Topics

Tracking your AI visibility in Traqer is based on topics, instead of individual prompts. This page will explain why and how to create or edit new topics. ## Why Traqer Uses Topics Traqer is built around \\\\\\\_topics\\\\\\\_ because tracking visibility on single prompts is extremely misleading and inaccurate. You can read the full explanation of why here: [Topic-based GEO](https://www.growandconvert.com/ai/topic-based-geo/), Invisible Prompts. But here is a short summary:  - No one knows what users are typing into LLMs like ChatGPT. There’s no “search console”. Other tools that claim to have “real chatGPT search” data, get this to people who volunteer to have their web browsing tracked, so it’s an extremely small slice of the population.  - Users don’t enter short, repeatable keywords into LLMs. They use longer prompts or in tools like ChatGPT have long, non-reproducible conversations.  - LLMs include a ton of chat history, context, and preferences of each user when generating responses. So two people who both ask ChatGPT for “the best running shoes” can get dramatically different responses based on what ChatGPT knows about them (their gender, budget, preferences, habits, favorite brands, location, etc.)  - LLM responses have natural variability and randomness, so showing up once in a response doesn’t mean you will show up every time.  As a result, **tracking your GEO visibility based on individual prompts like traditional SEO keywords doesn’t make sense for AI search.**  As a result, we think it’s far more accurate to see \\\\\\\_how often\\\\\\\_ your brand is mentioned when a certain \\\\\\\_topic\\\\\\\_ is broached. While this is still not 100% accurate (nothing in AI search can ever be), we think it’s a far more correct and accurate way to view your visibility than thinking you simply are or are not visible for a given prompt.  ## How Topics Work A topic in Traqer is a collection of prompts that all ask about the same thing. Traqer then calculates what percentage of those prompts your band was mentioned or cited.  For example here are 5 prompts all related to the idea of a user asking an LLM about “commercial service software”.  So they are grouped under the topic “commercial service software” ### Topic Level Visibility Percentage Then for each LLM, we can add up how many of the prompts in that topic had your brand mentioned or cited and give you a \\\\\\\_topic-level\\\\\\\_ \\\\\\\_visibility percentage\\\\\\\_.   This is a simple percentage of the prompts in that topic where your brand is either mentioned or cited. We calculate it separately for each LLM.  For example, three out of the five prompts in the topic above have a brand mention or citation in ChatGPT and therefore the ChatGPT visibility percentage for this topic is 60%.  ### Why Topic Level Visibility Is More Accurate  This is how Traqer solves the issues with single prompt tracking mentioned above. By covering a topic from multiple angles, with multiple prompts, Traqer shifts a brand’s thinking from “Are we visible for \\\\\\\_this prompt\\\\\\\_?” to “How likely are we to be mentioned in this topic area?”  For example, the second prompt above is a reasonable way to ask about commercial service software: "What's the best software for managing commercial service contracts?" If this brand had only tracked that prompt, it would mistakenly think it has \\\\\\\_no\\\\\\\_ visibility in ChatGPT for commercial service software. However, with topic-level visibility, we see that actually 60% of the times we asked about commercial service software to ChatGPT, the brand was either mentioned or cited.  This is a far more accurate way to view your AI search visibility.  \\\\\\\_Note: You can measure topic level visibility for just brand mentions or just citations by flipping the brand versus citation toggle at the top of the page.\\\\\\\_ Adding or Removing Topics  Use the edit button on the top of the brand overview page to add or remove topics: