Theresa Wilkinson

Articles & Publications

Would You Trust an AI-Built Homepage? What 54 Survey Participants Told Me.

An expert review and a survey of 54 participants reveal the work still needed after AI generates a website.

When I reviewed the remodeling website generated by Lovable, the photographs were appealing and the navigation was clear. The text immediately concerned me.

The service descriptions were vague. They provided little specific information about the work offered or why a prospective customer should choose the company. Important credibility information, including licensing, insurance, and credentials, blended into the surrounding text.

My earlier experience in website conversion shaped that reaction. I look at whether a website gives people the information they need to make a decision. Does the company offer what they need? What supports its claims? Do visitors have enough confidence to contact it?

My approach was influenced by Bryan and Jeffrey Eisenberg's work on website conversion and Persuasion Architecture: considering how customers make decisions and what information they need to move forward. That perspective shaped my review of the AI-generated page.

From that perspective, the AI-generated website needed substantial human editing before I would consider it ready for a real business.

Looking beyond the appearance

I compared the main page of the Lovable prototype with the main page of J.S. Brown & Co., an established remodeling company with a non-AI-generated website.

The difference in written content stood out. J.S. Brown & Co. provided specific information about their services and experience. Recognition and accreditation logos made its credentials noticeable. Its project photographs appeared connected to the company's work.

Lovable produced an attractive layout, but its brief descriptions left important questions unanswered. There was room on the page to explain the services, demonstrate the company's capabilities, and give visitors reasons to choose it.

Credentials were present, but their presentation did little to draw attention to them. Contact prompts also needed greater visibility throughout the long page, with clearly labeled alternatives to calling.

These were findings from my expert review. I wanted to examine prospective customers' impressions separately.

What 54 participants thought

I conducted a survey in which 54 participants viewed a screenshot of J.S. Brown & Co.'s main page. Participants were recruited through LinkedIn and Facebook and were not paid.

They rated their trust in the company, confidence in its qualifications, perceived professionalism, and confidence that it would deliver quality work. They also rated their likelihood of contacting the company and requesting a quote.

Their impressions were consistently positive: