Research says AIO/LLMO method is useless. Should I do it?
Yes, there’s currently no substantial data to support GEO strategies. The reaons is that not enough time has passed to conduct large-scale research and analysis.
SEO and marketing professionals often compare GEO to SEO. In SEO, studies are present for multiple variables, whether it’s about meta descriptions, titles, schema, content, or even hyperlink colors. Neil Patel, Brian Dean, Rand Fishkin, Tim Soulo, ViperChill, and hundreds of other marketers have crunched numbers to find correlations that could give us insight into how SEO works. Many Chrome extensions, free tools, surveys, add-ons, and scripts were made and are running on the web, for free, with only one purpose. To track the user’s behaviour on the web and in the search. Expert data scientists have leveraged this information to determine what works in search.
All this SEO and marketing research didn’t happen overnight; it took decades and people to build the knowledge base. We should apply the same to GEO, AIO, and LLMO. It will take time to identify the right controls, reactions, units, and dimensions to measure and test, and to establish proper benchmarks. What we are seeing now is early-stage research, which is mostly based on limited datasets and done by biased researchers.
So, should I start implementing that GEO/LLMO/AIO strategy?
Research-backed strategies are not always on time, as collecting relevant data and developing variables and controls requires thorough investigation. The lack of supportive research shouldn’t be your major concern when experimenting with a relatively new marketing channel. You are the first one there, and you will have a first-mover advantage if you move fast.
Whether you do it or not, it depends on the efforts and resources you could put aside to implement that GEO/LLMO optimization. Also, do a threat check. Most GEO methods do not hurt SEO, but still, use your mind. Things change case by case. So there is no general advice. But do not limit yourself because of a lack of research evidence.
If you are a startup or a new business, SEO could be challenging, but AIO/LLMO would make a dent much faster.
What if the research shows that this LLMO strategy does not work?
Many SEO and marketing companies are conducting research on a specific LLMO strategy. The findings usually show that the said method is not helpful. If you read their research methodology, you will find how shallow it is.
- Research conducted on a sample of 100 people does not give enough data to generalize to the entire market
- Data from GA4 of one or 100 websites does not reflect the industry
- Replication has not been tested at all
- Survey filled by 50 people at the event cannot be called respected primary data
- Whether a technical optimization helps or not cannot be decided by the crawl popularity of those pages
- Comparing the users from ChatGPT and Google is not fair, as Google’s purpose is to send users to websites, and ChatGPT’s purpose is to provide information
Schema, text chunks, llm txt files, prompt injections, question pages, llm feeding, and automated conversations — for many, they are working, but general preliminary research shows they made no impact.
Marketing tools and SEO agencies are only doing it for the sake of doing it, as journalists need statistics on this subject. Marketing tools and SEO agencies need brand mentions from journalists on traditional media websites. MarTech knows that the standards of journalism are so poor that they will put any numbers into the article without reading the methodology. So MarTech does the research without putting in any effort, and everyone runs with it.
So even if the research shows no, you should not stop yourself from doing it, because research happening on GEO/AIO/LLMO and AI marketing is not good enough to take seriously.
