“`html
If you have spent any bmmagazine.co.uk time in the SEO or content strategy space over the last twelve months, you have undoubtedly noticed a shift in the pitch decks landing in your inbox. Where we once discussed “SERP visibility” and “organic share of voice,” we are now seeing a new, shiny metric: brand radar impressions. You’ll see it promised in SaaS dashboards, marketing reports, and high-level strategy presentations. But before you start building your quarterly reporting around it, we need to ask the golden question: Where does the data actually come from?
As someone who has spent over a decade navigating the messy intersection of enterprise search and BI dashboarding, I’ve seen enough “visibility scores” to know that when a metric feels too good to be true, it is usually because it is hiding a lack of methodology. Let’s strip back the marketing jargon and look at what the AI impressions metric really entails, and why measuring it is more complex than simply tracking a blue link.
Beyond the Blue Link: Traditional SEO vs. AI Visibility
For years, SEO professionals have relied on platforms like Ahrefs to tell us where we rank. That data is relatively binary—you are either at position 1, 2, or 10, or you aren’t there at all. It is measurable, verifiable, and—most importantly—consistent. AI visibility measurement is a different beast entirely.
When we talk about “AI impressions,” we are moving away from a deterministic list of search results and into the realm of probabilistic generation. Whether it is Google AI Overviews or a direct query to ChatGPT, the “answer” is dynamic. If your brand is mentioned, cited, or surfaced within an LLM response, that counts as an impression. But unlike a traditional SERP position, an AI impression can change based on the user’s conversation history, the time of day, and the specific prompt used.
The Methodology Gap: Who Is Watching the Watchers?
The market is currently flooded with tools claiming to solve this black-box problem. Emerging players like Peec AI and Otterly.AI are doing interesting work in attempting to quantify how brands appear in generative contexts. However, the industry is rife with “hand-wavy” methodology.
When evaluating these tools, look for the following red flags:
- Lack of Source Transparency: If a dashboard gives you a graph showing a 20% increase in brand mentions, ask what the crawler is. Is it a live browser session? Is it an API call? Is it a simulated persona?
- The “Per-Seat” Trap: I’ve seen too many tools that charge per seat, only to hide the meaningful “export to CSV” or “API access” features behind a prohibitively expensive enterprise add-on. If you can’t get the raw data into Looker Studio for a sanity check, you don’t own the data—you are just renting the chart.
- Answer Engine Coverage: Are they tracking Google AI Overviews exclusively, or are they checking across ChatGPT, Gemini, and Perplexity? A brand radar that only looks at Google is a legacy tool dressed in AI clothing.
The Perils of Regionality and Prompt Injection
This is where most of the current reporting tools fall apart. If you are a UK-based retailer, but your “brand radar” is tracking impressions from a data centre in Virginia via a US-based proxy, your data is garbage.
Regional tracking done via prompt injection without clear documentation is a major issue. Many tools use standard prompts like: “List the best [Product Category] in the UK.” However, LLMs are sensitive. If a tool doesn’t explicitly state that it is using localised IP addresses and regionalised user-agent strings, it is likely hallucinating your visibility data based on US-centric training weights.
The Prompt Injection Pitfall
Some automated tools attempt to “force” visibility by injecting specific brand-related prompts into the LLM during the collection process. This is not organic discovery; it is a synthetic test. If your report shows a high AI impression rate because the tool is literally asking the AI about your brand, that is not a measure of market share—it is a measure of the tool’s own query bias. Always insist on seeing the exact prompt set used to gather the “visibility” score.
Comparison of AI Measurement Approaches
To help you distinguish between high-quality data and “fluff” metrics, I have put together a breakdown of how we should be evaluating these platforms:
Bringing AI Impressions into BI Dashboards
Once you’ve found a tool that actually discloses where the data comes from, don’t leave it in their siloed dashboard. The biggest mistake B2B teams make is viewing “AI impressions” as a standalone metric. It needs to be connected to your other internal data.
I recommend pulling the raw data from your chosen AI tracker and pushing it into Looker Studio. Why? Because you need to overlay it with your internal conversion data. If you see a spike in “AI brand impressions” but a flatline in direct traffic or branded search volume, you are looking at a vanity metric. If the AI is talking about your brand, but nobody is clicking through, you have a conversion problem, not a visibility problem.
Final Thoughts: Don’t Buy the Hype
We are currently in a period of intense experimentation. Companies like Peec AI and Otterly.AI are pushing the envelope, and tools like ChatGPT and Google AI Overviews are changing the rules of the game daily. But as someone who has lived through the rise and fall of countless “marketing tech” solutions, my advice remains the same:
AI visibility is not the new SEO; it is a new layer of the user journey. Don’t settle for “hand-wavy” visibility scores. Insist on data that you can verify, regionalise, and ultimately, use to make actual business decisions.
“`
