New findings suggest domains with links from a wide range of websites are cited more often than those with Google page-one rankings.
In traditional SEO, startups need to compete against established brands with bigger budgets and years of investment to get ranked on Google.
AI search offers a new way into the visibility game. And getting noticed by AI engines depends less on where you rank on Google and more on how often you’re linked on third-party platforms.
Our latest analysis at Earned Media examined 360 AI search responses spanning ChatGPT, Gemini, Claude and Perplexity in Australia, across several verticals.
One finding stood out among the more than 3,000 citation events: unique referring domains predict AI citations better than Google rankings and Ahrefs’ Domain Rating, a metric that measures the strength of a site’s backlink profile.
A referring domain is a distinct website that links to your own domain. Let’s say a brand is featured in 10 articles on the same news site. That counts as 10 backlinks, but just one referring domain. If that brand earns coverage across three publications, with each platform linking to the brand’s domain, that’s three backlinks from three unique referring domains.
The pattern held across 137 domains we measured, with our analysis finding the relationship between unique referring domains and AI citations statistically significant, with less than a 0.1% chance of coincidence. We ran a similar test in the US, and the same pattern showed up across 141 domains, with less than a 0.01% chance.
Our analysis found domains with links from a broader range of websites tended to receive more AI citations. It’s important to qualify this finding: correlation doesn’t establish causation. A broad backlink profile doesn’t automatically mean your pages will show up in AI search results every time. But the pattern is consistent enough to warrant a closer look.
More than a third of Australians online start a new search with an AI assistant, according to a recent YouGov survey. And that search can often stop there, with Ahrefs reporting that the presence of AI Overviews reduces clicks on top-ranking content in Google by 58%.
That’s a fundamental shift in how people discover products and services, and it creates a new consideration for brands vying to get in front of potential buyers.
While established brands may have bigger budgets and years of SEO investment behind them, they can also be slower to adapt. Startups, on the other hand, often have fewer legacy processes to push through, making them agile enough to adapt to new shifts faster.
Our analysis adds weight to this opportunity. Earned Media found that even with big-name sites like Reddit, Forbes and Canstar in the mix, the top 10% most-cited domains captured only around a quarter of total citations. Most activity was spread across a long tail of smaller sites.
Given AI citations aren’t already confined to a handful of dominant domains, this presents a competitive advantage for startups to get in early and start building a wider spread of relevant links and mentions from third-party platforms.
For entrepreneurs, the opportunity here is to be deliberate about where you build credibility. Here are four ways to get started:
- Identify sources AI already trusts in your category by testing relevant buyer questions across ChatGPT, Gemini, Claude and Perplexity.
- Build relationships with journalists, industry associations and relevant publishers to create opportunities in the future to get referenced on their sites.
- Create unique content worth citing, whether that’s original research, expert commentary or practical insights.
- Look for knowledge gaps in AI search, which is what we call negative space. This is where AI invents or contradicts information when answering high-intent questions.
The biggest takeaway for startups wanting to show up in citations across ChatGPT and AI Overviews is this:
Build a backlink profile that’s focused on building links from a broad set of trusted referring domains, not just one or two sites.
- Nick Brogden is the founder of Earned Media.
