I remember the early 2020s vividly, the narrative was shifting. Marketing wasn’t just lead gen play anymore. It was accountable for closed-won, retention, expansion, the whole customer journey.

It also spun a lot of new and funny titles. Pipeline Marketing Manager. VP of Customer Marketing. Full Funnel Lead.

Fair enough. That push wasn’t entirely wrong.

But I think it sent a lot of marketing leaders in the wrong direction.

The instinct it created was to design every program, every dollar, every campaign around moving the full funnel. And if you couldn’t show how it connected to the aggregate metric, it felt hard to justify. Especially when the board and the investors push for LTV over anything else.

Here’s where I think that breaks down. Most of your dollars are going into one specific stage. They should be. A webinar for mid-pipeline prospects is an acquisition play. Run it like one. The moment you start asking how it contributes to retention or long-term customer economics, you’ve muddied what you’re actually trying to do.

I know what you are thinking. Yes, early decisions have downstream effects. Good acquisition feeds better activation. Sharp onboarding feeds retention. Those patterns are real. But noticing them isn’t the same as designing programs to move everything at once.

The aggregate metric is the output of individual stages done well, over time, in parallel. I wouldn’t go at it directly. I’d find the stage that’s most broken or most opportunistic, put focused dollars there, and run it long enough to actually see results.

In my experience, that’s always where the returns have come from. Not from trying to attribute everything to everything.

I’ve sat across enough founders to know what borrowed confidence looks like. The words land well. The framing is tight. But ask one follow-up question and you can feel the floor give way. No scar tissue underneath. No story. Just a clean answer that nobody actually owns.

This is happening more. And I think AI is why.

We’ve seen this before

This is what happened to corporate executives in the 90s when management consulting became a default. McKinsey would come in, synthesize everything, and produce a beautiful recommendation. The executive would present it. The team would nod. Everyone would mistake the clarity of the output for the quality of the thinking.

It worked until something broke. Because when things break, frameworks don’t hold. Conviction does. And conviction is the one thing you cannot outsource.

We’re doing it again.

AI is a better consulting firm than McKinsey ever was. Faster, cheaper, always available, and it argues in your voice if you prompt it right. Ask it what you should build, who you should sell to, where your market is going. It will answer. The answer will be good. Good enough to feel like something you figured out yourself.

That’s the trap.

The gap you can’t see until you need it

A founder’s understanding of their customer isn’t a research output. It forms through repetition and discomfort. Through a hundred conversations where someone said no and you paid attention to exactly how they said it. Through deals that closed for the wrong reasons and ones that didn’t for reasons that turned out to matter. You don’t generate that kind of understanding. You earn it slowly, and it lives somewhere below the surface.

AI compresses that into a paragraph. And founders take the paragraph and run.

What makes this harder to catch than the McKinsey problem is that executives knew they were outsourcing. They said it openly. With AI there’s no distance. The output arrives in first person, shaped to your context, sounding like a thought you had. Founders think they’re thinking.

That gap is invisible until you actually need what’s underneath.

You need it when a prospect says “why would this ever work for us” and the deck isn’t enough. You need it when your first big product bet is failing and you have to decide whether to push or fold. You need it when you’re recruiting someone who has options and they’re trying to figure out if there’s something real here. People can feel conviction. They can also feel its absence.

The input has to be yours first

I’m not saying don’t use AI. That’s not the point.

The point is that there are a few questions, just a few, where your input has to come before anything else. What do I actually believe about who needs this? What am I building toward, and why that and not something else? What would I hold onto if everything else got challenged?

These aren’t research questions. They’re belief questions. AI can sharpen a belief. It cannot form one.

The executives who got real value from McKinsey were the ones who already knew what they thought and used the consultants to pressure-test it. The ones who got the least walked in with an open mind and walked out with someone else’s strategy.

The founders who will get the most from AI are the ones who bring something real to the conversation. Already formed, possibly messy, maybe half-wrong. But theirs.

The ones who walk in empty and walk out confident are building on borrowed ground.