I run WordPress sites. I build plugins. I mess with AI. I like websites.
So naturally, when I get a business idea, one of my first instincts is to build the website.
Domain. Logo. Theme. Homepage. Plugins. Forms. Pricing page. Maybe a community. Maybe turn it into an app while I am at it.
Before long I have built an entire machine for a business that has not made one dollar.
If you are a builder, you probably know exactly what I mean.
The website feels like progress because you can see it. The problem is that a beautiful website can hide the fact that you still have no idea whether strangers want what you are selling.
So here is the rule I am trying to follow more often now.
Test the business idea before you build the full website.
The website is not the business
I say this as somebody who loves WordPress.
A website is infrastructure. It can explain the offer, capture leads, take payments, rank in Google, host a community and automate half the business.
But it cannot manufacture demand.
I have spent years building online. I have had projects work, projects fail and projects change direction halfway through. I have dealt with Apple reviews, app development, WebToNative, WordPress, plugins and AI builders. I have also spent hundreds of dollars on Lovable credits because once you are in build mode, one more fix always seems reasonable.
The danger is that building becomes a substitute for testing.
You tell yourself you will market it when it is ready.
Then “ready” moves.
My favorite form of procrastination is productive procrastination
Regular procrastination is easy to spot. You are watching videos instead of working.
Productive procrastination is sneakier.
You work for six hours and feel great about it. You changed the colors. Installed analytics. Rewrote the about page. Compared membership plugins. Made the mobile menu better.
Everything you did was technically productive.
None of it answered whether somebody will buy.
I am extremely capable of doing this because I enjoy the technical side. I can always find another thing that needs fixing.
Validation forces a different question. What can I do today that gives me evidence from outside my own project?
Start with the problem and the customer
Before the site, write down who the customer is and what problem they already know they have.
Not “entrepreneurs.” That is basically everybody with a LinkedIn account.
Something specific enough that you know where to find them.
Then find evidence.
The U.S. Small Business Administration recommends market research around demand, market size and customer characteristics, along with competitive research into strengths, weaknesses, barriers and opportunities. This is boring compared with generating a slick homepage with AI. It is also much closer to the questions that decide whether the homepage will ever matter.
Search Google. Look at Reddit and niche communities. Read reviews of competing products. Look at what people pay for. Look at the ugly workarounds they tolerate.
If nobody appears to care about the problem, do not assume your website will teach them to care.
Build the smallest test that can get a real response
You do not need to avoid websites completely. Just stop building the whole damn company before the test.
A one-page site can be a test.
Explain the problem. Explain the outcome. Put a price on it if you can. Give the visitor one action.
Then get the page in front of the people you think need it.
If you cannot get them to care about the simple promise, adding a member dashboard is probably not the answer.
This is where AI and modern site builders are genuinely useful. You can create the test much faster than you could years ago. I just think we need to use that speed to test more ideas, not to overbuild unproven ones.
I recently wrote about my move back toward WordPress after spending heavily with AI builders in WordPress vs AI Website Builders. The lesson was not that we should stop using AI. I use AI constantly. The lesson was that speed, ownership, cost and control all matter differently depending on what you are building.
For validation, speed is fantastic. Build a test in an afternoon. Just do not confuse the test with a business.
Search demand can tell you whether the conversation exists
One of the easiest things to research before building is how people search around the problem.
Google’s own documentation on Google Trends explains how to compare search interest, explore related queries and see how topics change over time. Google also specifically recommends making sure a topic fits your audience and your own experience instead of creating content just because something is trending.
I like that advice because it applies to business ideas too.
A spike in searches does not prove a business. Zero search volume does not automatically kill one either. Some products create new categories. But search behavior gives you another piece of evidence instead of another opinion.
Try to get a stranger to do something
This is where the test gets uncomfortable.
Your mom liking the idea does not count. Your friend saying they would definitely use it someday does not count. Me asking an AI whether my idea is brilliant definitely does not count.
Get somebody who does not owe you encouragement to do something.
Join the waitlist. Book the call. Reply to the message. Try the manual service. Put down a deposit. Buy.
The closer you get to money, the stronger the signal.
SCORE’s market research guidance focuses on the same basic reality. You need to understand whether there is a real paying market, not merely whether the idea sounds plausible.
Do not build around assumptions you have not noticed
This one has bitten me in different forms.
An app idea can quietly assume Apple will approve the experience you designed. A web app can assume the wrapper behaves exactly how you expect. An AI product can assume API costs stay reasonable. A SaaS idea can assume customers tolerate another monthly subscription. A WordPress product can assume the plugins it depends on keep behaving the same way.
Every business has dependencies. The mistake is pretending they are implementation details.
They are business risks.
Write them down before you build.
Run a Reality Check before you open the builder
This is exactly the kind of behavior I want the free Marketur Reality Check to encourage.
Before you open WordPress, Lovable, Replit, your code editor or whatever you build with, run the idea through the tool.
See where it thinks the assumptions are weak. Look at competition, demand, monetization, distribution and dependencies. Then challenge the result with your own research.
If the idea looks stronger afterward, build a small test.
If strangers respond, build more.
If people pay, now we are getting somewhere.
And when the evidence finally gives you a reason to build the real website, WordPress is still sitting there waiting for you.
I am not trying to stop people from building.
Hell, building is my favorite part.
I am trying to stop us from spending three months building the answer before we bothered to ask the market the question.