At Profound, we help brands optimize their visibility in answer engines such as ChatGPT. One of the first products we built was Answer Engine Insights, which allows customers to track the prompts most relevant to their company (think “best running shoes” if you’re Nike) and create a visibility score showing how often their brand appears in AI responses.
However, the number one question we get is: Which prompts should a customer track?
Customers can track any prompt that they like, but if it’s not grounded on what users are actually searching, it could be fundamentally useless - content is created, optimizing for the wrong prompt and has limited business impact.
We solved this by leveraging our dataset of 100M+ answer engine prompts and responses from real user’s conversation data, which powers our Conversation Explorer product. By analyzing all prompts connected to a brand, we can recommend the most valuable ones based on the most common themes and questions users are asking. For example, if you’re focused on the business phone category, we look across all user conversations to identify the prompts driving engagement, and help you capture that attention.
How does it work?
First, we start with the topics and prompts that customers are already tracking. Using this list of relevant topics, we search across our dataset of real conversations to find prompts that align with these themes.
From here, we perform semantic analysis to identify the relevant prompts matching these topics and apply deep NLP techniques to then identify from here, the most relevant subtopics within this industry and quantify their relative share importance within the topic.
E.g. Sports betting might have subtopics about MLB bets, parlay strategies, etc.
Finally, we compare these subtopics to each brand’s existing list of tracked prompts. This allows us to spot the gaps, highlighting prompts that are clearly relevant but are not yet being tracked.
What’s next?
We started by recommending prompts specifically for ChatGPT, but we will soon expand to recommend prompts across multiple answer engines. We believe the intent and behavior of users can differ significantly between platforms.
So far, we have used topics provided by our customers as the starting point to identify the most important prompt gaps. Next, we plan to recommend entirely new topics that are relevant within your industry but may be missing from your tracking altogether.
Getting Started
Stop guessing which prompts to track. Transform your AI visibility strategy with recommendations powered by our unique dataset of 100M+ answer engine prompts and responses from real user conversations. Our prompt recommendation engine is available now on all plans through the Profound dashboard.
Get started today by scheduling a demo to see how we can help you capture the attention you're missing.
