Non-Commodity Content: How to Build a Brand That AI Search Can't Ignore
Your content is already inside the AI answers your customers read. It got summarized, blended with fifty other articles, and served up without your name anywhere near it. That is the deal most businesses have accidentally signed: you supply the raw material, the AI supplies the answer, and the customer never learns you exist. The only way out is to publish things that cannot be answered without naming you.
The real problem: AI answers are built from your work, and you get nothing back
Start with what the click data actually says, because it is worse than most owners realize. Pew Research Center analyzed the real browsing behavior of 900 U.S. adults in March 2025. When a Google search produced an AI summary, users clicked a traditional result only 8 percent of the time, compared to 15 percent when there was no summary. And the links inside the AI summary itself? Users clicked those on just 1 percent of the searches where a summary appeared.
Ahrefs ran the numbers from the website side and found the same story. Their April 2025 study of 300,000 keywords found that the presence of an AI Overview correlated with a 34.5 percent lower clickthrough rate for the top-ranking page. By their December 2025 update, that gap had widened to 58 percent. Similarweb data on news publishers shows zero-click searches for news climbing from 56 percent to 69 percent in the year after AI Overviews launched, and overall traffic to news sites dropping 26 percent over twelve months.
Here is what that means for your business. Ranking well no longer guarantees being seen. The AI layer sits between your content and your customer, and it answers the question using a synthesis of everything ever written on the topic. If your article is one of the fifty sources that got blended into that synthesis, you did the work and someone else, or no one, got the credit.
But the same data contains the opening. AI systems do cite sources. They name brands. They quote statistics and attribute them. The brands getting named are not the ones with the most blog posts. They are the ones whose content cannot be paraphrased without losing something. That is the difference between commodity content and non-commodity content, and it is the difference between being the source and being the raw material.
Why most businesses get this wrong: the commodity content trap
Most business content is written by reading the top ten results for a keyword and producing a slightly nicer version of the same information. For about fifteen years, that worked. Google needed pages to rank, and a well-organized page that covered the topic could earn its slot. So an entire industry trained itself to produce content that summarizes what is already known.
Large language models are summarization machines. Summarizing the consensus is the one job they do better, faster, and cheaper than any content team on earth. When ChatGPT or Google's AI Overview answers "what is email deliverability" or "how often should I service my HVAC system," it does not need your version of the consensus. It already has the consensus. Your beautifully formatted, professionally written, technically accurate article adds nothing the model could not generate itself, so there is no reason to cite you, mention you, or send anyone to you.
The scale of this problem predates AI. Ahrefs studied roughly 14 billion pages back in 2023 and found that 96.55 percent of them got zero traffic from Google. Most content was already invisible because most content was already interchangeable. AI did not create the commodity content problem. It just removed the last reason commodity content ever got clicks: someone had to visit a page to read the summary. Now nobody does.
Google has been telegraphing this direction for years. The company holds a patent, US 11,157,557, on what it calls an "information gain score," a way of scoring a document by how much new information it adds beyond what the user has already seen. Read that again from a business owner's perspective. The system is explicitly designed to ask: does this page tell me anything the other pages did not? For most business blogs, the honest answer is no.
The mistake, then, is not laziness or bad writing. It is a strategy error. Businesses keep investing in content whose entire value proposition, organized information, has been automated. They are competing with the machine at the machine's best game.
What the data actually shows about who gets cited
This is not guesswork. There is now a body of research on what AI systems actually choose to cite and name, and the findings are remarkably consistent.
The first large academic study of the question came from researchers at Princeton, Georgia Tech, IIT Delhi, and the Allen Institute for AI. Their paper, "GEO: Generative Engine Optimization," presented at KDD 2024, tested nine content tactics across 10,000 queries. The tactics that boosted a page's visibility in AI-generated answers by 30 to 40 percent were adding statistics, adding quotations, and citing sources. In plain terms: concrete, attributable, verifiable material gets pulled into answers. Generic prose gets absorbed and forgotten.
Then look at what predicts whether a brand shows up in AI answers at all. Ahrefs analyzed 75,000 brands in 2025 and measured which factors correlated with visibility in Google's AI Overviews. The strongest factor was brand web mentions, with a correlation of 0.664. Backlinks, the currency of traditional SEO, came in at just 0.218. Mentions beat links roughly three to one. Brands in the top quartile for web mentions earned up to ten times more AI Overview mentions than the next quartile. The model recommends brands it has seen discussed, repeatedly, across the open web.
Semrush's AI visibility research adds a sharper edge. Their study of how ChatGPT, Gemini, and Google's AI Mode handle brands found that ChatGPT cites sources in 87 percent of responses but mentions brands by name in only about 21 percent. Being a citation in a footnote is not the same as being the recommendation in the answer. Their data also showed that community and reference sources, Reddit and Wikipedia above all, consistently outrank corporate websites as citation sources across industries. AI systems trust what other people say about you more than what you say about yourself.
Mike King of iPullRank, named Search Engine Land's Search Marketer of the Year in 2025, has been blunt about what this moment demands from people who make content for the web:
"Do you all really want to stay the janitors of the web? This is our moment to really stand up and be something different."
Mike King, iPullRank, SEO Week 2025
Put the research together and the pattern is hard to miss. AI systems cite original statistics. They name brands that the wider web already talks about. They pull from sources that say something specific and attributable. Every one of those is something a business can deliberately build.
How to become the source instead of the raw material
None of this requires a technical team. It requires deciding to publish things only you can publish. Here are five moves, in priority order, that you can do yourself or hand to whoever runs your marketing.
1. Publish one original data asset per quarter
You are sitting on data nobody else has. Your sales records, your customer support tickets, your project outcomes, your pricing history. A roofing company that publishes "what 340 roof inspections in Ohio actually found, and what each repair cost" has created something no AI can generate and no competitor can copy. If you have no internal data, run a survey. Even 100 responses from your customer base, written up honestly with the numbers shown, gives AI systems a statistic they can only cite with your name attached. This is the single highest-leverage move on this list because the Princeton research showed statistics are precisely what generative engines reach for.
2. Name your method
If you have been doing this work for years, you have a process. Right now it lives in your head as "how we do things." Give it a name, define its steps, and publish it. A named framework is uniquely citable: an AI cannot explain "the 3-2-1 Backup Rule" or "Jobs to Be Done" without attribution, because the name is the content. The same applies to a named audit process, a named pricing model, or a named diagnostic checklist. Unnamed expertise gets paraphrased. Named expertise gets referenced.
3. Write from the first person, with receipts
Replace "businesses often find that" with "in March we tried this with a client and it failed, here is the invoice-level breakdown of why." First-hand experience, specific numbers, specific failures, specific timelines, is the one category of content a language model cannot synthesize, because it never happened to the model. It also maps directly onto the information gain concept Google patented: your experience is, by definition, information that exists nowhere else in the index.
4. Get talked about where AI systems listen
Remember the Ahrefs finding: brand mentions across the web predict AI visibility three times better than backlinks. So spend the effort you used to spend on link building on getting mentioned instead. Pitch yourself to industry podcasts. Answer journalist requests. Contribute named, quotable commentary to publications in your space. Show up usefully in the Reddit and forum threads where your customers actually ask questions, as yourself, without spamming. Ahrefs also found YouTube mentions showed the strongest correlation of all with AI visibility, so if you can put your data and frameworks on video, do it.
5. Make every key claim quotable
Audit your most important pages and ask one question of each: is there a single sentence here that someone could quote, with a number in it and our name near it? "Most websites lose traffic to AI" is not quotable. "Across our last 40 client audits, sites with original data assets earned 6 times more AI citations" is. Structure matters too: put the direct answer in the first sentence under each heading, keep statistics in plain text rather than buried in images, and attribute your own data to your own brand by name in the sentence itself.
What to measure, and when to expect results
The first thing to do is stop grading this work with the old report card. Total organic traffic is now a misleading number, because AI answers are suppressing clicks across the board even for content that is performing well. Rand Fishkin of SparkToro has been arguing this for years, and his framing is the right one: in a zero-click world, traffic is a terrible goal. Influence is the goal. Impressions in Search Console rising while clicks fall is not necessarily failure; it can mean you are being surfaced inside answers.
Here is what to track instead. First, AI mentions and citations: once a month, ask ChatGPT, Perplexity, Gemini, and Google's AI Mode the ten questions a real customer would ask before hiring someone like you, and record whether you are named, cited, or absent. Do it in a clean browser session so your own history does not skew it. Second, branded search volume: the number of people searching your name is the cleanest proxy for whether the mentions strategy is working, and Ahrefs found it among the top correlates of AI visibility. Third, referral traffic from AI assistants: ChatGPT, Perplexity, and Copilot referrals show up in your analytics, and while the volumes are small, those visitors arrive pre-sold. Fourth, the numbers that always mattered: inquiries, booked calls, and revenue from organic channels.
On timelines, be realistic. An original data asset can earn its first AI citations within 4 to 8 weeks of publication if it gets picked up and discussed. Brand mention building is slower; expect 3 to 6 months before you see movement on niche queries and 6 to 12 months for the broader, more competitive questions in your space. The vanity traps to avoid: celebrating raw traffic, counting published posts as output, and chasing rankings for keywords whose results are now fully answered above the fold. A site that publishes 12 cited, mentioned, quoted assets per year will beat a site that publishes 200 commodity posts, and it will cost less.
Frequently Asked Questions
What is non-commodity content?
Non-commodity content is content that cannot be reproduced by summarizing what already exists online. It includes original data and research, named frameworks and methodologies, first-hand experience with specific numbers, and documented expert positions. Commodity content restates the consensus, which AI systems can generate themselves, so they have no reason to cite it. Non-commodity content forces attribution because the information exists nowhere else. If a competitor or an AI could produce your article without ever having done your work, it is commodity content.
How do I get my business cited by ChatGPT and Google AI Overviews?
Focus on two things the research consistently supports: publish citable material and build brand mentions. The Princeton GEO study found that adding statistics, quotations, and cited sources boosted visibility in AI answers by 30 to 40 percent, and an Ahrefs study of 75,000 brands found web mentions predicted AI visibility about three times better than backlinks. Practically, that means publishing original data with your name attached, getting discussed on podcasts, YouTube, and community sites like Reddit, and making sure your key pages answer questions directly in the first sentence. Then test monthly by asking the AI tools real customer questions and recording whether you appear.
How long does it take to see results from a non-commodity content strategy?
A strong original data asset can start earning AI citations within 4 to 8 weeks if it gets picked up and shared. Building the brand mentions that drive consistent AI visibility typically takes 3 to 6 months for niche queries and 6 to 12 months for competitive ones. Measure progress with monthly AI mention checks, branded search volume, and AI referral traffic rather than total organic traffic, which is falling industry-wide regardless of content quality. The work compounds: every mention, citation, and named framework keeps working long after publication.
The web is splitting into two kinds of businesses: the ones whose work feeds the answer, and the ones whose name is in it. Commodity content was a fine strategy when humans did the reading, and that era is over. Publish the data only you have, name the method only you use, and get talked about in the places the machines are listening. Be the source, or be the raw material. Those are the options now.
