Why AI visibility metrics matter?
If you type into ChatGPT: What are the best brands in
But the main question is: what comes next?
If your brand is not showing in AI responses, you need to understand why. If it is showing, you need it to show up more frequently. Waiting too long can mean losing customers to competitors.
In this article, we explain the main AI visibility metrics, how to diagnose your brand’s performance, and how to take action to improve your visibility in AI.
What is the difference between GEO and AEO?
GEO stands for Generative Engine Optimization, which is used to improve the presence and order of the website in AI responses. AEO stands for Answer Engine Optimization, which is the process of optimizing to get mentioned in AI responses.
In short:
- GEO focuses on websites mentions AI responses
- AEO focuses on brand mentions in AI responses
Tracking GEO and AEO Metrics
AEO Metric
AEO, optimizing the brand to show in AI responses, can be measured as the frequency of responses citing the brand. More concretely, it is dividing the number of responses citing the brand by the total number of responses as follows:
$$AEO = RB / R$$
RB refers to Responses with Brand mentions, and R refers to the total number of Responses.
GEO Metrics
The GEO metrics we use here are inspired by the original GEO paper. One of them looks at the mentions from the prompts perspective, while the other looks at them from the sources perspective.
Prompt GEO Metric
The prompt GEO metric is the frequency of responses with at least one website citing the brand.
For a single prompt run, it can either be 0 or 1. If none of the cited websites mention the brand, then the prompt GEO is zero. Prompt GEO is one if at least one website cited the brand for that prompt.
For several prompts and runs, these numbers are summed together and divided by the total number of runs as follows:
$$GEOP = NC / R $$
NC, which is short for Normalized Citation, is the sum of zeros and ones of GEOP for different prompts. The R represents the total number of Responses (one response per prompt run).
$$NC = \sum_{i}^{n}{GEOP_{i}}$$
Source-level GEO Metric
This focuses on the brand mentions in the websites cited by the AI. Not only the websites cited in one prompt, but by all of them, combined.
Simply put, it is the frequency of brands showing in cited websites. It is computed as follows:
$$GEOS = WB / W $$
GEOS refers to GEO for Sources, WB represents the Websites citing a Brand, and W refers to the total number of Websites cited by the AI (or search engine).
AEO vs GEO Metrics
Since the AEO and GEO prompt metrics look very similar, some might wonder what the difference is between the two, as both compute the mention frequencies per prompt.
These are the main differences between the two:
- Pre-learned information. A model might use previously learned information and mention brands that are not on the websites. So it relied on the information the model was trained on rather than the websites. This can be the case for AEO, but not GEO.
- Missing Information. A model might miss information or brands already mentioned on the website, but it did not consider that information important enough to include it in the answer.
- External Information. Some websites fetch information from external websites (e.g., APIs), which might not be captured by GEO. This is a common issue faced by crawlers and search engines. Certain cases are captured by Google as they use some tricks to capture such information.
- Dynamic Loading. JavaScript is sometimes used to show the information gradually (e.g., with animation), which can cause website crawlers to miss such information, leading some brands to show in AEO but not in GEO.
How to take action for GEO and AEO
When the brand mentions is low for both AEO and GEO metrics, then it is probably an SEO-related issue (e.g., backlinks, keywords, structure, quality, ...). But when it is high for one and low for the other, then it is one of the two:
- Low AEO and high GEO. This is likely due to the content or the structure of the website, which led the model to decide not to include the brand in the response. Or maybe the model did not understand the content, or its relevance (context). To solve this, the content quality needs to be improved if that was the cause. If the relevancy is an issue, then working on the clarity and readability can solve this. Making sure the page is properly structured (e.g., h1, h2, …) so the model can clearly understand and match the relevance of the snippet to the user query and maybe even the user intent.
- High AEO and low GEO. This could be a crawling issue, where the relevant information couldn’t be fetched due to the way the information is being loaded into the page (e.g., external information or dynamically loaded content). The other cause is that the model is reciting information it was trained on, and in that case, the user needs to improve their SEO and optimize their website (e.g., keywords, backlinks, authority, sitemap, …).
We summarize these cases in the following table:
| AEO | GEO | Actions |
|---|---|---|
| High | High | You are good already. But try to optimize for order, to show first in the cited pages in responses, and increase your coverage. |
| Low | Low | Focus on SEO. |
| Low | High | Optimize your website content: high-quality, easy to understand, and proper keywords. |
| High | Low | Improve SEO and use a static/preloaded version of the website. |
Conclusion
AEO metric focuses on brand showing in the AI answers, while GEO metrics focus on brand mentions in cited websites. It is important to measure both, as different actions might be required depending on which is high and which is low.
But brands that are not tracking their presence in AI are missing out and might not know why more prospects are going to their competitors, and why their performance is lacking in the AI game.
To improve your presence in AI systems (ChatGPT and Gemini), you first need to see where you stand in AI visibility (AEO and GEO). Start tracking your brand in AI with buzzsense.ai, and let us help you win in the AI visibility race.