leverage. Apply the leverage to a broken circuit and you have purchased an accelerant.

That's why more than four in ten of these projects get scrapped. It's almost never the technology failing. It's the wrong solution bolted onto a process nobody ever defined.

The objection I get, and it's fair

"This is just 'people matter.' That's a greeting card."

It would be, if I stopped at people matter. I'm making a narrower claim, and it's falsifiable.

The claim is causal, not sentimental: the problem you named is never the problem, and you can test that yourself without hiring anybody. Think about the last three fixes you made that didn't hold. The process you rebuilt that decayed. The hire who didn't work out. The marketing push that produced nothing. Go one layer down on each and ask what actually broke.

If it was really an ops problem, fixing ops would have worked. It didn't work. It came back. That's evidence, and it's sitting in your own history.

The greeting card version says treat people well. My version says stop debugging the symptom, because the thing you keep fixing is not the thing that's broken. Those are not the same statement and only one of them costs you money to ignore.

"AI is replacing people anyway, so why are you building around them?"

Because that gets the direction backward.

Jobs are moving and reshaping. Some are getting consolidated fast and pretending otherwise helps nobody. But the businesses that win the next decade are not going to be the ones with the most AI. Everyone will have the same models. That's the one thing you can be certain of. The models are commodity and the price is falling and whatever edge you buy this quarter, your competitor buys next quarter for less.

What isn't commodity is an organization where capable people and capable machines can both do their best work. That is hard, it's specific to you, and it can't be downloaded.

The more powerful AI gets, the more valuable the people problem becomes, not less. When everyone has infinite leverage, the only variable left is what you point it at. And what you point it at is a judgment call made by a person inside a structure you either built or didn't.

What to do about it

Pick the process that matters most and write down how it is supposed to run. Who owns it. What done looks like. Not a manual. One page.

Then look at what only you can do, and be brutal about the difference between what only you can do and what you simply have not let go of.

Then break one impossible role into jobs a real person can own and win in.

Then, and only then, ask what a machine can carry.

None of that is clever. It's the boring work, and it's why so few do it. The clever work is the stuff you buy. The boring work is the stuff you own.

The thread

Twenty years in, every conversation lands in the same place. We talk ops, marketing, sales, revenue. We end up talking about people. Who you hire, what you build for them to stand in, who vouches for you, and whether you have earned it.

That's the business. The rest is instrumentation.

AI didn't change that. It just raised the price of getting it wrong.

Where did your last fix that didn't hold actually break?

They ask a chatbot. It gives them three names. They call one.

If your business depends on being found, that should get your attention. For twenty years the path was at least legible. Show up in search, get found, earn the click. It was slow and grindy, but you could see it. You knew what page one looked like. You knew roughly what it took to get there, and you knew whether you were on it.

Now somebody asks a machine who to hire and there's no page one. No list of ten. No comparing tabs. Three names, take it or leave it. And the machine only ever names a sliver of the businesses that exist.

So the question I'm getting from owners right now is the obvious one.

How do I get in front of it? How do I work the new system?

I have been answering a version of that question for twenty years. Back when it was just SEO. The answer has not changed, and I want to walk through why, because the reasoning matters more than the conclusion.

The short version

You can game a system. It works. For a while.

Then it stops working, and everything you built on top of the tactic goes with it. Meanwhile the person who spent those same years getting better at the actual work is still standing there, and now they are the one getting named.

I'm not making a moral argument. I'm not telling you to be virtuous and trust the universe. I'm telling you the math is against you, it's against you structurally, and it always has been.

Here's the part most people never look at.

What the system is actually paid to do

Every discovery layer ever built has had the same job: hand the user the best version of what they are looking for.

Not because the people building it are noble. Because that is the only thing that works.

Google was never trying to be fair to you. It was trying to give people good answers, because good answers hold attention, and attention was the product. Every improvement Google ever shipped to its ranking, every penalty it dropped on link farms and keyword stuffing and content mills, was self interested. A search engine that returns junk loses its users, and a search engine without users has nothing to sell.

So the entire history of SEO is one long story of people finding a shortcut and the system closing it. Keyword stuffing worked, then it didn't. Link buying worked, then it got you delisted. Content farms worked, then Panda ate them. Every one of those tactics had a shelf life, and every one of them was killed by the same force: a company protecting the quality of its results because quality was the business.

The people who did well over the whole arc weren't the ones who kept finding new tricks. They were the ones who kept being worth finding.

Where the models are different

Now here's the part I want to be careful about, because the easy version of this argument is wrong.

You will hear people say the models are running the same play as Google. They're not. Google's alignment was actually a little loose, and it's worth being honest about that.

Attention monetized through advertising is a volume business. A perfect answer on the first try costs Google an ad impression. Good enough to keep you coming back was enough, because for two decades you had nowhere else to go. That gap between "good enough" and "best" is the exact space where gaming lived. It's why the tactics worked as long as they did.

The models are playing a tighter game. Two reasons.

Switching costs are close to zero. You can move from ChatGPT to Claude to Gemini in about four seconds and lose nothing. No index to rebuild. No history you cannot live without. No habit deep enough to hold you. The only thing keeping a user is that the last answer was good. That is a harsher master than search ever had.

And the bigger one: the model's name is on the recommendation.

Google handed you ten blue links and let you sort it out. If the fourth one was garbage, that was your problem for clicking it. The search engine never claimed the fourth link was good. It just put it on the list.

The model doesn't do that. The model gives you three names and vouches for them. When it says "call this one," it just spent its own credibility on you. And if you hire that person and they are terrible, you do not blame yourself for picking wrong. You blame the machine that named them.

That is not an algorithm. That is a referral.

And anybody who has ever made a referral runs the same filter, whether they say it out loud or not: is this going to make me look bad?

That is the whole thing. The model has skin in the game in a way the search engine never did. It's not a new system to crack. It's a referral partner protecting its own reputation, which means it applies roughly the same test your best client applies before they give your name to a friend.

Which is why the way you get picked has not actually changed.

The two objections, and they are both fair

I want to take the counterarguments seriously, because I've gotten both and they're not stupid.

"Ads are coming to the models." Probably right. There is already movement toward shopping and sponsored placement, and I would bet money that some form of paid surface shows up inside these things. So do not let me sell you on the idea that the models are pure. They are companies, and companies monetize.

But it doesn't break the argument, because the argument was never about purity. It is about accountability. Whoever ends up paying the bill, the recommendation still carries the model's name. That's the part that doesn't change. An ad slot is labeled and discounted by the reader. The recommendation is the thing being trusted, and the moment it stops being trustworthy the product is dead. They know that better than we do.

"AI SEO is already an industry." Also right. There is a whole cottage business selling GEO and AI visibility optimization right now, and some of it works today.

So I won't tell you it's ungameable. That would be a lie, and you would catch me in it within a week.

What I will tell you is narrower and, I think, sturdier: it's gameable right up until it isn't. Correcting for manipulation is the one thing these systems are built to do forever, because their existence depends on it. Every dollar of that correction budget is aimed at the shortcut you just bought. You are renting an advantage from a landlord who is actively working to evict you.

That's a fine trade if you know that's the trade. Most owners buying it think they're building something.

So what do you actually do

This is where I lose people, because the answer is not clever.

Be worth talking about. Not a slogan. The actual work. Do the job so well that a normal person, unprompted, mentions you to someone else. Everything downstream depends on this, and there's no way around it. If the work is mediocre, the rest of this is decoration.

Make the proof exist somewhere other than your own website. Your site is you talking about you. Nobody weighs that heavily, and neither does a model. Real customers saying real things in public, on platforms you do not control, is the thing that carries. Most owners I meet have done real, good work for a decade and have almost no public record of it. That's not a marketing problem. It's a filing problem.

Be consistent everywhere someone might look. Same story, same offer, same name, same claims. Inconsistency reads as risk, to a person and to a machine.

Then run the test yourself. If someone recommended you tomorrow, would it make them look good? Would it make them look smart for knowing you?

If the answer is no, no tactic fixes that. If the answer is yes, you are already doing the durable version and the pipe does not much matter.

The part I keep coming back to

Do good work and people talk about it. They leave feedback nobody had to beg them for. That feedback gets seen. The people you want find you.

It's slower than the shortcut. It's always been slower than the shortcut. That's the trade, and it's the same trade it was in 2006, and the people who took it then are the ones who are still here.

Reputation was always the moat. The pipe changed. The moat didn't.

Who's the last person you recommended without being asked? And what made you do it?

That's the part wearing owners down right now. Not that prices are high. That the number keeps moving under your feet. New tariff rates went out this week. Different countries, different amounts, more on the way, and half of it changes again next month.

If you buy or build anything that crosses a border, you're trying to run a business on a cost that won't sit still.

The waiting trap

So you do the natural thing. You wait. Hold the order, the hire, the expansion, until things settle.

Here's the hard part. They're not going to settle. Not on your timeline. Waiting for stable conditions has quietly become a strategy of never moving, and the cost of never moving doesn't show up on any invoice. It shows up as the competitor who moved anyway.

I understand the instinct. Every owner I know is carrying some version of the same spreadsheet with three scenarios in it, and all three are guesses. The instinct says: get better guesses.

The instinct is wrong. Nobody is calling this right. The people paid millions to forecast policy are getting it wrong month over month. You will not out-forecast them from your kitchen table, and you don't have to.

What the durable operators have

The owners getting through this aren't the ones who called the policy right. They're the ones who can move when it shifts.

They can see their real margin this week, not find out three months from now that it got eaten. Their books are current, their job costing is real, and a cost change shows up in days, not quarters.

They can reprice or reroute in days, without every call running back through the owner. Someone owns pricing. Someone owns sourcing. The decision rules are written down, so when the input cost jumps 12 percent, the response doesn't wait for the one person who's been in meetings all week.

They keep slack in the system on purpose. Not fat. Slack. A second supplier who already knows them. Terms that don't snap under one bad month. Enough margin of maneuver that a surprise is a Tuesday problem instead of an existential one.

That's not a better forecast. It's a better operation.

The reframe

Stop trying to predict the weather. You're not in the weather business.

When you can't predict the weather, you build something that runs in any of it. The forecast-proof business isn't the one that saw the storm coming. It's the one where the storm hits and the work keeps moving anyway.

What did the last few months cost you that you can't even see on paper yet?

That gap, between what you're paying and what you feel safe charging, is the quiet thing squeezing owners right now. For the first time in a while, cash flow has passed inflation as the number one worry I hear when I sit down with owners.

Here's what most owners do with it. They stare at the price. Raise it, hold it, or eat the hit. All three hurt, and the debate between them can eat a whole quarter while the margin keeps leaking.

But price is only one lever, and right now it's the hardest one to pull.

The levers you actually control

The levers that don't require a customer's permission sit inside the business. Nobody votes on them but you.

How fast you get paid. If your average invoice takes fifty days to collect and your suppliers want payment in twenty, you're running a bank for your customers and paying for the privilege. Tightening that cycle is margin nobody can compete away.

How much of your week leaks into work that was never supposed to be yours. Every hour the owner spends doing a $25-an-hour task is an hour not spent on the three decisions that actually move the number. That leak has a dollar value. Most owners have never calculated it, because the answer is uncomfortable.

The jobs sitting half done because every call still routes through you. Half done work is cash you've already spent that hasn't turned back into money yet. The faster work moves through the system without waiting on you, the faster the same team produces the same revenue with less strain.

The uncomfortable comparison

I've watched owners find more margin by fixing their cash cycle and their systems than they ever would have squeezed out of a price increase nobody wanted to pay.

The price increase risks customers and takes one conversation. The operational fix risks nothing and takes discipline. Owners keep choosing the first one because it looks like one decision, and the second one looks like work.

It is work. It's also yours to keep. A price increase can get matched or shopped. An operation that collects faster, wastes less, and moves work without friction is invisible to your competitors and permanent for you.

When you can't charge your way out, you operate your way out.

Where is your business leaking money you're not even looking at?

Here's the part they leave out.

Analysts expect more than four in ten of these AI projects to get scrapped within a couple of years. Paid for, launched, then quietly shelved after the money is spent. You can see it already. Some of the companies that went loudest on AI a year ago are quietly walking it back.

I see a version of this up close, from two directions.

The two failure modes

Founders offer me equity and an advisor seat to help build their AI idea. I pass on most of them. What they're building is usually something the next version of ChatGPT or Claude just does natively. A feature waiting to get absorbed. The product isn't durable because it's a thin layer over someone else's model, and the someone else keeps shipping.

Same story one level down, inside operating businesses. Fly-by-night AI consultants are selling owners agent frameworks and automations with no real grasp of how a business actually runs. It works for a month. Then it breaks, needs constant babysitting, or gets ripped out. Money you don't get back, and worse, a team that's now allergic to the next attempt.

Two different failure modes, one root: something bolted on where nothing was defined underneath.

Why the tech isn't the problem

It's almost never the tech that fails. The models are good and getting better on a schedule you can nearly set your watch to.

It's that the wrong solution was bolted onto a broken process.

AI is like having an employee who never sleeps and rarely stops to ask questions. That sounds like a feature until you sit with it. Every good human employee quietly corrects for your bad instructions. They see the process doesn't make sense and fix it without telling you. AI doesn't. Point it at a job that hasn't been defined and it just runs the wrong way faster, at scale, with confidence.

The durable version

Here's how we think about it. Human first, AI native.

Fix the organization first. Get the direction and the strategy clear. Build a framework solid enough that AI can be woven into it, not bolted onto it. Woven means the machine sits inside a defined process with a defined owner and a defined standard. Bolted means it sits on top of a mystery and automates the mystery.

The owners getting real results do the boring part first. They pick one process that matters. They write down how it's supposed to run. Who owns it. What done looks like. Then they explore handing what they can to a machine.

The AI doesn't make them the money. The clarity does. The AI just runs it faster.

That's the game right now. Not who sprinkles the most AI on top. Who has an operation clear enough that AI has something solid to stand on.

You can't automate a process that only lives in your head.

Where have you added a tool, AI or otherwise, when what you really needed was a clearer process?

Either they've decided it doesn't apply to their business, so they ignore it.

Or they bought a few tools and they're winging it with no real plan.

I see both every week, and I want to be fair to both, because each one comes from a reasonable instinct that's aimed at the wrong target.

The ignorers

The owners avoiding it aren't wrong that half the hype doesn't fit them. It doesn't. If you run a plumbing company or a machine shop, most of what gets shouted about AI online has nothing to do with your Tuesday.

Where they're wrong is concluding that none of it does. Somewhere in their week, and usually somewhere in every employee's week, is work a machine should already be handling. The quote that takes forty minutes to assemble from three systems. The follow-up emails that don't get sent. The report someone builds by hand every Monday.

Ignoring the category because the hype is annoying means paying people to do machine work indefinitely. The hype being wrong doesn't make the tool useless. It just makes the loudest advice useless.

The wingers

The owners winging it have the opposite issue. Tools open, subscriptions active, nothing underneath.

They bought the thing because they felt the pressure to do something, and now AI helps them do the busywork faster instead of clearing busywork off their plate entirely. That's the tell, by the way. If the tool made you faster at a task you shouldn't be doing at all, it didn't help you. It anesthetized you.

Winging it feels like progress because there's motion. But pointing powerful tools at an undefined operation just produces undefined results at a higher speed.

The question that fixes both

Here's the thing. Neither one is a tool problem, so no tool fixes it.

The thing that actually moves the needle isn't the model. It's the call about where to point it. And that call comes down to one question you can answer with a legal pad:

What in your business can only you do, and what have you just never handed off?

Sit with that honestly and the two lists come apart fast. The "only I can do this" list is shorter than you think. Everything else is a candidate for a person or a machine, and once the job is defined, the tools almost pick themselves.

The ignorers discover there's real work worth handing over. The wingers discover the reason the tools weren't landing. Same question, both cured.

So which one are you right now: ignoring it, or winging it?

Big companies have spent the year laying people off. That means the hiring pool for a small business is better than it's been in years. People who never would have looked twice at your job posting three years ago are looking now. Operators, marketers, finance people, builders. Real ones.

Here's the catch nobody mentions.

It's still expensive to hire. This isn't a cheap-labor moment, and the people worth having know what they're worth. And great people don't stick around in a messy company. They've seen what good looks like. That's exactly what makes them valuable, and it's exactly why they won't tolerate chaos for long.

How the win becomes a loss

I see this play out constantly. An owner wins the hire, celebrates, and loses the person six months later.

Not because the person was wrong. Because there was no real system for them to step into. No clear decisions. No actual delegation. The owner handed them a title and kept all the levers. Just chaos with a new face in it.

And here's the quiet cost that never shows up on a spreadsheet: that person talks. They go back into the market and tell other good people exactly what it was like inside your company. The hire you fumbled doesn't just cost you a salary and a restart. It costs you the next three candidates you never hear from.

What "ready to receive" looks like

The advantage right now isn't more hiring. It's having a business that's actually built to receive good people.

Roles that are real, not five jobs taped together under one title. Decisions that don't all run through you, so a capable adult can actually use their judgment. A way to bring someone in, show them what winning looks like, and get out of their way. A team someone talented would want to stay on, because the work moves and the wins are visible.

None of that requires a big company budget. It requires you to write down the things that currently live in your head, and to actually hand over the levers you keep saying you want to hand over.

The order of operations

Fix the organization first. Then go get the talent that's finally available.

Do it in the other order and you're renting great people for six months at full price, plus the reputation cost when they leave.

The market handed small business a gift this year. Most owners will waste it on hires their business can't hold. The ones who do the boring work first get to keep what they catch.

What would your business look like to someone great walking in today?

I've heard some version of this more times than I can count.

You're paying a full team and still staring at half finished work everywhere. Projects that stall. Stuff that only moves when you touch it. From your seat, it looks like nobody is stepping up.

Here's what's actually going on.

The plan is in your head. All of it. You expect people to see what you see and just run with it. But they're not you. You never spelled it out, so the big things sit and wait for you.

The flinch

Then someone says the word "structure" and you flinch. I've watched it happen in real time.

It feels like red tape. Like the corporate nonsense you started this business to get away from. The thing that slows you down, boxes you in, and turns your scrappy company into a place where people fill out forms instead of doing work. So you keep it loose and tell yourself that's the nimble way to run.

It's not. Loose means nobody really owns anything. There's no bar to hit, so people fall short of a standard you never actually set. Then you're frustrated when they miss it, and they're frustrated because they can't figure out what winning looks like, and eventually the good ones stop trying to guess and go somewhere that tells them.

You can't hold people to a standard that only lives in your head.

What structure actually is

Strip the word of its baggage and structure is three things.

Real roles, so people know what's theirs without asking. Clear outcomes, so they know what finished looks like without a check-in meeting. And a way for work to move from one person to the next and actually get done, so it doesn't route back through you every time somebody finishes a step.

That's it. Not a bureaucracy. One page per process, if that.

Notice what none of that does: none of it slows you down. Every piece of it removes a wait. The waits were you. The approval that sat in your inbox. The question only you could answer. The job that stalled because two people each thought the other one had it.

You think structure will box you in. It's actually the one thing that finally lets you step back.

The tell

Here's how you know this is your problem and not a hiring problem.

The same issue keeps showing up with different people in the seat. You've replaced the role twice and the work still stalls in the same place. At some point the common variable isn't the people.

When the same problem keeps showing up with different faces, it stopped being about the faces a while ago.

How many good people do you have to lose before it stops being about the people?

Then they go to sell it and find out it isn't.

A lot of owners are heading for the exit right now. More than half of the people who own small businesses in this country are past 55, and plenty of them want out in the next few years. Which means a lot of hard conversations are coming, and most of them will happen in a broker's office about two years too late.

Here's what nobody warns them about.

You can't sell a job. And that's what a lot of these businesses are. The owner is the product, the sales team, the quality check, and the reason anything ships on time. Pull them out and there's nothing left to buy.

Buyers catch it fast. They don't need a forensic audit to see it. They ask who your biggest customers talk to when something goes wrong, and the answer is you. They ask who sets pricing, and the answer is you. They ask what happens when a job goes sideways at 6pm on a Friday, and the answer is you, again, every time.

If the whole thing lives in your head, they're not buying a business. They're buying your calendar. So the offers come in low, or they come with an earnout that keeps you chained to the desk for three more years, or they never come at all.

The math nobody runs

Here's the part that stings. Two businesses with identical revenue can sell for wildly different numbers, and the difference usually isn't the product, the market, or the margins.

It's whether the thing runs without the person who started it.

A business that runs on systems gets valued like an asset. A business that runs on the owner gets valued like a job with overhead. Same top line. Very different check.

And the discount isn't small. It's the difference between funding the retirement and going back to work at 64.

The fix is the same either way

Here's the good news, and I mean it as good news even though it's work.

The fix for the exit problem is the same work that makes the business better to run today. Real roles, so the team isn't waiting on you. Decisions that don't all route through your phone. Systems a stranger could step into and follow without an archaeology project.

Do that work and you get two things. A better life now, because you're no longer the load-bearing wall in your own company. And something actually worth selling later, whenever later comes.

You don't have to want to sell. Plenty of owners never will. But a business that could be sold is just a better business. It runs cleaner, it survives a bad month, and it doesn't collapse when you take two weeks off.

Build it like you'll sell it. Run it like you won't.

If you vanished for a month, what would your business be worth?