With the emergence of Google’s new generative AI search, AI more broadly, and the fate of SEO as we know it hanging in the balance, a lot of websites and link builders are wondering, understandably, what is the future of link building?
What link building strategies and approaches are going to have an impact on rank and organic traffic moving forward? How will artificial intelligence factor into my link building strategy and what do I need to know about it?
In the below article, we’re going to dive into where we see the future of link building going, how you can plan for it and what kind of changes you can expect to see as Google continues to upend traditional search.
Understanding Bard
If you haven’t already spent some time getting to know Bard–Google’s experimental generative AI, it’s worthwhile spending some time on it before moving onto link building.
In Google’s words:
Currently only being tested in the United States and the UK, there are a few important things to understand about Bard and how it works.
LLM Models
First, Bard is what’s known as an LLM (a Large Language Model).
A Large Language Model is a type of algorithm that utilizes deep learning, an AI method that teaches the algorithm to predict and classify information. Its efficacy is based on its initial programming with massive data sets (essentially all of the known web) and constant interaction with users.
What this means is that you ask Bard a question or give it a prompt.
And it provides you with suggestions–either unlinked information that Google itself has curated from its database or that same information with a short bibliography of “sources” to choose from.
This new search AI will only get better and better at understanding search queries and providing results–as well as filtering out irrelevant ones.
It’s also worthwhile understanding that Bard is a lot more complicated than just “ChatGPT”--a common misunderstanding right now.
Bard is an amalgamation of several AI types
Google’s new AI search is a combination of different AI models that all do different things. The end result: a new search experience that is vastly different from regular online search.
Standard world wide web search currently involves manual research.
You have to type a question or keyword into a search engine and let the search engine decide which pages to show you based on proprietary algorithm logic (Google’s RankBrain or Penguin, for instance).
The defining characteristic of traditional search is that it does not create content. It syndicates it from a variety of websites. You click the result and are taken to the website where that content is hosted.
With generative search, content will be limited to what the AI synthesizes and, perhaps (though not guaranteed), links to the source material.
Understanding Discriminative vs Generative AI models
Bard is a hybridization of various AI types, which do different things, all of them combining to create this revolutionary new way of querying and showing information.
Below we will touch on each of these AI models and what they do.
Discriminative AI vs Generative AI
The first component of Bard is generative AI technology.
In a discriminative model, human supervisors train AI to be able to distinguish between different objects in a sample. For example, if you gave a discriminative model 10 images of 10 different fruits as inputs, the discriminative model will tell you the difference between the various subjects.
Generative AI, on the other hand, using sample data, creates objects with little to no supervision.
Generative machine learning allows AI to understand inputs while preserving the understanding in its neural network memory so that it can call on this experience at any time in the future when similar or related challenges or queries arise.
Generative Adversarial Networks (GAN)
A Generative Adversarial Network (GAN) is a machine learning algorithm that is trained on both generative and discriminative models. The generative component creates samples from inputs like questions and keywords, while the discriminative model identifies whether the sample created by the generative model is fake or original input.
If the discriminative component determines the generative component is fake, then the generative model creates new output for the discriminative model to analyze. This continues until the generative model is capable of creating fakes that the discriminative model is unable to tell from the original input.
Transformer-based models (TBM)
Transformer models are deep neural networks that analyze inputs to predict what the output could be.
For instance, if you were to give a TBM a string of unrelated words or terms, it would analyze these words and then try to predict either prior or proceeding words that could be inserted to create coherent sentences.
ChatGPT is an example of a transformer model (the GPT in ChatGPT stands for “Generative pre-trained transformer model).
Generative AI search utilizes all of the above models in conjunction to provide you with images, videos and text that, in theory, better correspond to what you’re searching. Instead of providing you with a list of websites and content to choose from, it chooses the websites and content for you.
In theory, it’s all good news for users, but, of course, quite scary for anyone having to contend with what this will do to their online business model, especially if it is an SEO and organic traffic-reliant one.
Bard’s impact on traffic
If you’ve searched for something like “how to lose weight fast,” for example, what Google’s new generative AI search shows you is something that looks like the below:
Notice how the AI’s output is what dominates the top of the screen. Also, notice the three featured images and clickable links at the top right of that box?
This is what has a lot of SEOs quite nervous.
What is becoming quite apparent in the wake of all of this beta testing right now is that Google is going to be cannibalizing a lot of organic search traffic.
Not all of it, of course, because their business model still relies heavily on ads placed in and around content, but click-through rates are up in the air right now.
And that’s in the wake of already declining clickthrough rates as Google has been consistently pushing website links further down the SERP in favour of more ads, more video content, and more SERP features.
There is currently no Google Search Console data for Bard search performance and it’s not possible to say with any certainty how Bard will choose which sites to showcase and which to condemn to obscurity.
What is likely to happen
While it’s important to emphasize how unknowable the future is at this point, there are a couple of scenarios that most SEOs acknowledge are likely.
- Google could end up only providing links to the top 3-5 search results.
- There could be an even bigger emphasis placed on Google’s E-E-A-T guidelines (Expertise-Experience-Authority-Trustworthiness).
Both of these will like happen in conjunction and work dynamically. Based on these two likelihoods, the future of link building is a bright one.
A link building strategy as a means of improving E-E-A-T
Expertise, experience, authority and trust are the foundation of Google’s helpful, reliable, people-first content.
Inbound links definitely help send trustworthiness and reliability signals to Google when analyzing and ranking content.
If Google’s modus operandi with its new AI-powered search engine is to really weed out the unhelpful content and showcase the best of the best, then high quality links from reputable high domain authority websites in your niche are going to be what makes you stand out to Google.
What is likely to happen as search becomes a lot more granular and Google’s generative AI results dominate much of the SERP (displacing websites) is that domain authority (based on backlinks) as a proxy for ranking is going to become more more granular.
We never get tired of stressing the fact that not all links are good links. Websites that focus on a lot of low quality links over fewer high quality backlinks are going to struggle in the new era of search even more than they already do.
Black hat link building tactics, which are already pretty much useless, will be pointless.
Link building campaigns will have to become a lot more selective and well-informed in order to send the kind of authority and trust signals that Google will reward with even scarcer real estate at the top of the search engine results page.
A side note on anchor text
Another way in which link building will remain as if not more important than it already is, is through anchor text selection.
Google has already acknowledged that anchor text is a ranking signal for the target page and also helps to classify the entire domain it is pointing at.
Anchor text has also been referenced as a trust rating, which acts as a supplemental indicator that Google can use when ranking pages.
Anchor text is an important part of any good link building editorial policy, but it is likely going to become even more important as time goes on and Google’s traffic cannibalization via new generative AI starts to really hit its stride.
A diversity of different anchor text types, including natural, keyword-optimized, naked URL and branded anchors, is crucial for a good backlink profile and something that experienced SEOs and link builders can really help with.
Link building strategies as means of driving traffic to your most commercially valuable pages
Link building is also going to increasingly be important for driving traffic to a website’s most commercially valuable pages, especially those that stand to lose the most from traffic cannibalization.
What generative AI is likely to do to search engines
Ryan Stewart released a good breakdown of the new AI search and what he sees as the future of SEO.
Of the various takeaways from his video, perhaps the most important is the idea that Bard is going to end up eliminating or heavily reducing traffic potential for a ton of search terms moving forward.
Stewart’s caveat here is that there will still be plenty of search terms with strong commercial intent that AI will not completely takeover, but it’s going to be hard to predict which types of search terms are going to get hit and which won’t.
If commercial content that was formerly ranking is now no longer visible in the SERP, then you’re going to need to find alternative ways of driving traffic to those commercially valuable pages.
If you rely on organic traffic for leads, building links to other pages that you can rank for and for which Google doesn’t dominate the SERP with AI-generated answers and prompts, will be paramount. This will involve modern keyword research aimed at more specific long-tail keywords, a good understanding of Bard’s algorithmic decision-making (which we still don’t have) and an ability to find relevant, high quality link building opportunities.
The importance of user-centric links moving forward
Search engine rankings and results are likely to become more unpredictable moving forward as search morphs into something much more interactive and much better at understanding context, providing a better user experience.
It won’t be enough to simply get more links or to match your competitor’s backlink profile –although the backlink profiles of the top-ranking sites are certainly still going to be worth emulating.
Google, increasingly, is only going to show searchers relevant websites that really satisfy search intent. The SEO spam that often dominates the SERP likely isn’t long for this world.
Your inbound links from other websites will likely vouch even more for the relevance of your content and its appropriateness, meaning link building strategies are going to need to become even more selective.
Search engine rankings–or, perhaps, visibility is a better way to conceptualize it–will become more a function of how you build relevant links.
Start planning for the future now
The future is coming at us fast. The GPT models used to tailor search results will only get more sophisticated, meaning even more refined and accurate results.
We all, site owners and agencies alike, need to start future-proofing our processes and strategies so that we can continue to enjoy strong organic traffic.
Generative AI search is going to completely revolutionize how we access information and make economic decisions. No industry will be left untouched.
If you want to find out more about how an agency that is already planning for this future does user-centric link building, in a way that is performance-based and contract-free, get in touch with dofollow today.
Why Trust Us On SEO
Eric Carrell & Sebastian Schaffer have been working in SEO for over a decade, building their own projects - understanding and testing SEO strategy, along with building hundreds of white hat links per month for our projects. They take their learnings and experience and apply them to the strategy that drives our link building strategy for our clients.
Eric & Seb have always believed in quality over quantity, doing things the right way so we future proof our client’s websites against future Google updates and the evolving industry of search.
While Seb handles the company strategy around culture, processes and structure, Eric is constantly working to improve our service offering, customer experience, and following the industry in parallel with Google’s Quality Guidelines so that we are always one step ahead of our competition and aligned with what Google wants to see for your site to rank higher.