The help centre that answers before anyone calls.

Support that should have been self-serve

The client is a global accommodation booking platform. This project focused on its partner help centre: the support content for the property owners and hosts who list with it.

The goal was unusual for a search project. The platform wanted fewer partners contacting its support team by phone and web chat. Partners now look for answers in Google, AI Overviews, AI Mode, ChatGPT and Copilot before they think about calling. If the help centre's answers showed up there, a lot of those calls and chats wouldn't need to happen.

There was a second goal. Prospective partners research hosting and listing questions without any platform in mind, and other booking platforms were answering them. Winning more of those non-brand answers builds awareness with hosts before they decide where to list.

A year, starting with an audit

The project ran for a year. We started by auditing the existing support content, the information architecture and the internal linking.

The findings shaped everything after:

• Consolidation lists for articles that overlapped or competed with each other
• An internal linking plan that grouped content into topic clusters, so related answers support each other
• A review of existing articles, reworked to answer the question directly and to fit the way people search now
• New topics, plus new pillar pages that we briefed and that were written to those briefs

Most questions name the platform

A typical partner question is something like "how to manage guest credit cards". Most support searches include the platform's name, because they're about how its system works.

That changes the job. For a branded question, ranking isn't the hard part. What matters is where the answer comes from: the platform's own help centre, or a forum thread, a third-party blog or someone's best guess. We wanted AI Overviews, AI Mode, ChatGPT and Copilot to cite and reference the official solution.

The work combined our generative engine optimisation and SEO content practice.

How we measured it

We tracked AI referral traffic and engagement in GA4. AI referral traffic grew 70%, and visitors arriving from AI engines had an engagement rate roughly 30% higher than organic visitors. They arrived with context, and the article they landed on answered the question and linked to the next one they were likely to need.

We also checked the engines directly, running partner questions through them and confirming that the help centre's solutions were being cited, so partners could get the answer without navigating the site at all.

Internally, the support side reported that direct enquiries had come down. Nobody put a number on it, but it was the outcome the whole project was built for.