Hello Reader
Your website homepage can sound polished and still leave a buyer wondering what you sell.
I know because I built one.
I recently asked messaging and positioning consultant (and author of an excellent newsletter called Gobbledy), Jared Blank to review the Driver Labs homepage. I showed him the website before I explained anything about the business.
Here’s the first headline he saw:
“Now you can build a business your clients can't replace.”
His response:
“I've learned nothing.”
Ouch.
For context, Driver Labs is an online community that helps marketing leaders to use AI through live meetups, guides and peer learning—but you wouldn't have known that from the headline I wrote.
And I couldn't even blame AI for my messaging faux pas.
ChatGPT and I spent a lot of time collaborating on that copy. I gave it context, read everything it suggested and pushed back when I disagreed. We went through multiple iterations and I thought the copy read well.
What I should have done first was the positioning work that leads to clearer copy.
I let ChatGPT coax me into a false sense of accomplishment and the belief that my copy was doing a good job.
Luckily for me, Jared wasn’t shy about pointing out all the areas I’d gone wrong.
The danger of unclear positioning
As a marketer I’ve always felt the pressure to do more with less.
And even though our CEOs and boards expect us to do more, resources have barely budged. Gartner's 2026 CMO Survey puts marketing budgets at 7.8% of company revenue versus 7.7% last year.
At the same time, AI has made producing the work dramatically faster.
Pew Research Center found signs of AI authorship in more than a third of the webpages it sampled in July of this year.
So we're using AI to help write landing pages, campaign copy and websites.
Who can blame us?
Claude is very good at rolling out polished sounding sentences that let us tick another task off our list.
Unfortunately this polished sounding copy isn’t always effective, especially if our positioning needs some work.
Which was exactly my problem.
I had a whole page of phrases that included “complete workflows,” “real AI results” and “the fixes you’ve tried before”.
I know what all of these mean but someone with little to no context on what I offer would be lost in a sea of vagueness.
The worst of marketing crimes!
“Every company has a shorthand,” Jared says. “And when you're speaking externally, you have to break out of that, otherwise your reader will have no idea what you’re talking about.”
I'd given ChatGPT so much context that we'd developed our own shorthand.
Once an AI has your strategy, customer research, brand documents and previous conversations, it starts to understand your business and mimic the way you talk (and write).
That context is incredibly useful when you're creating, but it can also make AI a lousy proxy for a first-time visitor.
Don’t bury the lead
Jared's main issue with the Driver Labs copy was how much work I was making the reader do to understand my offer.
He says a homepage should orient the visitor immediately.
“They need to know they're in the right place,” he says. “Is this for someone like me? Will it work for my situation? How do I learn more?”
My headline didn't tell him who I was talking to, what I was selling or what benefit Driver Labs provided.
His suggestion was remarkably simple: tell people what you’re offering.
A peer community for marketing leaders.
Then explain the benefit as specifically as you can.
Phrases like “Real AI results” and “AI systems that work” are much too vague.
“What are real AI results?” he asked.
Good point.
If I'm promising members useful AI workflows, I need to show them one. If I'm promising practical implementation help, I need to explain exactly what that looks like.
Jared also prefers positive and direct language because negatives force the reader to translate the sentence into a benefit.
Every time someone has to translate your jargon, resolve a negative or infer what you mean, you’re forcing them to do the work for you.
“Why do I have to do the work?” he asked. “You do the work.”
Give AI a specific job
While Jared’s not a fan of AI writing, he thinks it’s a great collaborator on other tasks.
“I ask it to be a skeptical reader,” he says.
He gave me an example from a recent client project— a company that makes software to run and monitor large infrastructure systems.
Jared loaded Claude with client materials, recorded conversations and other background he’d collected during the project. Then he asked Claude whether anything in the messaging should concern him.
It flagged the word compromise that he’d used as part of a tagline.
Cybersecurity is critical in that industry and a compromised system carries a negative meaning.
Jared hadn't made the connection and appreciated Claude pointing it out.
“I don’t use AI to write but it’s a great partner for showing me things I might have missed,” he says.
You can ask AI to interrogate your copy choices, find assumptions and look for ambiguity. It’s also a great way to surface the areas where you're relying on knowledge the reader doesn't have.
Three checks before you rewrite
If you suspect any of your copy needs a rewrite, here are some useful things to keep in mind.
1. Get specific about who you help
A job title alone tells you very little about someone's situation, so before rewriting anything, make sure you can complete this sentence:
We help [person in a particular situation] do [specific job] through [actual offer].
If you’re struggling to fill in those blanks, spend more time on the customer and offer before working on the copy.
2. Ask someone unfamiliar with the business to explain it back to you
Give them your homepage without context and ask what you sell, who it's for and why someone would choose it.
If you need to explain any of your phrases or terms, you know these sections need a rework.
Record the conversation so you can listen for the moments when the person understands the offer. They’ll say something like, “oh, now I get it!”
Go back to the words and phrases you used that elicited that response. Those are the ones you want to use in your copy.
3. Put the proof next to the promise
If you promise useful workflows, show one.
If you promise implementation support, explain what happens.
If you promise results, name them (preferably with a quantifiable metric).
Make it easy for the reader to connect what you're promising with what they actually get.
Use AI to assess your messaging
Once you've rewritten your copy, use AI to verify that you’ve hit the mark. Here are two tests you can run:
Test 1: The Stranger
Open a new AI conversation and give it zero context about your company.
Don't upload your brand guide or explain your audience. If you already have this loaded in your custom instructions or skills, tell it to forget what it knows.
Then paste in your homepage or copy and ask:
Don't reference your custom instructions or skills. Pretend you know nothing about my business.
Based on this page:
- What does this company sell?
- Who is it for?
- What problem does it solve?
- What would I actually receive or be able to do?
- Why would I choose this over an alternative?
Don't fill in missing information using assumptions. Flag anything you have that’s unclear, vague or generic.
I recommend you use a high reasoning model like Astra or Opus for this analysis.
Here’s what my output looks like in ChatGPT. I used Astra High:
This gives you something close to the experience of a first-time visitor.
Test 2: The Skeptical Reader
Now load your context. Give AI your customer research, positioning, product information, sales calls, industry terminology and brand documents.
Then give it the same copy and ask:
Act as a skeptical reader and answer these questions:
- What assumptions are we making about what the reader already knows?
- Which claims are vague or unsupported?
- Which words could have another meaning in this industry?
- Where are we using internal shorthand?
- Where does the reader have to infer the benefit?
- Which claims need an example, demonstration or evidence?
- What could a first-time visitor reasonably misunderstand?
Here, context works in your favor.
AI knows enough about the business to spot inconsistencies, implications and missing evidence.
Here’s a term and phrase analysis I got from ChatGPT: