Growth & marketing fundamentals

A few years ago I watched a startup raise money, hire a “growth hacker,” and burn most of it in four months. Smart people. Big budget. They had every tool. What they didn’t have was the boring stuff: a clear idea of who the customer was, where those customers actually hung out, and a way to tell a real win from a lucky week. The hacks went out. Nothing stuck. The tools weren’t the problem. The fundamentals were missing.

I keep coming back to that, because in 2026 the exact same trap has a new coat of paint. Swap “growth hacker” for “AI.” Same story. People buy the tools, skip the basics, and wonder why the output is fast and useless.

So let me say the plain version first, before any tool talk. Growth marketing is running small experiments across the whole customer journey to find what actually grows the business, then doing more of what works. That’s it. Not a department. Not a campaign type. A way of working: test, measure, keep the winners, drop the rest. Marketing is the wider craft underneath it: knowing your customer, picking the right channels, and saying something they care about. Growth marketing is just that craft run as a tight loop instead of a once-a-year plan.

And here’s the part the AI hype gets backwards. AI changed the tools, not the fundamentals. The job is still finding what works and doing more of it. AI just lets you run more experiments, faster and cheaper than before. The people winning with AI right now are almost all people who knew the fundamentals first. They had the loop. They bolted a faster engine onto it. Bolt a fast engine onto no loop and you just crash quicker.

The numbers say everyone has the engine now. In recent survey data, around 88% of marketers use AI tools daily and roughly 9 in 10 say they’re actively using it in their work. So the tool isn’t the edge any more. Everyone has the same ChatGPT. The edge is whether you know what to point it at, which is exactly what the fundamentals teach you.

This page is the base layer for everything else on the site. It’s the part you read before the AI-specific stuff makes sense. I’ll define the real terms plainly (growth marketing, growth hacking, the funnel, the channels) so a beginner can follow, then put a practitioner’s opinion on top, because a textbook recap helps no one. Where there’s a deeper essay, I’ll point you to it. Where there isn’t yet, I’ll say so.

What is growth marketing, really

Most definitions of growth marketing are technically true and totally useless. “Data-driven, full-funnel, iterative.” Sure. That’s like defining cooking as “heat-based food transformation.” Correct, and it teaches you nothing.

Here’s the version I’d give a friend. Growth marketing is the job of growing a business by running lots of small tests across the entire customer journey, then putting more money and time behind the ones that win. The phrase “entire customer journey” is the load-bearing part. Old-school marketing mostly cared about the top of the funnel: ads, awareness, getting people in the door. Growth marketing cares about the whole thing. Do people find you, do they sign up, do they stick around, do they pay, do they tell someone. A win at the top that leaks out the bottom isn’t a win.

The reason this matters is dull and important: most experiments fail. Look at any serious testing program and the win rate is brutal. Across one large analysis, only about 1 in 7 A/B tests is a clear winner, and a big chunk of the rest land inconclusive. So roughly six out of seven things you try won’t move the needle. That sounds depressing until you flip it. The whole job is running enough tests that the winners, rare as they are, compound. You’re not trying to be right every time. You’re trying to be right often enough, cheaply enough, that the wins stack up.

My take: A lot of people think “growth marketing” is a fancier word for “marketing.” It isn’t. The difference is the loop. A regular marketing plan is a guess you commit to for a quarter. Growth marketing is a guess you test in a week and kill on Friday if it’s wrong. Same skills, completely different metabolism. The first one bets big and slow. The second one bets small and fast, a lot.

For the long version of this, with the frameworks and the hiring angle, read what is growth marketing. It’s the practitioner’s definition, including why most experiments fail and what companies actually test for when they hire a growth person. Everything on this page sits on top of that essay, so it’s the natural next click.

Growth marketing vs growth hacking vs traditional marketing

This is where people get tangled, so let me untie it.

“Growth hacking” came first as a term. Sean Ellis coined it in 2010 while he was advising a wave of early startups, including Dropbox. He defined a growth hacker as “a person whose true north is growth.” Notice he avoided the word “marketing” on purpose. Marketing meant a department and a budget. Growth meant a single goal that cut across marketing, product, and engineering. The famous example is Dropbox’s referral program: give free storage to both sides of an invite, and every happy user brings in two more. That’s not a trick. It’s a clean experiment that happened to win big.

Then the term got hijacked. By the mid-2010s “growth hacking” came to mean clever loopholes and shady tactics: fake countdown timers, pre-ticked boxes, dark patterns to juice signups. A lot of it worked short-term and aged badly. So the serious people quietly dropped the phrase and started saying “growth marketing” instead. Same job. Less baggage. If you hear someone say “growth hacking” in 2026 with a straight face, they’re usually selling a course.

And traditional marketing is the wider, older craft both of these grew out of. Brand, positioning, channels, the message. It’s not worse. It’s slower and broader. Growth marketing didn’t replace it, it put a testing loop on top of it.

Here’s the honest comparison, side by side:

The pattern across that table: it’s the same underlying job getting more disciplined over time. Tricks became systems. Signups became retention. One clever person became a repeatable loop. The deeper history of how the name changed, and why it matters for how you work, is in what is growth marketing and across the growth marketing and growth hacking essays.

My take: The thing people miss is that a “growth hack” was always just a fast experiment that happened to win. There was never any magic in it. Dropbox’s referral loop wasn’t a hack, it was a good idea, tested cheaply, that paid off. AI doesn’t change that one bit. It just means you can run ten of those experiments in the time it used to take to run one. So the operators who’ll look like wizards in 2026 aren’t the ones with the cleverest trick. They’re the ones running the most tests with a clear head about what counts as a win.

The experiment loop, step by step

Strip everything else away and growth marketing is one loop, run over and over. Here it is, plain:

1. Idea. A guess about what might grow the business. “If we change the headline, more people will sign up.” It should be specific enough to be wrong.
2. Build it small. Make the smallest version you can actually test. Not a rebuild, a variant. One headline, one email, one landing page. The cheaper the test, the more of them you can run.
3. Measure honestly. Run it long enough to trust the result, and judge it against what would’ve happened anyway. This is the step people fudge, because we all want our idea to win. The data doesn’t care.
4. Double down or drop. If it won, do more of it and bank the learning. If it lost, kill it and write down why. Both are useful. A dead idea you understand is worth more than a live one you don’t.

Then you go back to step one with what you learned. That’s the loop. Run it fifty times a year instead of five and you’ll out-grow a smarter team that only plans.

Where the loop breaks for most people is step three. Honest measurement is hard, partly because the win rate is so low. When most of your tests come back inconclusive or losing, the temptation is to call a fluke a win and move on. Don’t. A coin flip looks like a streak half the time. The discipline of saying “that didn’t actually work” is most of the skill.

The loop again as a table, with what AI changes in each step:

Look at the right column. AI makes steps one, two, and three dramatically faster and cheaper. That’s huge, because the slow, expensive part of experimentation was always building the tests and reading the results. But step four, the decision, stays human. AI will happily tell you a fluke is a winner if you ask it nicely. Deciding what a result means, and whether the experiment was even worth running, is judgment. That doesn’t come in a model.

The channels and the funnel (the map everyone skips)

Before any of the AI stuff, you need two mental maps. They’re old, they’re unglamorous, and skipping them is why so much “growth” work is just noise.

The funnel. People don’t go from stranger to customer in one jump. They move through stages: they become aware you exist, they get interested, they consider you against the alternatives, they buy, and if you’re good, they stick around and tell others. Marketers have drawn this as a funnel for over a century. The AIDA model (awareness, interest, desire, action) dates to 1898. The growth-era version everyone uses is Dave McClure’s AARRR, the “pirate metrics,” from 2007: acquisition, activation, retention, referral, revenue. Different letters, same idea. The point of the map is simple: you can’t fix what you can’t locate. “Sales are down” is useless. “People sign up but never come back” tells you exactly which stage to test.

The three channel types. Every way you reach a customer falls into one of three buckets, and the classic split is owned, earned, and paid:

• Owned is what you control: your site, your email list, your blog. Slow to build, yours forever, doesn’t disappear when you stop paying.
• Earned is what others give you: word of mouth, press, someone sharing your post. The most trusted, the hardest to force.
• Paid is what you rent: ads. Fast, predictable, and it stops the second you stop paying.

Most beginners pour everything into paid because it’s the fastest to start. Then they’re shocked when the growth vanishes the day the budget does. The operators who win build owned channels that compound while using paid to speed things up, not to replace them.

My take: If I had to teach one thing to a beginner, it’d be this map, not a tool. Almost every “our marketing isn’t working” problem is really a “we don’t know which funnel stage is leaking” problem, dressed up. You can have the best AI stack on earth and still be pouring water into a bucket with a hole in the bottom. Find the hole first. The map shows you where to look.

This is the territory digital marketing fundamentals covers. If you want to go from “I have a vague plan” to “I have a real one,” the AI marketing strategy generator essay is a good practical next step. It walks through what an AI tool actually gives you (a skeleton: audience, channels, a content calendar) and, just as importantly, the strategic thinking it can’t give you, which is exactly the funnel-and-channel judgment we just covered.

What AI changes, and what stays the same

This is the heart of the whole page, so let me be blunt about it.

The hype says AI is a revolution that rewrites marketing from scratch. It isn’t. AI is a massive speed-up on the parts of marketing that were always slow and boring: research, first drafts, pulling data, sorting keywords, writing the follow-up. The thinking parts, the parts that decide whether any of it works, are exactly where they were.

You can see this in the data. Nearly everyone now has AI, but almost nobody gets real results from it. McKinsey’s 2025 State of AI survey found that while most organizations now use AI somewhere, only around 5.5% see meaningful bottom-line impact from it. Near-universal access, rare results. If the tool were the answer, that gap couldn’t exist. The gap is the fundamentals. The teams getting results know what to test, which customer to chase, what a real win looks like. The tool just lets them do it faster.

The split, laid out:

The left column is the 80% you should hand over today. The right column is the 20% that’s still your entire job, and probably always will be. If a tool promises you the right column, it’s selling you something.

My take: “AI replaces marketers” is the lazy frame, and it’s wrong in a specific way. AI replaces tasks, not the judgment that picks which tasks are worth doing. The marketers who lose to AI are the ones whose whole job was the left column, the drafting and the formatting and the data-pulling. The ones who win move up to the right column and use AI to do the rest at scale. So the real question was never “will AI take my job.” It’s “am I still doing work a model can do for a dollar?” Learn the fundamentals and you live in the right column. That’s the whole game.

What a beginner should learn first

If you’re starting out, the internet will scream at you to learn fifty AI tools. Ignore that. Here’s the order I’d actually learn things in, because each one makes the next one make sense.

1. The funnel. Learn the stages a customer moves through (aware, interested, considering, buying, staying). Everything else hangs off this map. Without it you can’t even say what’s broken.
2. The three channel types. Owned, earned, paid. Know the trade-off: owned compounds slowly and is yours, paid is fast and rented. Most beginner mistakes are a channel-mix mistake.
3. One real customer. Pick the actual person you’re trying to reach and learn them properly. Not a demographic. A person, with a problem, in a place you can find them. This is the input every AI tool needs and the one it can’t invent for you.
4. The experiment loop. Idea, build it small, measure honestly, keep or kill. Run it once, by hand, start to finish, on something tiny. You’ll learn more from one real test than from ten courses.
5. Then, and only then, the AI. Once you have the map, the channel, the customer, and the loop, AI becomes a force multiplier. Point it at the grunt work and keep the judgment. Before that, it just helps you make mistakes faster.

The reason this order matters: AI is leverage, and leverage multiplies whatever you point it at. Point it at a clear plan and it’s a superpower. Point it at confusion and it scales the confusion. The fundamentals are what make the leverage worth having.

If you want a practical on-ramp to step three and four, the AI marketing strategy generator piece shows how to use a tool to get a first draft of a plan fast, then where you have to add the real thinking yourself. It’s a good way to see the “AI gives you the skeleton, you add the strategy” idea in action.

Frequently asked questions

What is growth marketing, in plain terms?

Growth marketing is running lots of small experiments across the whole customer journey to find what actually grows the business, then doing more of what works. It’s a way of working, not a department or a campaign type: you test an idea, measure it honestly, keep the winners, and drop the losers, over and over. The “whole journey” part matters, it covers not just how people find you but whether they sign up, stick around, pay, and refer others. Most experiments fail, so the real job is running enough of them, cheaply enough, that the rare winners compound.

What’s the difference between growth marketing and growth hacking?

There’s almost no real difference in the work, only in the name and the baggage. “Growth hacking” came first (coined by Sean Ellis in 2010) and originally meant exactly what growth marketing means now: fast, data-driven experiments to grow a business. The term got hijacked to mean clever loopholes and shady tactics, so serious practitioners dropped it and started saying “growth marketing” instead. So if someone draws a hard line between the two, be a little skeptical. A “growth hack” was never magic, it was just a cheap experiment that happened to win.

Do marketing fundamentals still matter in the AI era?

Yes, more than ever, because AI changed the tools, not the fundamentals. The job is still finding what works and doing more of it, knowing your customer, picking the right channels, and saying something they care about. AI just makes the busywork (research, drafts, data) fast and cheap. The proof is in the data: nearly everyone now uses AI, but only a small slice get real results from it, and the difference is the operator’s grasp of the basics. Point a powerful tool at a clear plan and it’s a superpower; point it at confusion and it just scales the confusion.

How do you use AI in digital marketing without losing the basics?

Start with the fundamentals, then add AI as a speed-up, not a substitute. First get clear on the funnel stage you’re trying to fix, the channel you’re using, and the actual customer you’re reaching. Then point AI at the grunt work inside that plan: drafting variants, researching, reading data, repurposing content. The rule that keeps you honest is to keep the judgment human, AI should do the doing while you keep the deciding (which experiment to run, what the result means, what to do next). If you skip the basics and hand AI the strategy, you get fast, confident, generic work that doesn’t move anything.

What should a beginner learn first?

Learn the funnel, the three channel types (owned, earned, paid), and how to run one small experiment end to end, in that order, before touching any AI tool. The funnel tells you where things break, the channels tell you how to reach people and what each one costs you, and the experiment loop teaches you to tell a real win from a lucky week. Pick one real customer and one tiny test, and run it by hand once. You’ll learn more from that than from any tool tutorial, and it makes every AI tool you add afterwards actually useful instead of just fast.

Can AI replace a marketer?

No. AI can do most of the tasks inside marketing (research, first drafts, data pulls, repurposing) but it can’t do the job, because the job is judgment: who to target, what to say, which bet to make, when to change course. AI has no point of view and no stake in the outcome, so it’ll happily produce a polished campaign aimed at the wrong people. The honest model is AI as a very fast, very cheap junior that needs a clear brief and a human editor. The marketers at risk are the ones whose whole job was the busywork; the ones who learn the fundamentals and use AI to scale them are safer than ever.

Want a second pair of eyes on the basics?

If you read this far, you’re probably trying to actually build the thing, not just read about marketing. The outcome worth aiming for is simple: the fundamentals in place (a clear funnel, the right channels, a real testing loop) and an AI-native way to run them, so one person gets the output of a small team. I’ve spent about ten years running growth this way, and rebuilt the whole thing around AI more recently, breaking plenty of it along the way.

If you want to sort out where your basics are leaking, or just check you’re not about to point AI at the wrong thing, I do a free 15-minute spar. No pitch, no slides, just a straight read on your setup. Come grab a slot and bring the one thing about your marketing that’s bugging you most.