For about three years I tried to learn marketing from books.
Good books, a stack of them on the shelf. I read carefully, took notes, underlined the parts that felt important. And almost none of it moved. I could describe a positioning statement. I could not position anything. I knew the word “funnel” and could draw a clean one on a whiteboard. The funnel in my actual business stayed broken the whole time.
Then I started doing it. Not reading about it - doing it. Writing the thing, putting it in front of a real person, watching what actually happened, changing it, putting it out again. And it started to work. Slowly, unevenly, with a lot of getting it wrong first. But it moved.
Here is the honest version, because it matters for everything below. The books probably helped. When I finally practiced, I had vocabulary, I had frames, I could recognize what I was looking at instead of staring at it blank. The books were not useless. They were the floor I was standing on. But standing on a floor is not the same as walking somewhere. The years I lost were the years I quietly believed that reading was the work.
I think about this a lot now, because there is a harder question sitting under it. If I were raising a kid today - someone who will be looking for their place in the world ten or fifteen years from now - what would I actually tell them to learn?
What AI took
Start with what is changing, because it changes the answer.
For most of the last few decades, the path to being valuable ran through execution. Learn to write the code. Learn to do the math. Learn to draft the contract, build the model, render the design. Execution was the hard part - the part that took years, the part that separated the people who could from the people who couldn’t. We even named it. We called it “hard skills,” and hard meant both difficult and load-bearing.
AI is eating that part. Not all of it, and not evenly - I will come back to where it stops. But the floor of execution, the entry-level version of nearly every craft, is collapsing in price. A model writes the first draft of the code, the contract, the campaign. The thing that used to take a junior person two years to get passable at, a model now does passably in seconds.
Software always automated work. The difference this time is what it automates. It used to take over the repetitive doing. Now it takes over the skilled doing - the part we were proud of, the part we charged for.
I had a conversation last month with a colleague, an anthropologist I work with on research. At one point, independently, in the same second, we both said the same phrase: “the hards are there, the skills aren’t.” We both meant hard skills and soft skills. But the language had already split them for us. Hard skills had shrunk to just “hards” - the predictable apparatus, the part you can look up. Soft skills kept the word “skills,” as if the language itself had decided which of the two was the real one.
That throwaway phrase is the whole argument. If the hards are being commoditized, the question for a kid gets very simple: what is left that stays scarce?
What is left is the ability to make another human being want something, trust you, and act. That is marketing. That is sales. And almost nobody teaches a kid either one on purpose.
They are not the same skill
People say “marketing and sales” in one breath, as if it were one long word. They are two different muscles, and treating them as one is the first mistake.
Marketing is one-to-many. You are speaking to a crowd you cannot see, asynchronously, and you will never watch a single individual react. The question marketing answers is: why should a whole category of people care about this at all? Its raw materials are attention, framing, positioning, narrative. Its feedback is slow and statistical - you learn from cohorts and numbers, weeks after the fact.
Sales is one-to-one. There is a specific person, in front of you, right now. The question sales answers is narrower and sharper: why should you, this person, act on this today? Its raw materials are reading a face, hearing the thing that was not said, handling the objection that actually exists rather than the one in the script, and closing. Its feedback is instant and high-bandwidth - a pause, a glance away, a shift in tone.
| Marketing | Sales | |
|---|---|---|
| Who | A crowd you can’t see | One person in the room |
| When | Asynchronous, no live reaction | Synchronous, reaction right now |
| The question | Why should a category care? | Why should you act today? |
| Core muscle | Framing, positioning, narrative | Reading people, objections, closing |
| Feedback | Slow, statistical, weeks later | Instant, high-bandwidth, in the face |
One line holds it together: marketing is framing for strangers, sales is adapting to the person in the room.
Here is why that gap is real and not just a tidy definition. The two jobs share a goal - make someone want this and act on it - but their feedback runs on opposite clocks. Marketing feedback arrives late, aggregated, and stripped of the individual: a number, a cohort, weeks after you committed to the framing. Sales feedback arrives now, specific, personal - a face shifting while you are still mid-sentence. And feedback latency is what shapes a skill. Long latency forces you to think like a forecaster: get the framing right before anything comes back, because nothing is coming back for a while. Zero latency lets you think like an improviser: be wrong, watch it land badly, and fix it inside the same minute.
That is why the strong marketer and the strong closer are usually two different people. One spent years training the reflex to commit and wait. The other spent years training the reflex to adjust and recover. You cannot run both at full strength in the same moment. A kid should know these are two skills, will probably feel which reflex comes more naturally, and that pull tells them something real about themselves.
”But AI does marketing and sales too”
Here is the obvious objection, and it is the right one to raise. AI is already writing copy, scraping leads, sending outreach, booking meetings. If AI does marketing and sales, why teach a kid to do them?
Because the same split runs straight through marketing and sales themselves.
There are hards inside marketing - generating variations of copy, building lists, running the mechanical A/B grind. AI is taking those, quickly. And there are skills inside marketing - knowing what a moment means to the people living through it, judging which framing will land in this culture, this week, with these people. AI has no read on the room. It can produce a thousand competent sentences and not one of them will know what it feels like to be the reader.
Sales is the same story. The fully autonomous AI SDR - deploy the bot, remove the humans - has mostly not worked. The emails come out generic, and buyers have learned to smell them and filter them before they are read. What works is the opposite shape: AI does the research and the first draft, a human carries the judgment and the actual relationship. The people winning at outbound are not doing less marketing and sales. They are doing the judgment part of it and letting the machine do the mechanics.
So the answer to “AI does this too” is not to retreat. It is to aim. Do not teach a kid the mechanical layer that is already a commodity. Teach the judgment layer - reading people, framing, earning trust - the part that is getting more valuable precisely because the mechanical layer got cheap.
Why you can’t read your way in
Now the part I learned the slow way.
Why can’t a kid - or an adult - simply learn this from books?
It is not that the books are wrong. It is that marketing and sales both live in a place a book cannot take you: the zone where there is no correct answer printed at the back.
Think about what school trains. School is a long sequence of problems that each have one right answer, and the answer is in the back of the book. You converge on it, you get the mark, you move on. This is genuinely useful. It builds literacy and numeracy, the floor that everything else stands on, and I am not making an argument against school. But it trains one very specific reflex: there is an answer, your job is to find it.
Marketing and sales punish that reflex. There is no answer at the back. There is a person, or a crowd, and they will not tell you the truth - not because they are lying, but because people are genuinely bad at articulating what they want. They will give you a clean, confident reason that is not the real reason. The channel is noisy. The real signal is a hesitation, a thing left unsaid, a glance. A book can hand you a script. The human in front of you has not read the script and never will.
So the skill is not knowledge. It is calibrated intuition. And intuition calibrates exactly one way: reps, plus feedback, plus being wrong out loud in front of someone. A book gives you the map. It is a real map, worth having. But it cannot give you the territory, and it certainly cannot give you the thousand small corrections that only come from walking the territory and getting it wrong.
This is why marketing and sales sit closer to playing an instrument or jumping out of a plane than to a subject you study. You can read everything ever written about freefall. The skill shows up on the third jump, not in the chapter. You can read everything ever written about persuasion. The skill shows up after the tenth time someone says no to your face and you have to decide, in that second, what happens next.
None of this is an argument against reading. I would still hand a kid the books. The books made my practice faster once I finally started practicing - they were necessary. They are just nowhere near sufficient, and the failure mode, my failure mode, is mistaking the necessary part for the whole thing.
How a kid actually trains it
So what does this look like, concretely? Not a curriculum. Reps.
A lemonade stand is not a cute weekend activity. It is a kid setting a price, watching people walk past, and discovering in a single afternoon that the sign mattered more than the lemonade. That is marketing and sales in one sitting, with the feedback arriving in minutes instead of years.
Reselling something works the same way - buy a thing, describe it, price it, talk to a buyer who pushes back on the price. That is a complete loop. So is pitching: let a kid make the actual case for the trip, the later bedtime, the thing they want, and make them handle the “no” instead of escalating to a tantrum. So is any school project where the real task is getting other people to care about something and join in.
The common thread is small, real stakes and fast feedback. A kid needs to put something in front of a person who is allowed to say no, hear the no, and go again tomorrow. That is the rep. You cannot assign it as homework. You can only leave room for it and resist the urge to rescue them from the no.
I am not telling you AI makes school worthless, or that every kid should grow up to start a company. Most of this matters even for someone who never founds anything - a career is mostly the marketing and sales of one product, yourself, repeated for decades.
What I would tell a kid is narrower and harder than a subject. The execution is being handed to the machines. What stays scarce is being able to make another person want something, and trust you enough to act on it. You can read about that. I did, for years, on a comfortable shelf full of good books. You will learn it the afternoon you put something in front of someone who can walk away - and then set the stand back up the next day.