Vibecoding and the Illusion of a Single Lead

For me, social media is what I scroll when my brain is too exhausted to process complex information and just needs to idle. I don't want to read a book, I don't want to write work notes, and my brain flatly refuses to build another marketing strategy.

I just want to relax, scroll through short videos, or watch strangers on the internet explain to other strangers how to live their lives properly.

At some point, Threads started filling up with posts from "vibecoders" sharing their rapid success stories.

The storyline is almost identical every time. An idea pops up in the evening. At night, they open Claude or Cursor, and a few hours later, they’ve hacked together a new service. By morning, the site is live. And there’s always the mandatory line: "A year ago, this build would have taken three months and thousands of dollars."

I love tech that lets you build in an evening what used to require weeks of approvals, a designer, a developer, a QA engineer, and someone who insists on scheduling a meeting about it all. I have zero issues with vibecoding itself.

But curiosity eventually wins over laziness, and I start reading the comments. Inevitably, someone asks: — Can we see what you built?

The response is either total silence or a link to a clean, slick landing page. A big bold header, smooth gradient, feature cards, three pricing tiers, and a "Start Now" button. A couple of years ago, you actually needed a frontend dev for that. Today, all it takes is an evening and a detailed prompt.

The site looks great. The problem is, it looks like it could simultaneously sell accounting SaaS, a meditation app, or a service for finding lost socks.

So I ask a question about conversion rates. Once, I got this exact reply: — Just got a lead yesterday.

Now things get interesting.

A Single Lead

Years of working in advertising and analytics ruin a person. When someone tells you "we got a lead," you don't feel the expected burst of excitement. You immediately want to know: how many visitors actually came through? Where did they come from? How much did that traffic cost? What kind of lead was it, did they buy anything afterward, or did the founder just fill out the form themselves while testing the submit button?

A single lead tells you almost nothing on its own. If three people visited the site and one left a lead, that’s one story. If ten thousand visited, that’s another. If you paid $50 for that visitor in ad spend, that’s a completely different story.

In social media posts, all of this easily compresses into a neat formula: come up with a product → build a service → get a customer → drop a bold claim: "I have a business now, I’m a founder."

Startup analytics tell a colder story. Over 80% of these AI micro-projects on Product Hunt have a median revenue of exactly zero dollars. They get stuck at the "one test user" stage, who often turns out to be the author.

Building Got Trivially Easy

Development has fundamentally changed. Someone with zero programming background can now build a functional prototype. Not a Figma mockup, but an actual working tool where users can click buttons, submit data, and process payments.

In the past, high development costs acted as a natural filter. To test an idea, you had to find a developer, pay real money, and wait weeks for the result. According to CB Insights, the number one reason startups die is a lack of market need. 42% of projects fail simply because nobody wants what they built. That filter used to kick in before development started. And honestly, that was useful.

Now the technical barrier is gone. You can invent a service in the morning and send the link to your network by evening.

Along with the barrier, however, we lost the forced delay that used to make founders ask a critical, uncomfortable question — does anyone actually need this? AI can write the code today. Answering that question still requires human effort.

A Website Is Not a Product

Marketing suffers from an old, similar disease. An ad campaign launches, clicks start coming in after three days, and everyone celebrates. Then you look at the analytics — there are clicks, there are add-to-carts, but zero sales.

The same thing is happening with AI projects. The mere existence of a working interface is mistaken for a business result.

"I built an app." No, you deployed an interface.

In product analytics, keeping at least 15% of your users returning after a month (Retention Day 30) is considered a good benchmark. For most of these fast AI services, that metric is sitting at zero. A user visits once out of curiosity and never comes back.

A product emerges later, when you prove that people genuinely use the tool repeatedly. A business emerges even later, when you prove you can acquire those users regularly without spending more than their average order value.

Nobody wants to post about that part. It’s far more satisfying to share a screenshot saying "Launched a startup over the weekend" than to spend six months figuring out why user retention is dead in the water.

The Cheapest Part of the Enterprise

When engineering was expensive, code was viewed as the main asset. Today, that relationship is completely inverted. Assembling a working tool is the cheapest part of the entire enterprise.

Then comes everything that doesn't fit into a nice social post:

• Finding an audience and understanding their actual problem.
• Driving traffic and observing what users actually do.
• Redesigning half the features you thought were obvious.
• Calculating unit economics and deciding if the project is worth continuing.

Claude is great at helping with code. But it won't create demand. Building a service in two nights feels great psychologically. Discovering over the next two months that nobody cares about it feels much worse.

Competition has shifted from the ability to build to the ability to understand what is actually worth building. AI hasn't automated that tedious work yet — analyzing data, questioning your own assumptions, and refusing to celebrate a single accidental lead as a victory.

I like vibecoding. Given the choice, I’d rather spend an evening hacking together a working prototype than writing a twenty-page specification document. But the ability to write code has officially uncoupled from the ability to build a business.

Users do not care how long it took to build your product or how many prompts you sent. They either need your tool, or they don't.

We gained the ability to turn ideas into working services at incredible speed. But building fast doesn't mean it was worth building in the first place.

And a single lead on a Tuesday evening doesn't change that reality.