ChatGPT Business Hub discovery topics: how to use them for content planning

OpenAI’s help article defines discovery topics as common themes in answers that mentioned or cited your business. It adds two limits: themes do not reveal individual prompts, and they do not directly measure customer intent. Businesses never receive individual prompts or conversations. Launch reporting from LLM Pulse and Sanbi, both vendors, describes a table of the top five discovery topics in the Brand insights tab.

So a discovery topic is closer to a category label than to a keyword. “Marketing attribution for B2B SaaS” as a theme could cover dozens of different questions with different intent. That makes discovery topics useful for direction and useless as a keyword list. If you have not set up the profile, start with our guide to claiming your business in ChatGPT.

This is the workflow our content team uses. It treats discovery topics as a starting signal and your own testing as the evidence.

1. Log the themes: copy the top discovery topics into a sheet each week, with the date range, because the window is short and the list changes.
2. Expand each theme into questions: write three or four questions a buyer in your ICP would actually ask ChatGPT about that theme, in their words.
3. Test and map: run those questions in ChatGPT, note whether you appear, who appears next to you and which page of yours is cited.
4. Classify: for each question, decide whether you need to strengthen an existing page, create a new page or earn third-party coverage.
5. Measure: after changes, rerun the same questions monthly and watch mentions and website clicks in Business Hub.

The test in step three gives each question one of four outcomes. Each one leads to a different piece of work.

Keep the mapping honest. A single run in ChatGPT is one sample; answers vary between sessions. Our guide to measuring share of answer explains how to size a question set so the trend means something.

Here is how the steps look in practice, using a made-up example rather than client data. Say a B2B data platform sees a discovery topic like “data integration for manufacturers”. That theme could hide very different questions. We would write four: which data integration tools work with SAP for a mid-sized manufacturer, how long an implementation takes, what it costs, and how the platform compares with a named competitor.

Running those four in ChatGPT might show the company named for the first, absent for the cost question and outranked in the comparison. The plan then writes itself: keep the integration page current, publish a page that explains cost drivers and price ranges honestly, and build a fair comparison page plus third-party coverage for the head-to-head. A month later, the same four questions and the Business Hub numbers tell you whether it worked.

Notice what the theme did and did not do. It pointed at an area where the company already appears. The questions, the testing and the page decisions came from the team.

Discovery topics only show where you already appear. The bigger opportunity is often the themes that should be there and are not. If you sell to CFOs and no theme mentions cost, ROI or pricing, that is a signal. Publishing clear pricing context is one of the strongest moves we see, and our guide to publishing pricing for AI search covers how to do it without giving away your price list.

Build the missing list from sales calls, lost-deal notes and the questions your sales team answers every week. Then check those questions in ChatGPT the same way. Our explainer on how B2B buyers research vendors in ChatGPT shows the question patterns we see most.

Weekly logging, monthly decisions. Copy the themes every week so you have a history, because the window is short and the dashboard does not keep one for you. Make content decisions once a month, when you can see whether a theme held for several weeks or appeared once.

Review the full question set behind each theme each quarter. Products, competitors and buyer language change, and so do ChatGPT’s answers. A quarterly review keeps the plan tied to what buyers ask now, not to what a dashboard showed at launch. Share the summary with sales: the themes often explain questions they are already hearing on calls.

A few mistakes are already common in launch week.

• Treating themes as prompts and writing one article per theme title.
• Reading the top five as demand volume; OpenAI says themes are not a direct intent measure.
• Reacting to a single week; themes depend on thresholds and change with activity.
• Stuffing theme words into the profile description; write it for buyers, not for the dashboard.
• Ignoring the pages ChatGPT already cites; improving them is usually faster than writing new ones.

Discovery topics finally give content teams a first-party signal from ChatGPT itself. They do not replace your own question set, analytics or CRM data; they make them easier to prioritise. Our generative engine optimization service combines all three and turns them into a monthly content plan.

If you want help turning your first discovery topics into a plan, talk to our team.