Why Your Everyday AI Tool Isn’t A Good Source for AI Visibility Data
Now that AI overviews have taken over the top half of almost every page of search results, insurance executives are increasingly interested in what it takes to be cited or recommended in that coveted space. In the quest to find out how their company stacks up, and how AI engines perceive them, they turn to AI.
The problem? They ask the same AI tool they rely on every day, which is a little like asking your mom. It’s not objective at all.
For example, if you use ChatGPT every day, it may remember your company, your projects and how you like your answers. It's built to be helpful, so it tends to tell you what you want to hear. It already knew your company name before you posed the question. Of course, your company will be on the list of top performers in its space.
We tested this on our own brand. When we asked our AI assistant what it knew about Inbound Insurance Marketing, we got a robust report of how we appear online. When we tracked the questions our buyers actually ask, using neutral visibility tracking tools, we got a reality check.
The elements of AI visibility
To find out how you are really perceived by AI engines, it’s important to use an AI visibility tracking platform. Many are now available. We recently tracked the Inbound Insurance Marketing brand using three AI visibility trackers: Scrunch, HubSpot's AEO tool and Peec.
AI visibility tools measure several things, and each answers a different question. Here's what the most common terms mean.
| Term | What it measures | What you need to know |
|
Prompts |
The questions a tool asks AI platforms on your behalf, such as "Who writes case studies for insurance companies?" | Your results are only as good as your prompts. They should match what your buyers actually ask. |
| AI engines | The AI platforms being checked, such as ChatGPT, Google AI Overviews, Gemini, Perplexity and Copilot. | Each engine answers differently, so a brand can be strong in one and missing in another. |
| Visibility (also called mention rate or brand presence) | The share of AI answers that mention your brand. | It’s important to look both at your visibility using branded prompts (that include your name) and unbranded prompts that ask open questions. |
| Position | Where your brand appears in an answer when it's mentioned, such as first in a list or further down. | Higher positions are more advantageous. |
| Sentiment | Whether AI answers describe your brand positively, neutrally or negatively. | Each tool scores this on its own scale. This is not the same as satisfaction. A neutral or negative score could mean there is ambiguity surrounding your brand. |
| Share of voice | Your mentions compared with your competitors' mentions on the same prompts. | This is how you stack up, but it’s important to include actual competitors for an accurate comparison. |
| Citations (also called sources) | The web pages AI tools link to or draw from when they answer | This tells you which sites AI trusts on your topics, including your own pages, LinkedIn, publications and competitors. |
No single number tells the whole story. Visibility shows whether you're in the conversation. Sentiment and position show how you're presented. Citations show when and why you appear.
IIM's quick case study: Three tools, three different answers
| Tool |
IIM visibility How often your name appears |
Sentiment The tone used when mentioning your brand |
Share of voice (SOV) % of mentions your brand receives compared to competitors |
| Scrunch | 15.8% mention rate | 100 out of 100 (positive) | We scored 13.8% SOV. One brand edged us out at 18.3%. We chose the competitor list. |
| HubSpot AEO | 19.0% mention rate | 1.39 on their scale (neutral) | We scored 76% SOV. HubSpot suggested competitors based on our prompts, including some UK companies. |
| Peec | 14.1% mention rate | 60 out of 100 (neutral) | IIM had 63.3% SOV compared to 6 competitors they identified. |
Across all three tools, Inbound Insurance Marketing’s visibility landed in a similar range. The other metrics varied, based on differences in how the tools run their calculations and differences in user set up. Another thing to consider is that these metrics were derived from a short window of time. If we tracked over a longer period, more similarities might develop between the three tools' findings.
Why one search can't tell you where you stand
- AI answers change every time. This is not like SEO, where you might achieve a top 3 page rank and stay there for the next six months. AI tools write a new answer for each question received. Scrunch logged 831 responses from 117 prompts in one week.
- Your chatbot knows you. Your AI tool can draw on memory, past chats and the context of how you’ve collaborated in the past. The answer you see may not be the one anyone else sees.
- The question matters more than the brand. AI tools named us in 100% of responses to "who offers insurance blog writing services." On "how to" questions, such as turning SME interviews into articles, we showed up 0% of the time. Buyers ask both kinds of questions. Once you identify questions where you don’t show up, you can start plugging the holes by adding new content to your website and LinkedIn feed.
- Fame doesn't carry over. Two respected insurance marketing firms showed 0% presence in Scrunch. Being known in the industry isn't the same as being cited by AI.
- Each engine is different. Every AI tool shows different results. Being strong in one doesn’t mean that you’ll be strong in another.
- You can change your results. When you see that you're not being cited for a certain prompt, you can work to influence those results by publishing outcome-driven content on your website, your LinkedIn channel and in other places. Ongoing tracking enables you to see what's working.
What every tool agreed on
LinkedIn was the most-cited domain in both HubSpot and Peec. Our own website came second. This tells us that in the insurance marketing space, AI tools are reading what you publish on LinkedIn and on your site. Those are the two places to invest first within our industry. We ran a similar test for an estate planning law firm, and the results were not the same. In that case, many AI answers were sourced from Quora, the state bar and law directory websites.
Essential practices
- Track a fixed set of prompts over time, not one search.
- Include "how" and "why" questions, not only "who is the best."
- Add your real competitors so share of voice means something. At the same time, don’t be too narrow. We specialize in insurance content, but one of the competitors that AI cites for our prompts is a industry-agnostic content provider. A video content provider also appears. I would never have considered these companies to be my competitors, but this is another reality check. If they are cited when potential buyers search AI with insurance content prompts, then they are, indeed, my competitors.
- See which sources get cited, then build content assets that answer the questions where you're missing.
- Review the trend monthly. One snapshot is a guess; a trend is a signal.
Want to know where your company actually stands?
Request a free 7-day AI Visibility Snapshot (limited to 5 a week, first come, first served).
We'll track your key buyer questions for 7 days within our agency version of Peec and show you where AI tools cite you and where they don't.
The seven-day snapshot is a quick test. After the 7 days, you will have the opportunity to continue tracking and improving your visibility, with an ROI-focused insurance content program.