Off-site SEO in the age of AI, information retrieval & brand identity context

Off-site SEO was considered the fuel for rankings for the vast majority of SEO history, due to the PageRank algorithm and the very foundations on which search engines like Google were built. Recently, this has changed significantly. Links are still a very important factor (or group of factors) that build authority and allow websites to rank at the top of Google and other search engines, but the impact of off-site presence is now much broader. Off-site SEO determines discoverability and indexation, but the most important shift is within the “trust” group of factors.

From this article, you will learn:

• What the role of links in current SEO is
• How links and off-site presence impact AI search and recommendations
• What the difference between links and mentions in SEO is
• What tactics you can use to gain mentions and links
• How a well-crafted message on third-party websites can increase your ability to rank

Editorial note

I created this article as a sequel to my talk at the Chiang Mai SEO Conference in November 2025. Below, you can see the title of the talk, its description (as published on the conference website), and later you will find the slides with my comments. Full slide deck attached in the end.

Talk title: Offsite SEO in the AI era: Links, authority and control over chaos

Talk description: AI is changing and expanding the SEO ecosystem, but authority signals still run the game. In this session, Szymon shows how controlled link building based on paid media and even PBNs can fuel both rankings and LLMs. Expect real cases, practical tactics, and proof that the same methods deliver results across markets, from Poland to the US.

AI search era and FOMO that comes along

Let’s start with a fact: a large part of the SEO community feels overwhelmed by the flood of new insights, studies, reports, case studies, and tools. It is all about AI. The year 2025 was full of new terms and acronyms created to describe this “new reality” of organic traffic acquisition, or more broadly, organic marketing:

• AEO – Answer Engine Optimization
• GEO – Generative Engine Optimization
• LLMO – Large Language Model Optimization
• AIO – AI Optimization (although some people also use it to describe AI Overviews, which were rolled out globally at the end of March 2025)

This led me to ask the general question:

My answer was intentionally ironic: I called it LMAO. Why? Because focusing on naming conventions like this is not serious. It is mostly about selling products and services, not about improving quality or shared understanding, either for clients or for other SEOs.

I also called it the FOMO era of search, because nobody truly knows what is going on. Well, some SEOs do. Those who track changes, observe SERPs, analyze traffic data, and follow new snippets and channels can see how all these elements layer on top of one another. For example:

• Featured snippets + passage indexing = direct answers
• Direct answers + LLMs = AI Overviews

These formulas are not only logical, they are technologically connected. They are built by the same groups of engineers and supported by the same patents. They are also a consequence of one overarching goal: satisfying user intent without letting users leave Google.

Zero click searches, baby! I talked about that earlier:

• Redefining SEO in the New Era of Traffic Generation (SEO Estonia talk)
• Zero-Click Searches: How Google’s AI Mode and AI Overviews Are Redefining SEO (October 2025)

There is no “new SEO.” Things have just accelerated.

You need to understand that:

• Google has been an AI-powered search engine for over a decade. AI Overviews and AI Mode simply made this visible to the masses.
• What really matters is whether you are legible to Google’s semantic, retrieval, and quality systems.
• Google and other LLM-based tools learn about you not only from your website. They pull data from other documents, your off-site context.

Here is my simple framework, a mental model that helps me avoid getting distracted by “new shiny things.”

Whenever a new AI-search-related concept appears, I ask myself one question:

Does it improve semantics, retrieval, quality, entities, authority flow, or discoverability?

These are the foundations of search. If you understand how these systems work at their core, you gain a powerful filter for evaluating new trends. New concepts must be compatible with the existing system. There are exceptions, but they are rare and usually clearly communicated by Google or serious SEO researchers. In those cases, we are talking about real technological shifts supported by whitepapers, patents, or official announcements on the Google Search Blog or DeepMind Blog.

Examples include Gemini, MUVERA, BlockRank, or MUM being incorporated into Google Search.

Be reasonable, don’t panic, and verify your sources.

I strongly believe that the “AI vs SEO” dilemma is fundamentally false. I really admire Rand Fishkin’s opinion on this topic, which he shared in a comment on a Datos report about AI and search.

I recommend reviewing these reports:

• https://datos.live/report/state-of-search-q1-2025/
• https://datos.live/report/state-of-search-q3-2025/

Look at real data before claiming that ChatGPT is overtaking Google or repeating similar nonsense.

Now let’s focus on what actually matters: why off-site SEO is becoming more prominent and why this is happening.

Growing prominence of off-site SEO tactics

In recent years, content has become extremely cheap and scalable. If you have basic technical skills and access to your own data (preferably through RAG enriched with company knowledge), you can grow and maintain content efficiently.

Some may argue that off-site SEO is also scalable through PBNs or link blasts. However, there is a key difference. Google actively targets these tactics, and it is much harder to bypass anti-spam systems. On the other hand, if you scale content while respecting systems and attributes such as content effort, originality, semantic clarity, and proprietary insights, you are relatively safe. You can manufacture content, keep it fresh, and optimize it semi-automatically at scale, all without violating Google’s guidelines.

As a result, many competitors reach similar levels of topical authority. This is why off-site signals have become more valuable. They are often what truly makes the difference.

Off-site signals

What are the main elements of the off-site SEO puzzle? There are many, but during my talk I highlighted four:

• Backlinks
• Mentions
• Users (behavior)
• Brand identity

Traditionally, backlinks fueled authority. Mentions were an unclear part of the equation, but over the last year they have increasingly been described as “new links,” something almost equivalent. I strongly disagree with this interpretation.

User behavior translates into branded searches, clicks, and engagement. Brand identity refers to consistent messaging, visual identity, naming, and, most importantly, clear positioning and values, similar to the E-A-V model.

Whether brand identity is a direct ranking factor is debatable. I see it more as a catalyst or prerequisite that enables other signals to work effectively.

Ranking in Google is no longer about collecting ranking points from different domains or disciplines. It is a complex system of layered and conditional mechanisms. Scalable factors stop working if quality and trust thresholds are not met. The same applies to off-site SEO.

Are the brand mentions new links?

Treating mentions as equivalent to links is, in my opinion, a dead end.

Links are still essential for authority.

Mentions act as additional trust signals that provide context and reduce uncertainty, especially when they appear in relevant surroundings. Google wants to understand your business the same way it understands other entities.

How does Google learn about the world? Through co-occurrence of entities, their attributes, and their relationships. This is how the Knowledge Graph is built.

Read more: How Google Uses Entities to Understand Content

That is why some websites or pages rank better when a brand receives more mentions. Google can classify them more easily, lowering uncertainty and increasing trust.

Mentions are not an alternative to links. They are a parallel signal that confirms authority and provides context through third-party sources.

It is also worth noting that the majority of backlinks deliver similar value and primarily pass authority once quality thresholds are met.

Authority and trust are not the same.

Trust is critical at the retrieval stage (of pipeline), determining whether a document can even be considered. Authority determines who ranks above whom once trust is established.

Trust has a threshold. Authority is what secures the top position.

I believe it’s useful to look at those different signals as supporting each other.

Backlinks without trust signals look artificial. How often does a genuinely recommended brand exist without being discussed naturally, without links? Mentions are more natural than links, especially on social platforms, which Google increasingly targets, for example through the December 2025 BCAU update.

It is similar with branded traffic. If links and mentions are genuine, a brand’s popularity should be manifested in parallel by search and click activity in Google.

Then, the more people find out about the brand, the more people visit the website, and the more chatter around it should be measurable. This should result in mentions and links. Consider it an SEO momentum flywheel.

Off-site SEO momentum flywheel:
Mentions → recognition
Backlinks → authority
Brand-related search → confirmation

Brand mentions alongside branded search make growing link velocity appear natural.

And one rhetorical question to all those who argue that a mention is equivalent to a backlink from the same source…

If you could change an unlinked brand mention from a good source into a dofollow backlink from the same source, wouldn’t you? I thought so.

Off-site context

No matter whether we talk about links or mentions, they should never exist without context. Information is always retrieved within some context. Therefore, I believe that some links, and some mentions too, acquired deliberately, may do harm.

Nowadays, it is all about trust, clear positioning, context, and reducing uncertainty. This is why:

Random, irrelevant links and brand mentions without context can create noise.

If you heard my talks in Brighton (BrightonSEO 2024, about guest posts versus other tactics) and/or Saigon (SEO Mastery Summit 2024, about the use of sponsored articles), or if we have simply had the opportunity to discuss off-site tactics, you probably know that I am an advocate of pragmatic and controlled SEO, whether it is on-site or off-site.

Link building tactics that may be useful

I hate randomness when it comes to signals about a brand. It weakens positioning, and it is not good, especially if you are a smaller brand fighting against behemoths.

My take on specific link building tactics:

Digital PR is ok when:

• it is passive (I highly recommend Alex Horsman’s talk from Chiang Mai about resource-page-based link building without outreach or link begging)
• or it leads to cost-effective acquisition of links from reputable sources that couldn’t be acquired within a reasonable price range or are impossible to buy at all

Niche edits are ok when:

• they are contextual
• they come with a significant document update
• they come from a page with traffic and other solid quality metrics

Cheap guest posts may be ok if:

• they index
• they aren’t spammed (there is a limit to outgoing links)
• they are used as second-tier links (to reduce the risk of a bad link neighborhood for your website)

PBNs are usually ok if they meet the conditions above and use solid expired domains with maintained brand signals, which becomes harder and harder in the long term.

Sponsored articles are great if:

• they are published within real media, industry magazines, or blogs (otherwise it is guest posting on content sites or article submission on PBNs)

I really love the potential of sponsored articles, as they allow you to control all major factors: placement, link surroundings, context, anchors, longevity, and more.

Friendly reminder: the majority of these tactics violate Google’s guidelines, including most digital PR campaigns, so it is all at your own risk. If you decide to follow this path, at least try to deliver real value to users; it may protect you to some extent.

Sponsored articles platforms

If you want to buy sponsored articles to boost your marketing effort, look here: https://www.szymonslowik.com/buy-guest-post-links-why-how-and-where/#Link_building_platforms_marketplaces_to_buy_links_from

AI Overviews + sponsored articles

From my observations, high-quality, well-written sponsored articles published on reputable sources can boost your exposure in AI Overviews. Think about it as parasite SEO 2.0.

If you want to try this method, follow these steps:

1. Pick link placements (websites) already ranking in your topical domain.
2. Target long-tail, query fan-out-style topics.
3. Treat AI Overviews as parasite SEO placements.

I believe it makes sense to target mid-funnel terms with this tactic, because:

• more and more queries with mixed intent (informational-commercial) result in AI Overviews being displayed,
• AI Overviews kill CTR in classic rankings,
• it is relatively easy to get into AI Overview snippets by targeting query fan-out,
• you can gain exposure with both the sponsored article and the target page simultaneously and cover your competition.

How to position your brand within ChatGPT, Gemini, AI Mode, Perplexity, and others

To be served in AI results, you need to meet several conditions:

1. Retrieval: information about you should be easy to find (query augmentation or fan-out) and collect (avoid JS-heavy websites and focus on semantic clarity).
2. Consistency and co-occurrence: to be positioned alongside certain values, there has to be a clear message delivered consistently with your brand.
3. Trustworthiness: in the long run, it is not enough to simply say “this brand is the best for…”. LLMs need a reason to believe you. Craft messages that say “brand X is best for Y because Z,” where Z includes social proof, case studies, reviews, rankings, physical evidence, and other “reason to believe” marketing tactics.

So it is not only about mentions within some listicles. Obviously they work, but only because of points mentioned above. It is about retrieval, clarity, consistent positioning and trust.

Also remember that to be retrieved, you first need to be, and remain, indexed. You can use these tactics on PBNs or even forums, but in the long run they need to be maintained very well to stick. This is why it is worth considering sponsored articles.

During the talk, I cited our case study of our client, a dietary catering delivery service, for which we created a campaign that amplified the fact that they have outstanding online reviews. Those reviews themselves were not picked up by LLMs during the retrieval process. This is why we published dozens of articles on relevant topics that underlined their high ratings.

I checked while writing this article, and it ranks well, in a fresh browser without history or login, for various Polish and English prompts in AI Overviews, ChatGPT, and Perplexity.

AI visibility tracking & traffic analytics

I mentioned using a clean browser without history or login, which leads us to the next point: AI tracking. There is an ongoing discussion about proper ways to track AI visibility. Experts argue whether a UI or API perspective is better, and both sides have valid points. I am not here to take sides, because things are still evolving. I simply believe there is no straightforward way to track and measure AI exposure.

AI results in Google or ChatGPT are probabilistic and heavily personalized. Therefore, it is less about tracking and more about observation and pattern recognition.

Nevertheless, it is worth collecting data to compare results over time. I currently use two tools: Surfer and Ziptie.

Surfer is very easy to implement and has clear, simple dashboards. Additional value comes from many other core features that help with competition analysis and content design.

What I like about this tool is its ability to track all the sources that were pulled by Google or OpenAI to build AI search results. It is a great reference point for link prospecting.

I also recommend tracking AI traffic and its impact on conversions using Steve Toth’s Looker dashboard:

https://seonotebook.notion.site/The-Detailed-LLM-Traffic-Dashboard-with-a-Two-Click-Set-Up-1e38c368519180f297dec82a4f959a2d; it looks like that:

Third-party websites for brand discovery, recognition, and positioning

During the last part of my talk, I focused on showcasing real-life examples of sources from which LLMs retrieve information. It is a replicable approach. Just remember: it is not only about being mentioned or securing a link from a DR XYZ website. It is all about structured information about your brand that shapes understanding for LLMs, reduces uncertainty, and increases your chances of being present in Knowledge Graphs, LLM entity-entity systems, and the query fan-out process.

It is also smart, in my opinion, to reverse-engineer other brands’ identity-establishing sources (besides their own websites, obviously). You can ask Grok, Gemini, ChatGPT, Copilot, Perplexity, Meta AI, and others what they know about a specific brand and then check the sources.

You can even index some of those results, like Grok’s answer. I even got a dofollow backlink from this domain:

In my opinion, it is worth looking for reputable Web 2.0 platforms, content submission websites, video, podcast, image platforms, wikis, and others. You will find many examples in my Chiang Mai slides.

MY CHIANG MAI SEO CONFERENCE 2025 SLIDE DECK

FAQ: Off-site SEO, Google Systems & AI Search

Because they allow control over retrieval context, authority flow, entity framing, and additionally they are often sources used by AI Overview grounding.