Real businesses. Real AI search results.

Turning a neglected Google listing into a name AI assistants recommend

Harbour Collective Coworking · Flexible workspace · 10 week engagement

The challenge

Harbour Collective had built a loyal member base almost entirely through word of mouth. Their Google Business Profile hadn't been touched in over two years, their site didn't answer the practical questions people ask before booking a tour, and when prospective members asked ChatGPT or Perplexity about coworking spaces in the city, the business simply wasn't part of the answer.

What we did

1. Rebuilt and verified the Google Business Profile properly: categories, services, photos, hours, and a structured FAQ section.
2. Wrote direct-answer content for the questions people actually ask before booking, day passes versus membership, pricing, parking, nearest station.
3. Added LocalBusiness schema and corrected inconsistent business details across the directories still listing them.
4. Refreshed and structured recent reviews so trust signals weren't two years stale.

The results

Getting found at the exact moment someone needs help

Whitfield Therapy · Integrative therapy, solo practice · 3 month engagement

The challenge

Clare is a one-person practice with no marketing budget and no time to spare on anything that needed constant upkeep. Her site answered nothing specific. It didn't speak to the exact way someone searches, or asks an AI tool, when they're trying to find a therapist for anxiety nearby, often at a difficult moment.

What we did

1. Rewrote the key pages to directly answer the specific questions people type and ask AI, by concern and by location.
2. Added clear, named-practitioner content: qualifications, approach, areas of specialism.
3. Built FAQ and schema markup around the questions people have before they book a first session.
4. Kept the scope deliberately small. A handful of genuinely useful pages, not a content factory, because it had to run without taking over her week.

The results

Helping residents find one page, fast, inside a fixed deadline

Elmsworth Borough Council · Public consultation · 6 week consultation window

The challenge

Residents needed to find the consultation page itself, not the homepage. The information that mattered, opening dates, how to respond, when it closed, was buried several clicks deep, and the council needed broad reach across the borough with no advertising budget and a fixed window to gather feedback.

What we did

1. Restructured the page hierarchy so the consultation itself was the destination, not a sub-page of the homepage.
2. Wrote direct-answer content for the real resident questions, what's happening, when it opens, how to respond, when it closes.
3. Added FAQ and structured markup so the details could be read and cited accurately by AI assistants, not just search engines.
4. Made sure the same dates and details were consistent everywhere the council published them.

The results

Ranking well on Google wasn't the same as existing in ChatGPT

B2B SaaS company · Name withheld at client's request · 2 month engagement

The challenge

This client came to us convinced their AI visibility problem was Google's problem to fix. It wasn't. They ranked well, page one for the terms that mattered, but when their own sales team tested ChatGPT with the exact questions their prospects ask, the company barely showed up. Out of 50 questions we tracked, they were cited in AI Overviews for 4 of them.

The real obstacle wasn't technical. It was trust signals. Years of blog content had been written by a rotating cast of freelancers with no named authorship, and their company information was inconsistent across more than 30 directory listings, G2, Capterra, Crunchbase, old PR mentions. To a model trying to work out who to trust, that looked like noise, not authority.

What we did

We ran them through the same six checks we'd put any business through before spending on AI visibility work: what AI tools currently say about you, whether your site answers the exact questions buyers ask, whether your expertise is traceable, whether your information is consistent everywhere, whether you're the source being cited or just another ignored page, and whether your name would come up if a competitor asked. They failed five of the six.

1. Reinstated named authorship across the entire content library.
2. Cleaned up company data across every directory listing we could find.
3. Built structured answers for the questions prospects were already asking AI tools.
4. Implemented schema properly, rather than bolted on as an afterthought.

The results

None of it came from a technical trick. It came from making the business genuinely easier for a model to trust.