Bridging the gap between Google and AI search for BreakTime Coffee & Vending
Testing whether targeted improvements to existing website content can strengthen visibility in AI recommendations.
Client: BreakTime Coffee & Vending
Industry: Commercial coffee and vending machine supplier
Location: Ripon, North Yorkshire
Project: Improving visibility in AI search recommendations
Website: https://www.breaktime.co.uk/
The Challenge
BreakTime Coffee & Vending was already performing strongly in traditional Google search.
We’d spent considerable time building and improving the website’s content, including location-specific landing pages covering the areas BreakTime serves across Yorkshire, the North East and Lincolnshire.
That work had produced strong Google visibility. But as more people start using AI tools such as ChatGPT to ask for business recommendations, I wanted to find out whether that strength was also being reflected in AI search.
I began by testing a series of questions that a potential customer might realistically ask ChatGPT.
The initial results were mixed rather than poor. BreakTime appeared when I asked for commercial vending suppliers across Yorkshire, for example, but it wasn’t included in some of the more specific recommendation searches.
One of those was:
“Can you recommend a vending machine company covering Leeds?”
Despite BreakTime serving businesses in Leeds, it wasn’t included in ChatGPT’s original shortlist (see image below).
Rather than seeing this as a problem with the whole website, I wanted to understand why BreakTime was appearing in some AI searches but not this particular one.
The Solution
I looked more closely at the businesses ChatGPT was recommending and the information it appeared to be using to support those recommendations.
Rather than creating lots of new content specifically for AI search, I concentrated on strengthening the useful content BreakTime already had.
The work included:
- reviewing the existing Leeds landing page and making BreakTime’s local service coverage clearer;
- giving greater prominence to BreakTime’s 50+ years of experience and four-hour fault response time;
- adding stronger local evidence from genuine customers;
- adding context around a Google review from a Leeds charity, where a late-afternoon request resulted in a BreakTime engineer arriving the following morning;
- strengthening relevant service content, including the commercial coffee machines page; and
- reviewing other recommendation searches to identify where BreakTime’s existing website could provide clearer evidence of its experience, services and geographical coverage.
The aim wasn’t to stuff pages with references to AI or produce content purely for ChatGPT.
It was to make the existing evidence about BreakTime clearer, more specific and easier to understand.
Once the changes had been made and given some time to be discovered, I repeated the original Leeds test in a fresh Temporary Chat.
Importantly, I asked exactly the same question:
“Can you recommend a vending machine company covering Leeds?”
The Impact on BreakTime Coffee & Vending
This time, the answer was noticeably different.
ChatGPT opened its recommendation with:
“For a Leeds workplace, I’d put BreakTime Coffee & Vending high on the shortlist.”
It then highlighted BreakTime’s experience, Leeds and West Yorkshire coverage, range of vending and coffee solutions and four-hour response time.
More interestingly, it also picked up the Leeds charity example we’d deliberately added to the website, referring to the occasion when an engineer attended the following morning after a late-afternoon request.
ChatGPT concluded that if it were choosing between the suppliers, it would speak to BreakTime first, particularly where reliable after-sales service mattered.
That’s a significant change from the original test, where BreakTime wasn’t included at all.
What does this actually tell us?
It’s important not to read too much into a single test.
AI-generated recommendations can vary, and many different factors can influence the businesses ChatGPT chooses to include. I can’t say that the website changes alone caused BreakTime to move from being absent from this particular result to being recommended first.
What I can see, however, is that information we’d deliberately strengthened on BreakTime’s website was subsequently being found and used by ChatGPT to explain its recommendation.
For me, that’s the particularly interesting part of the experiment.
BreakTime didn’t need a completely new website or dozens of new articles written for AI.
The website was already performing strongly in Google and appearing in some AI recommendations. By testing specific customer questions, we were able to identify an area where the evidence wasn’t coming through as strongly, improve the relevant existing content and then test again.
The result gives us good reason to continue that approach: test, identify gaps, strengthen useful existing content and measure what happens.
Google search still matters
None of this means businesses should suddenly forget about Google and concentrate solely on ChatGPT.
The work that had already helped BreakTime perform well in traditional search gave us a strong foundation to build on.
And many of the things we’re now looking at for AI search are simply good website practice: clear information about what a business does, where it operates, genuine customer experiences, specific evidence and useful content that answers real customer questions.
The opportunity is to make sure that strong website content can work as hard as possible wherever potential customers are searching.
Could your existing website be working harder?
You might already have a good website and reasonable Google visibility but still wonder how well your business is represented when people search using AI.
My free personalised website review looks at what’s already working, what may be holding your website back and the improvements I’d prioritise.
Not sure what your website needs?
Whether you're planning a new website or trying to improve the one you've already got, tell me what's not working or what you'd like to achieve. I'll take a look and give you a straightforward idea of what I'd recommend.