You're Still Fighting for Rankings. Others Are Already Fighting to Be Cited by AI

A while back, a friend of mine, a dozen-plus years in export trade, asked me out for tea.

He said: something's wrong. The site's Google rankings are still there, the reports still look great, but inquiries have fallen off a cliff. The money keeps going out, and the anxiety keeps going up.

I asked him: search one of your core product keywords and see what comes up.

He pulled out his phone on the spot. After the search, he froze.

At the very top, there were no ten blue links. There was an answer generated directly by AI. What the product is, which brand is good, roughly what it costs — the user's question was fully answered on the first screen.

His website link sat below. Nobody scrolled down.

And this is no longer just his problem.

A Twenty-Year Understanding, Dissolved

For the past two decades, businesses and search engines operated on an unspoken contract.

You worked the keywords, Google gave you traffic. You optimized your pages, Google gave you rankings. Everyone got what they wanted, and the partnership ran smoothly.

That contract has now been dissolved.

Google's AI Overviews, ChatGPT, Perplexity, Bing Chat — these generative engines are replacing "a list of links" with "a single answer." Some say traditional search is dead. That's half exaggeration — but the other half is true: what you're fighting for has changed. It used to be clicks. Now it's citations.

What does fighting for citations mean?

It means when AI answers a user's question, the company it casually mentions is you.

Get mentioned, and you're part of the answer. Don't get mentioned, and you never even qualified for the game. The user will simply never know you exist.

From "Strings" to "Entities"

So how do you get AI to mention you?

To answer that, you first have to understand one thing: AI search and traditional search run on two entirely different underlying logics.

A traditional search engine is, at its core, a retrieval system. A user searches "best enterprise management software," and it goes looking for pages containing that string of words, then ranks whoever has more backlinks higher. It matches strings.

Generative engines are different. They're understanding systems. They don't look at words; they look at entities.

What's an entity?

It's a "thing" as the machine sees it. It knows CRM is a category of software, knows Salesforce is a brand, knows "pricing" is an attribute of software. These things and their relationships form a web, and AI uses that web to understand the world.

So getting your brand treated by AI as "the standard answer to a certain question" doesn't come from repeating a keyword five times in an article. It comes from structuring your data so machines can easily read your authority. This approach has a name: GEO, Generative Engine Optimization.

Put bluntly: you used to serve the library's card catalog. Now you serve a thinking brain.

Serving a brain takes three things.

First, build a knowledge graph. Use schema markup to write your company, your founder, and your products plainly into your website's code. Don't make AI guess who you are — tell it outright.

Second, do entity salience. In your content, core concepts must be clearly defined, backed by exclusive data no one else has. AI loves "incremental information" — Google even runs an incentive mechanism that specifically rewards content adding something new to the internet, rather than copying the old stuff yet again.

Third, accumulate brand authority. AI looks at "real-world signals": whether credible media cite you, whether you have expert content under a byline, whether your information is consistent across platforms.

Skip all three? Then AI simply routes around you. Your keyword rankings may survive, but by the time a user could ever see your link, their question has already been answered by AI.

Traffic: zero.

The Three Tiers of Agencies

Hearing this, you may want to ask: so what is my outsourced marketing agency doing about all this?

Now that's the right question. This is exactly where it hurts most right now.

Someone has mapped this industry, sorting digital marketing agencies into three tiers. I find this framing remarkably sharp, so let me share it.

Tier one: keyword merchants. Roughly 75% of the market. Five years ago they were good hands; today their playbook is selling backlinks and mass-producing blog posts, producing reports keyed to keyword density and domain authority. In their eyes, Google is still that library card catalog. In the eyes of 2026's AI algorithms, the "thin content" this playbook produces is not just useless — it gets penalized.

Tier two: AI spammers. About 20%. The crowd that pivoted after ChatGPT took off. They saw the old road dying, so they pivoted to speed: mass-generating content with AI — hundreds of articles, prices slashed to the bone. Sounds tempting.

It's a trap.

They're running the old strategy with new tools. The content reads fluently, but the informational value-add is zero — people in the industry call it "gray sludge." Google is purging this kind of content at scale. The more they pump out, the faster they die.

Tier three: generative engine architects. Under 5%. They don't write content; they design data. They know AI must first "understand" you before it can cite you, so they build semantic networks and knowledge graphs, aiming for every single piece of content to add one new data point to the internet.

Put the three tiers side by side and a brutal fact emerges: the middle ground is vanishing. Your agency is either moving forward, or dragging you backward.

Five Questions to Test Them on the Spot

So how do you tell which tier yours is? Don't listen to their salespeople — salespeople can all talk. Just ask five technical questions, directly.

Question one: can you show me our knowledge graph strategy?

If they can't answer, or start chatting about social media management, walk away. The right answer should be a visual graph or a code file showing how they use Organization schema and the SameAs attribute to link your company and founder together.

Question two: what's the information gain on our flagship content?

If they say "we look at word count," walk away. A decent answer is: analyze the existing search results and make sure at least twenty percent of the content is exclusive data or original insight. AI favors what's new; parrots need not apply.

Question three: zero-click searches — how do we deal with them?

If they say "we're after clicks, we don't care about that," walk away. The reality to take seriously now: some share of traffic simply isn't going to click anymore, and what you're competing for is being named inside the AI's answer.

Question four: do you write nested schema markup?

If they say "just install a plugin," walk away. Enterprise-grade GEO requires hand-written JSON-LD that embeds your customer reviews into your product entities. Plugins can't do that.

Question five: since Google SGE launched, what have you changed in your playbook?

If they say "SEO is always changing, we stick to the fundamentals," walk away. Companies actually doing the work will tell you: budgets have shifted from mass-produced blogs to high-authority digital PR and data structuring.

Ask all five questions, and you'll know exactly what you're dealing with. No need for politeness. Politeness gets paid for out of your market share.

The Window Is Closing

One final thought.

The shift to AI-first search isn't something about to happen. It has already happened.

Every day your website sits still on old code and an old content strategy, your digital trust score keeps dropping. Meanwhile your competitors — the companies that did GEO early — are being cited by AI again and again, becoming the "industry answer" in users' minds.

Every citation adds a point of authority. Add enough points, and eventually, when AI talks about your industry, it opens with them.

History is being rewritten. And you're not among the ones writing it — that is perhaps the one position a business owner should never accept.

So stop paying monthly fees for metrics that no longer matter. Go ask those five questions. If the answers come out stumbling, that's your signal to change partners.

Here's to becoming the company the AI names.