LLMO is the practice of getting your company name and pages mentioned in the answers of conversational AI such as ChatGPT and Gemini. Whereas SEO aims to rank high in the list of search results, LLMO aims to get your name mentioned within the AI's answer. What small and medium-sized businesses can start today is aligning their company facts and writing in a way that answers questions.
LLMO is about getting your name mentioned in conversational AI answers
LLMO stands for Large Language Model Optimization. It refers to efforts to get text-generating AI such as ChatGPT and Gemini to use your pages as a source and name your company and services when answering questions.
Imagine an AI asked, "Tell me about website production companies with clear quotes," listing and explaining several company names. Whether your company's name is among them is the result of LLMO. Unlike a list of search results, there is no ranking table; the difference is simply whether you are named or not.
Why you need to think about LLMO now
Because the starting point for research has expanded beyond the search box. More people type a question in sentences and receive an answer, and companies whose names appear in the answer are remembered by readers as candidates for comparison.
That said, LLMO does not replace SEO. Google Search Central's official documentation explains that there are no special technical requirements to appear in Google's AI features and that the basics of traditional search apply. Pages on the web are also the material when AI builds an answer, so the work of making pages easy to read as material overlaps greatly with SEO.
Differences between SEO and LLMO
| Item | SEO | LLMO |
|---|---|---|
| Where it appears | The list of search results | Inside a conversational AI answer |
| Form of result | Your ranking rises | You're named or cited as a source |
| Reader's action | Choose a page from the list | Read the answer and, if needed, go to the source |
| What works | Content that matches search terms, internal links, page speed | Consistent facts, conclusion first, organized questions and answers |
| How to check | Measure search ranking | Ask AI actual questions and check |
What they share is writing original content that is useful to readers. The difference is that "ease of extraction" strongly affects AI answers. A page with the answer right after the heading is more likely to be chosen as material for an answer than a page where the answer is buried deep in long text.
6 tasks small and medium-sized businesses can start today
- Write your basic company information on a single sheet: company name, location, phone number, year of founding, business description and representative's name
- Align how your company is written on your website, Google Business Profile and social media with what you wrote down
- On your main service pages, add one sentence stating the conclusion right after the heading
- Pick 10 questions you have actually received from customers and put them in your website's FAQ
- Write prices, timelines and service coverage with numbers and units. For items where you have no number to write, leave them blank rather than filling them in
- Before and after the work, enter questions about your company into ChatGPT and Gemini, save the answers and compare them
These six cost almost nothing. Steps 1 and 2 in particular prevent AI from mistaking the same company for a different one, and are where results come most easily.
Add structured data after organizing the content
Structured data is a way of adding a page's content in a form machines can read. It may help with LLMO too, but what you should get in order first is the main text. After organizing the text and making it consistent with the information displayed on screen, add the type that fits the content, such as Organization for company information and FAQPage for FAQs. Google's guidelines do not allow writing content in structured data that is not shown on screen.
Common LLMO mistakes
The company name or address is written differently in different places
If the position of "Ltd.," whether the building name is included, or the way the phone number is separated differ between your website and social media, AI may treat them as different companies. Settle on a single list of facts and use the same wording everywhere.
Starting with a company introduction instead of answering the question
If you answer "How much does it cost?" by starting with the company's history and philosophy, the answer becomes hard to find. Make the question a heading and answer it in the sentence right after.
Relying on a proposal that guarantees rankings or inclusion
AI answers are decided by the company that runs the AI, and an outside provider cannot guarantee inclusion. If someone proposes that you will "definitely be shown," ask for the basis.
Building it once and never updating
If outdated prices or business hours remain, AI may use the old information in its answers. Get in the habit of fixing your website and outside listings at the same time whenever something changes.
Check results by entering the same questions into AI every month
LLMO has no fixed measurement screen like search ranking. So decide on about five questions customers would actually ask, enter the same wording into ChatGPT and Gemini every month, and save the answers. You look at three points: whether your company name appeared, whether your page was listed as a source, and whether the company information given is correct. If incorrect information appears, find the page it came from and fix it. Continuing with the same wording lets you compare before and after your efforts.
How to judge
If the following three apply, you can start on your own.
- You can edit the text on your website yourself
- You can find time to put your basic company information on a single sheet
- You can try questions to AI yourself once a month
If you need structured data implemented or a review of your entire existing site, it's faster to consult a production company.
Dividing roles inside your company
LLMO lasts longer when you divide the roles rather than shouldering it alone. Even in a small company, deciding on the following three roles makes it easier.
The first is the fact keeper. This person holds the correct company name, address, phone number, prices and business hours on a single sheet and tells everyone when something changes. Someone in general affairs or accounting suits this role.
The second is the question collector. This person jots down daily the questions customers ask by phone or email. Someone in sales or reception suits this role. Once a month, pick 10 of the collected questions and add them to your website's FAQ.
The third is the checker. Once a month, this person enters the chosen questions into AI and saves the answers. They look at whether your name appeared and whether the information is correct, and share the results.
One person can take on all three roles. What matters is deciding who does it and when. If the person and the date aren't set, things stop after the first round.
Also, LLMO work is never finished in one go. Each time your services or prices change, update the fact list and your website. The effort of updating is small, so build it in as a regular monthly task.
How to run the first month
Run the first month of LLMO like this. In week 1, create the fact list and fix how your company is written on your website and social media. In week 2, decide on five questions to enter into AI and save the current answers. In week 3, add a one-sentence conclusion at the top of your service and pricing pages. In week 4, pick 10 questions you've received from customers and put them in your FAQ. Entering the same questions after a month and comparing shows what changed.
Don't stop after one month; from the second month on, make entering the questions and updating the information a once-a-month routine. Even if the change is small, keeping records gives you material to judge the results later.
FAQ
Yes. AI answers are built using pages on the web as material. The work of building pages that can be found through search with SEO overlaps with the work of making pages easy to extract with LLMO. You don't have to choose just one.
We can't give a set number of days. It depends on when AI refers to the information and the state of your pages. Keep asking AI questions once a month and record how the answers change.
Yes. AI looks not at a company's size but at whether the page has the information that answers the question. It's quite conceivable that a small shop's page that describes its services and service area concretely would be picked up in an answer.
It has been talked about as a guide file for AI, but Google has indicated that no special file is needed to appear in its AI features. Prioritize organizing your main text first.
The main work is diagnosing the current situation, unifying company information, revising copy, implementing structured data and checking results. We've summarized what to confirm before hiring in a separate article.
You can consult us about AI search optimization for your current website
At Merci Ltd., we build structure that is strong for SEO, GEO and LLMO into new websites as standard. We also take consultations on diagnosis and improvements for websites that are already live.
We've put how to proceed with AI search optimization on a single page.
