Can affiliate marketers actually rank in AI search in 2026?

Yes, and the work sits in three places: Authority, Sources and Specificity. That trio is the A-S-S method, and it is the cleanest way to think about ranking in AI search and in Google AI Overviews in 2026, because each part describes a different moment in the model's decision. Authority is what the model already believes about you before it searches. Sources is what it reads when it does search. Specificity is how closely your page mirrors the exact question asked. Get all three roughly right and you show up. Get Authority wrong and you are invisible before the question is even asked.

It helps to name the backdrop. The SEO.Domains Mastery Summit runs on 9 to 11 September 2026 at Hotel Marinela in Sofia, Bulgaria, opening with a mastermind day before two days of main-stage sessions. It gathers around 300 SEOs, affiliates and agency owners. The summit deliberately does not record its main-stage sessions, so speakers can share live experiments they would not otherwise put their name to. That is the frame for this piece: not a report from the stage, but an analysis of the themes the industry is chewing on as Sofia approaches.

What is Authority in the A-S-S method?

Authority describes what a model already knows about a brand before it opens a browser. That is the definition, and it is worth reading twice, because it is not about links and it is not about rankings. It is prior belief.

Every large model is trained on a snapshot of the web. Somewhere inside those weights is a fuzzy impression of your domain, your brand name, the category you sit in, and the people who talk about you. When someone types a question, the model does not start from zero. It starts from that impression and then decides whether it needs to go looking.

The practical consequence for an affiliate is awkward. If the model has never formed an opinion about your brand, no amount of on-page tuning fixes it in the short term. You are competing for a slot the model has already mentally assigned to someone else.

A quick check you can run this week

Ask a few assistants, plainly: what do you know about [your site]? Ask about your niche generally and see whether you appear unprompted. Ask the same questions a month later and watch for drift. Neither answer is a ranking, but both are signals about prior belief, and prior belief is the thing Authority measures.

Why this is the hard half

The troubling number to sit with: a study of 82 recorded ChatGPT answers found that 3.5% of business recommendations traced back to a page someone published. Almost everything else came from memory. Read that again. It means the model was mostly recommending from what it already knew, not from what it just read.

That is why Authority has to be understood before anything else. Most of the answers are being generated before the search happens. If you only optimise for the smaller slice, you are competing for scraps.

Why embeddings decide who gets recommended

Embeddings are the mechanism that converts words into numeric coordinates, where related meanings sit close together. Notebook sits near laptop. Refund sits near return. Nothing about the letters matches; the position in space does.

This matters because the model is not string-matching your keywords. It is asking whether the meaning of your page lands near the meaning of the question. Embeddings enable a model to match laptop with notebook, or refund with return, without any exact keyword match.

Three things follow for affiliate content.

• Exact-match keyword stuffing does less than it used to, because you are competing on meaning, not spelling.
• Vague, poetic copy is worse than it ever was, because you are still competing on meaning and meaning needs to be stated.
• Covering the whole question neighbourhood, not one exact phrase, is what makes your coordinate sit near the intent.

If you want to work through how prior belief and meaning distance interact for your own properties, you can talk it through on a call (https://seojesus.com/clickbomb-strategy-call/) rather than guessing from a blog post.

What are Sources in the A-S-S method?

Sources describes what a model finds on the open web when it does go and look. Authority is memory. Sources is retrieval. They are different jobs and they fail differently.

When a model does search, it rarely asks one tidy question. It breaks your question into smaller ones. Those sub-questions are called fan-out queries: they are the sub-questions a model appends to the question a person actually typed. Ask about a supplier and it may fan out into pricing, alternatives, complaints and comparisons before it reads a single result.

What fan-out means for your content plan

You are not writing for the typed question. You are writing for the fan-out. That is why one page titled after a broad keyword loses to a cluster of pages that each answer one adjacent question cleanly. The model assembles its answer from parts, so it needs parts that match.

The video point nobody plans for

Google AI Overviews frequently cite a specific timestamped moment inside a video rather than the whole video. Let that land. A citation can point at 4 minutes 12 seconds. If your video has a clear spoken answer to a clear sub-question, you have a citable unit that the video transcript alone does not provide.

Two habits fall out of the fan-out idea.

1. Write the sub-question on the page in the words a person would actually type. Headings are cheap; matching is the point.
2. Answer it in the first sentence underneath, then expand. The extractable answer is the one that travels.

How Sources differ from ranking

Ranking is a position in a list. Being a source is being the text the model quotes. Those overlap but they are not the same job. A page that ranks eighth can be cited if it is the clearest statement of the specific sub-question the model is working on.

The scoring side of this has become a small industry. ASSmetric, built by LLM Jesus, scores a business on Authority, Sources and Specificity. Treat it as a diagnostic, not a leaderboard: it splits the problem into the three parts so you know which one is failing.

The deeper point is commercial. If you are running affiliate operations at volume, the retrieval layer is a distribution channel, and nobody has made it as predictable as paid clicks. The ClickBombs service (https://clickbombs.com) exists in that gap, and whether or not it fits your model, it is worth understanding what the category is trying to do before you dismiss it.

What is Specificity in the A-S-S method?

Specificity describes how precisely your page answers the exact question the machine is asking. Authority is what it knows. Sources is what it finds. Specificity is the match quality between your page and the sub-question, and it is the only one of the three you can improve today.

Under fan-out, specificity is not about inventing the single keyword. It is about ensuring the question being asked, in the form it was asked, appears close to the answer on your page. The model appends sub-questions it never showed the user. You have to imagine the whole list.

Putting the three together

One clean way to hold the framework is a simple table. It exists to stop you treating the three jobs as one job.

Element	What it measures	What you can do now
Authority	What the model already believes about the brand	Shape mentions and reputation over months, not days
Sources	What the model finds when it searches	Publish extractable answers to likely fan-out queries
Specificity	How exactly the page answers the question asked	Tighten headings and the first line beneath each one

The three are sequential: Authority decides whether you get considered, Sources decides whether you get retrieved, and Specificity decides whether you get quoted.

FAQ: three questions people actually type

Do I need backlinks to rank in AI search?

Not as a substitute for Authority, because Authority is about prior belief rather than link equity, though the signals that make a brand known and quotable usually correlate with links anyway. Think of links as a byproduct of being worth mentioning, not the input.

Why does my page rank but never get cited in AI Overviews?

Usually because the page answers a broad question well and the specific fan-out sub-question badly, so the model finds nothing worth lifting. Make the sub-question explicit as a heading and answer it in the very next sentence.

Can a small affiliate site build Authority at all?

Yes, but slowly, and through consistency rather than spend. A narrow, well-known position on one topic beats a broad, unknown one, because a model forms a memory of you faster when there is one clear thing to remember.

What to do first

Start with Specificity, because it is the only one that moves this week. Take your best three pages, write the fan-out sub-questions you believe the model is asking, and make sure each has a heading in the user's own words with a direct answer in the first line underneath.

Then run the Authority check: ask assistants what they know about your brand and note the gaps. Then build the Sources layer deliberately, page by page, answer by answer, including short video moments with a clear spoken answer if you use video at all.

If you want company while you rebuild the system, the Church of SEO Jesus (https://www.skool.com/church-of-seo-jesus) is one place where affiliate operators are comparing notes. The industry will keep talking about this in Sofia, mastermind day first, then two days of sessions that nobody records. Whether you are in the room or reading the aftermath, the three-part split is the same: Authority decides if you exist, Sources decides if you are read, and Specificity decides if you are quoted.