
AI search is growing, and people want to know what traditional SEO strategies still matter and which can be left behind. AI Search still has a long way to go before it will be a true rival to the traditional Google search, but it is important to be prepared, as it is where things are moving. Ultimately, the underlying problem will remain the same: the system needs a way to decide what to trust.
Backlinks are still one of the cleanest shortcuts the web has to trust. Even when an AI system is generating an answer, it still has to ground that answer somewhere. In most cases, that “somewhere” is a set of sources that already earned visibility and credibility on the open web.
Even if you don’t buy into the SEO buzzwords, links have always functioned like third-party validation. Not every link counts the same, but the concept is simple: when reputable sites reference you, search engines treat you like you are more likely to be legitimate. The more of these, typically the more visible your site has been in traditional SEO.
In classic SEO, people obsessed over anchor text and raw domain authority. In AI search, the sentence the link lives in often matters as much as the link itself (which is where traditional SEO was trying and actively moving anyways). Models attempt to understand “why was this referenced” not just “was it referenced.”
That is why an editorial mention inside a real explanation tends to outperform a random link dropped in a directory list. The model can extract meaning from the surrounding text, and that meaning becomes part of how it frames your brand in future answers.
The pattern that keeps showing up in studies is that AI answers tend to cite a smaller set of sources repeatedly. Forums, Wikipedia-style resources, and established publishers show up a lot. In other words, the model is not “discovering” random new websites every day. It is leaning on sources that already have strong trust signals.
That matters for backlink strategy because you are not only trying to rank. You are trying to become referencable. You want your links to be on pages an AI system would feel safe pointing to. This is why you see so many turning their focus to Reddit and Wikipedia references as a link strategy.
Entity-level SEO is basically the internet agreeing you exist and agreeing on what you are. If your brand name, services, people, and locations are mentioned consistently across multiple trusted sites, it becomes much easier for search engines and AI systems to connect the dots.
This is where backlinks and brand mentions start to blur together. A linked mention is great, but consistent unlinked citations across credible sources still build the “entity footprint” that AI systems can recognize. Not every Reddit reply you leave needs to have a link, sometimes a mention can be seen as more authentic!
Digital PR is one of the more “future-proof” link plays because it creates the exact pattern AI systems like: a credible site, an expert quote, and a clear attribution trail.
It also forces you to show your work. You are building a link as well as a record that your business is a legitimate source in a niche. Over time, that turns into repeat citations, compounding on the one-off wins.
The biggest misconception is that you need a totally separate playbook. You do not. The best practices are converging: be useful, be cited, be consistent, and earn mentions in places that already have trust.
Yes, you should pay attention to how AI interfaces display answers and citations. But if your approach is “we do SEO over here and AIO over there,” you are going to split effort and get mediocre results in both.
Backlinks are still critical, but the job is shifting. It is less about pushing a page up a list and more about becoming a source that the answer layer feels safe pulling from.
If you build links that come with real context, real relevance, and real third-party validation, you are building the footprint AI systems will keep rediscovering and reusing. For now, keep steady with your traditional strategy with continual adjustments based on what is proven to have worked over 6+ months, there are still a lot of kinks being worked out by these AI tools, don’t jump on every trend.