If you’re worried about if you’re showing up in AI search, there’s a really simple way to check on your visibility. In fact, you can do it yourself, today, for nothing – assuming you already use at least one platform in your day-to-day work.
Here’s how to do it, and how to know when you need to act.
Start with what your buyers are asking
While it’s great to understand where you perform from a brand point of view – such as asking ChatGPT “what does [your company] do?” – it’s not going to tell you much about what prospects are asking in the initial research stages.
What’s more important is whether you appear in the generated response when someone asks about a problem you solve, or a product or service you offer, without being familiar with your brand yet.
To understand this, you need to generate some test prompts that sound like your prospects (not your own marketing content). If you’ve done keyword research for SEO purposes, or if your sales team is willing to share some insights from their inboxes, these are good places to start. If not, try to put yourself in the shoes of your buyer and ask things such as:
- “Who are the leading suppliers of [your product or service] in the UK?”
- “I need a [service] for a [sector] business – who do you recommend?”
- “Best UK companies for [specific technical capability]”
If you use a chat agent, you could link to your website, your keyword research, and any other information that might help, and ask it to generate the prompts based on what buyers ask.
Run the prompts more than once
AI answers are volatile because they’re probabilistic rather than deterministic, which means the same prompt can return different brands on different days – even between moments. This is because indexes change, models get updated, and there’s a degree of randomness to how responses are generated. Run your prompts just once and you get a snapshot of a single moment, run it three times in one sitting and you start to get an idea of trend.
Monitor if you appear, where you appear – at the top, middle, or bottom of a response – what sources are cited, and what competitors appear. The consistency (or lack thereof) is what you’re looking to measure, which is why AI visibility is measured as a percentage – the chance of an agent mentioning you in a generated response – rather than a hard figure.
Before you start, turn and turn off as much personalisation and memory recall as possible, as well as turning off your location settings. All these parameters will bias your answers, based on your past chats with the agent or where you are based. It’s difficult to get a completely unbiased response, but you can do as much as possible to mitigate the issue.
Additionally, make sure you check across more than one platform, as they often offer wildly different answers. ChatGPT, Gemini, Perplexity, and Google’s AI Overviews are a good start as they’re currently the most popular platforms. Being strong in one and missing entirely from another is normal as they add different weights to different sources. The gap here is what you’re trying to understand – and how to bridge it.
Ideally, you also want to not just run the prompts three times in a single sitting, but then carry out the same test over the course of a month to see how the trend holds up over time.
Analysing what comes back
You want to understand three things, which all have different meanings:
- Are you cited?
Cited is what we’re looking to achieve. Your brand is mentioned in the generated answer, and the sources/citations link to your owned or earned content. This means it’s treating you as evidence, and you’re controlling the narrative. - Are you mentioned?
You’re mentioned by name but nothing links back. You’re in consideration, but not a leading source. - Are you absent?
Your competitors are mentioned but you’re noticeably absent. Note which competitors are mentioned, what sources are cited, and what types of content are cited, and where.
Absent is where you need to act immediately. Mentioned is where you need to up your content game. If you’re cited and highly recommended, you need to ensure you’re continuing with regular and consistent comms that drive the same narrative.
What to do next
Study the sources behind the brands that do appear. Rarely will it be their own website doing the heavy lifting. It’s often third-party coverage that is doing the real work – trade press, expert commentary, industry reports, whitepapers and original research. These are the sources a human buyer trusts, and therefore are credible to the LLMs.
While technical and on-page SEO really do matter – a clean site structure and schema go a long way to helping a machine understand who you are and what you do – they can’t create authority that doesn’t exist anywhere else on the internet. This is a PR problem, but one that can be easily solved through a strategic media relations and content programme.
When DIYing is not enough
The method above can highlight any glaring issues, but for deeper insight and understanding, and for support in developing and implementing a plan to help bridge your gaps, and monitoring its effectiveness, get in touch.