How to Audit and Expand Your Prompt-Tracking Setup With AI (Practical GEO Advice)

An artistic representation of expanding a prompt-tracking setup with AI.

This won’t be a comprehensive guide to prompt selection. There are already some great articles out there for that. (See Aleyda’s guide, for example.)

Rather, this article will be a high-level overview of how to audit and expand the prompts in your prompt-tracking tool (be it Profound, Peec AI, or another tool) to get more representative (and actionable) AI visibility data.

The reason for writing this is practical: I was recently advising a consulting client on their Profound setup, and we encountered optimization opportunities around the prompts they were tracking. This advice is born out of that experience.

And here’s a video guide to accompany the article:

The first step in coming up with prompts for your prompt-tracking setup is to cover the buyer’s journey for your target personas for your main lines of business.

Depending on your current plan within your prompt-tracking tool, you’ll have a certain number of prompts allotted to you.

The number of prompts you’ll want to track depends on how many personas you have, how many main lines of business you have, and how many steps users have in their buyer’s journey.

For an exact number of prompts to track, you’ll want to consult with an expert directly, as you don’t want to overspend on prompt tracking. Having said that, I’ve seen a number of setups where brands weren’t tracking all of the prompts allotted to them. So at a minimum, you want to take full advantage of the allotted prompts in your plan by tracking meaningful prompts.

I’ll use my prompt-tracking setup in Peec AI to run through the example process outlined in the rest of this article. My setup will be good for this purpose because I don’t currently follow my own advice — the prompts I track are more for content inspiration than AI visibility reporting, which means we’ll have a lot of gaps to find.

Assuming you have prompts already set up in your prompt-tracking tool, the next step is to export your prompts and import them into an AI assistant (like Claude or ChatGPT).

*You can follow this same step even if you don’t have any prompts uploaded but simply have a list of candidate prompts you plan to upload.

First, visit the prompts page in your prompt-tracking tool and find the export button. Here it is in Peec AI:

Prompt export button in Peec AI.

That’ll give you a CSV document that you can upload to Claude or ChatGPT.

Use the following prompt, with your information filled in:

“I’m uploading a list of prompts I currently track in (tool). Based on the buyer’s journey of my target personas for my main lines of business, identify gaps in my current prompt-tracking setup.

My main personas are (personas).

Their buyer’s journey includes (steps).

My main lines of business include (products or services).”

Here is the example that I put into ChatGPT:

ChatGPT prompt for finding gaps in prompt-tracking data.

This will return an analysis of current gaps in your prompt-tracking setup.

Here are gaps in my setup, for example:

ChatGPT response for gaps in my prompt-tracking setup.

At this point, you have two options for how to fill in those gaps …

The preferred option is to come up with your own prompts.

Ways to generate prompt ideas include talking to your customers (like interviewing them directly or even reviewing sales calls), mining Reddit threads (to see how your customers describe their goals or issues in natural language), or looking at keyword data (like queries in GSC).

This method is preferable because the prompts are based on real user language and behavior, especially when they come from customer interviews or sales calls.

The second option is to have an AI assistant generate prompt ideas for you based on the current gaps.

For the sake of instruction, that’s what I’ll do here.

You can prompt Claude or ChatGPT with this follow-up:

“Based on the gaps you’ve identified, create prompts so that I am more accurately targeting the buyer’s journeys for my target personas for my main lines of business.”

For bonus points, you can also have Claude or ChatGPT assign topics and tags to the prompts, to save you time doing that later.

Here are some of the suggested prompts ChatGPT gave me, based on gaps in my current tracking setup:

Suggested prompts in ChatGPT.

The final step is to audit the list of new prompts, ensure you have a number that fits within your current allotment in your prompt-tracking tool, and then upload them.

After enough data accumulates, you’ll have more accurate AI visibility data from which to take action from.

If you’re curious how to take action from prompt-tracking data, I’ve written several guides on that on my blog, including an overview of how to use Peec AI.

So what have we learned, in summary?

Basically, you want your prompt-tracking setup to be representative of the buyer’s journey for your target personas for your main lines of business.

You can identify gaps in your current setup by exporting your prompts, uploading them to an AI assistant like ChatGPT or Claude, and asking where the gaps are.

Based on that data, you can ideally come up with new prompts to track based on real customer data, or you can get suggestions from ChatGPT or Claude for new prompts.

Following this process will result in more accurate AI visibility data from which to do reporting or create strategies from.

If you want help with this type of work, feel free to contact me. I’m an independent SEO/GEO consultant helping brands and agencies with their AI visibility goals.

Until next time, enjoy the vibes:

Thanks for reading. Happy optimizing!

Related posts

Editorial history:

Created by Ethan Lazuk on:

Last updated:

Need a hand with your SEO/GEO strategy?

I’m an independent SEO/GEO consultant based in New York City. Contact me for more information!

Leave a Reply

GDPR Cookie Consent with Real Cookie Banner