Does more = better? In this test, the number of brand mentions does impact visibility, and brands with fewer mentions barely show up in AI answers.
One question we had was, “how many mentions does it take to show up?”
We decided to run a test by building our own small language model from scratch. To do this, we needed to create a ‘world’ of knowledge that lived inside the LLM, which we decided to focus on project management software.
We used known brands like Trello and Asana and included five companies we completely made up. All in all, there were 33,000 sentences across our training text written about various project management tools, use cases, summaries and more inside the LLM designed to replicate on a small scale what is found on the web.
Look at the breakdown:
| Company | Mentions |
|---|---|
| Company A: | 1600 mentions |
| Company B: | 20 mentions |
| Company C: | 59 mentions |
| Company D: | 533 mentions |
| Company E: | 178 mentions |
| Asana and Trello: | 1400 mentions each |
We trained our LLM on this text and made sure to control that the only thing different about the companies was how often they were mentioned, not what was said about them.
The experiment:
We then asked the LLM questions similar to what a consumer would ask, testing how likely the LLM was to mention each brand in the context of project management, instead of counting mentions in the generated answers. We did things like ask which project management software was best, which project management software had X feature, and other similar questions.
The results:
Company A came out as the brand most likely to be recommended, even surpassing Trello and Asana, even though it was a completely made-up company that didn’t exist before that morning.
The companies mentioned 20 and 59 times barely registered at all. When we asked the model which company fits, it gave them almost no chance, as if they didn’t exist.
To test this, we ran the test four separate times and got the same overall pattern every time.
The conclusion:
In this test, brands with more mentions had stronger visibility. In this case, it would appear that more = better. Of course, this doesn’t account for weighting in the mentions, meaning, it doesn’t tell us if a mention on certain sites matters more than other sites. However, getting a mention here or there may not be enough to improve your visibility.
If you have been trying to increase your brand mentions and not seeing results in terms of growing visibility, you may want to look at the quality and volume of your mentions.
This is not all that different from link building for SEO. Sites are often measured by site strength (DR if you are looking at an Ahrefs metric). More high-quality links will likely increase your DR. A stronger backlink profile can help you rank for more keywords.
If you are struggling to scale brand mentions, you may need some help in this area. Reach out for a demo of Seed to see how scaling quality mentions can increase your AI visibility.
Written by Jason GilbertBrand placements. AI visibility. A clearer way to see the whole picture.