Everyone pretends to be an expert in AI Search, but in reality, it’s a very recent topic which is constantly changing.
However, some truths exist.
In this article, Antonis Dimitriou, Organic Growth Strategist at Omniscient Digital shares his tiered list, based on his experience working with SaaS and Enterprise clients. These beliefs are supported by various studies across the web.
The article was edited by Nick Malekos, the founder & editor of Marketing Experts Hub.
The Tiered List
Tactics with strong evidence and effectiveness towards ranking of AI Search, and are the most important factors LLMs will use to surface citations and mentions.
- Building brand mentions with co-occurring terms
- Targeting queries plus modifiers
- Consistent brand positioning
Tactics which strongly impact AI Search results.
Often used for citations and influencing the AI Chatbot’s answer to a query, but not always effective.
Impactful up to a point, for specific cases.
Possibly impactful, there is evidence crawlers related to LLMs visit the pages, and sometimes mention them, but not as impactful.
Often recommended by agencies, but with little evidence of impacting AI Search.
S Tier
These tactics teach models what your brand is, in relation to what, for whom. This is what the systems reach for when they decide who to cite and who to show up in responses.
Building brand mentions with co-occurring terms
This is the single highest-leverage thing you can do, and most teams do a diluted version of it.
Models learn associations from the text around your brand name, not from the name in isolation. If your brand keeps appearing beside “SOC analyst training” or “sales engagement platform” across enough independent sources, that association becomes stronger, and LLMs learn to associate it with it.
Pick the three to five terms you need your brand welded to. Then create mentions which carry those terms in the same sentence or paragraph: podcast show notes, roundup contributions, expert quotes, partner pages, community answers, comparison posts on third-party sites.
Targeting queries plus modifiers
Buyers do not ask LLMs one-word questions. They ask long-tail ones. For example “Sales engagement tool with HubSpot integration and SSO for enterprise.”
Build your target list from modifiers, not keywords. Pull them from your own sales calls, your G2 category filters, your competitor comparison pages, and your Search Console long tail. Then create pages that answer the head term plus the modifier explicitly and completely, in the page’s own words, high on the page.
Consistent brand positioning
This a company-wide problem and usually sits under PMMs or Brand. Consistency of how the brand is represented creates stronger associations and assists retrieval for niche cases.
Vague and varied responses create confusion and your brand will get replaced with a competitor who was clearer.
Write one positioning line. Use the same text everywhere: site, review profiles, LinkedIn company page, partner directories, press boilerplate, podcast intros, guest bios.
Enforce consistency on a company-level.
A Tier
These strongly impact AI search results, and the evidence behind them is solid. They sit one step down because they work by improving the conditions for visibility rather than by directly teaching a model what you are.
Link building (Digital PR)
Links got rebranded to Digital PR and citations. Backlinks still matter, even if they are unglamorous by building the authority and connections required for discovery.
The essentially improve the conditions and chances, but do not directly impact HOW you are being mentioned.
Targeting fan-out queries
AI Mode and similar systems decompose a query into sub-queries, run searches based on them, and synthesise the results. Those sub-queries are called fan-out queries.
Fan-out tactics target the use-cases, niches, and different ways people may search, and are informed by the memory of the chat for you. You can’t easily “know” what fan-out queries are, but much research shows the structure they take and can be replicated when doing SEO research.
Homepage messaging
Your homepage is the highest-authority page telling models what you are. Treat it as a primary source document, not a conversion asset with a hero image and three adjectives.
If your H1 is a slogan and your subhead is a benefit statement, there is nothing concrete to extract. If it names the category, the buyer, the core capability and the differentiator in plain text, then LLMs understand what you actually do.
LLMs prefer usability over brand statements.
B Tier
Both of these show up in citations regularly and both may decide what content is being used in the decision-making process.
Reddit mentions
Reddit is heavily cited by several AI search systems.
Active participation is the best way to get Reddit mentions, while many SEOs tend to try manipulative and automated tactics.
Use it with caution.
YouTube videos
Very good for Google’s AI search surfaces.
Video assets get surfaced in AI Overviews and adjacent Google experiences, so if Google is your primary target, then it’s own video platform is the right source to train.
C Tier
Impactful up to a point, but may not work for all kinds of businesses and niches. Use when needed.
Schema
Structured data helps Google understand entities and can support your presence in Google’s AI surfaces. Product, Organization and FAQ markup are the best ones to use overall, and they help to get various rich results on Google, they may as well help with AI Overviews.
What it does not do is affect visibility in non-Google LLMs like Claude, ChatGPT, Perplexity etc.
D Tier
Possibly impactful, there is evidence crawlers related to LLMs visit the pages, and sometimes mention them, but not as impactful.
Creating an AI info page
There is something here, but pretty uncertain. The problem is that it’s probably not needed, but crawlers still visit the page with little evidence it works.
Editor’s note: I strongly believe that a well written Home, About, and Contact pages are more than enough for the purpose, an AI info page may not be a complete waste of time but it probably does little more than a blog post or a product page.
F Tier
Often recommended by agencies, but with little evidence of impacting AI Search.
llms.txt
Many SEOs, consultants, and agencies implement llms.txt, but as proven by cats.txt, a fun experiment by Mark Williams-Cook, even an irrelevant page will be crawled and cited.
No major AI search system has committed to consuming it as a ranking or retrieval input. It has not been adopted the way robots.txt or sitemaps were, and publishing one has no demonstrated effect on whether you appear in AI answers. It is a proposal with strong marketing behind it, and it survives mostly because it is easy to sell and easy to tick off.
How to sequence this
Tiers S to A are actions which must be in everyone’s to-do list, while actions B and C depend per case and resources.
The rest are pretty much optional.
But, each tactic can support a wholistic SEO / GEO / AEO strategy to rank for search and various AIs.
Nick Malekos is the Head of Growth & Demand Generation at Cyberbit, with a background in SEO, Content Marketing, and Performance. He is specializing in helping SaaS startups and scale-ups grow.
Antonis Dimitriou is an SEO and organic growth strategist specializing in B2B. He is currently the Organic Growth Analyst @ Omniscient Digital, where he is helping B2B and SaaS companies worldwide grow through organic search.
