Large Language Model Optimization

What is LLMO?

Large Language Model Optimization (LLMO) is the process of optimizing content to be found and displayed in LLMs like ChatGPT and Google’s Gemini. The key benefit for this optimization is getting an organization or brand associated with a query that a user would ask the LLM.

In this way, LLMO seems similar to SEO. Afterall, it’s a way to get your information in front of users when they search for something. However, the methods for optimizing content and online authority for LLMs differ from SEO in a number of ways.

But first, the similarities.

LLMO Similarities with SEO

Both SEO and LLMO have overlapping tactics and strategies that currently go hand-in-hand.

Many search LLMs still use search results as a basis for what information they display. The LLM search engine, Perplexity, cited the top 10 ranking sites in Bing or Google in 42-70% of their responses according to a study done by Semrush.

Share of Perplexity Citations by Google Ranking - Semrush
The AI search engine, Perplexity, uses Google search results as its backbone…for now.

Based on this, a solid SEO strategy is essential to help your brand get into LLMs and ultimately in front of your audience. If you don’t have a solid SEO strategy, then both SEO and LLMO should be done in sync, as SEO will help LLMO.

Like SEO, identifying keywords and queries that are most valuable to your audience is also extremely important in LLMO. The process for identifying these keywords for LLMs is very similar as well, as it should be focused on your audience’s behaviors and pain points.

Maintaining a fast and easy to crawl site will help in both SEO and LLMO. Slow page speeds or poor readability will negatively impact both search engines and LLMs. So maintaining a quick site with clear, readable content is key to succeeding in both disciplines.

Creating great content that is unique differentiates a brand in search and LLMs, and is a key SEO tactic. Poor content that is stuffed with keywords doesn’t work anymore. Now content must be approachable, in-depth, and easy to consume for both SEO and LLMs. So no shortcuts! (Looking at you AI-generated “thought leadership”.)

Building authority for your brand and your content is also vitally important to both SEO and LLMO. A well written piece of content is one thing, but without other signals from around the web that the content author is an authority, that content is just words on a page. Establishing authority is essential. This includes getting backlinks to your site from established websites (your typical backlink strategy that is essential for SEO) and to build up the overall authority of your domain. (This will also be different in LLMO, as you’ll see below.)

LLMO Differences from SEO

When getting a mention in an authoritative site, SEO pros will work as hard as they can to get a precious backlink from that authoritative site. That link is a key to giving their own site more authority, and increases its likelihood to rank well in search. LLMs, on the other hand, don’t need a back link. Instead they rely on contextual authority. If LLMs train on the data from the authoritative site and understand the connection between your brand and the other concept or topic in the article, that could be enough for the LLM to attribute these two things to each other and get you into the answer. It may not even need to know the information on your own website to make that connection. This means that more LLMO may take place off of your site than on it.

That is a huge difference with LLMOs; you no longer need to build authority to your own site to get an LLM to attribute a concept to you. You just need your brand and that concept mentioned together in enough places that have authority.

Another difference is unlike search engines, which display a nearly endless number of results for a user to choose from, LLMs give one single answer. That single shot at getting into the LLM response means that if you aren’t present in that initial reply, you or your brand are out of luck. There’s no page 2 of LLM results. Heck, there’s not even a link 2. It’s an all or nothing play for many keywords and queries.

The other big difference between LLMO and SEO is that LLMs will rarely send traffic to your website. SEO has been a major traffic driver for companies and has been one of the main metrics of success used by businesses who implement it as a strategy. But now, LLMs simply give an answer to the user without sending traffic to the source’s website. That means there’s no way for you to easily measure the results or effectiveness of an LLMO strategy and you will then most-likely rely on the user to find your site via a branded keyword search.

Because LLMs can extract information from unstructured data, the use of schema markups may not be as important as it is in SEO.

How to Show Up in LLMs

A lot of what was just covered can help you show up in LLMs. A strong SEO strategy should be one of the roots of an LLMO strategy.

Just like SEO, make sure you know the keywords and queries your audience is searching for. Without this as a starting point, LLMO is essentially worthless. Luckily, many companies that have a strong SEO strategy will have a good place to start. But it goes well past keywords. Think about the actual questions your audience will be asking LLMs.

For example, if you’re targeting the keyword “camping supplies”, you will want to make sure that your content about this covers contextual questions such as:

  • What are essential camping supplies?
  • What camping supplies do I need for different climates?
  • What camping supplies do I need to camp at a national park?
  • Where can I buy camping supplies?

Making sure that these types of questions are answered in your content either as FAQs or as their own posts can help LLMs properly train on and serve up your content to users.

Comprehensively answer these questions. Be straightforward with these answers as well. LLMs are looking for the most direct answer to the problem the person is asking, and they are not great at metaphors. So get directly to the point with the answers. (You don’t have to sound like a robot, but remember that you’re making content for a robot.)

Building authority in LLMs may not require getting backlinks from authoritative sites anymore. Mere mentions on authoritative sites where you or your brand are being associated with the keywords and queries you want to show up for are probably enough, as LLMs will put together that you or your organization are authoritative experts on the subject because an authoritative website attributed both you and the subject together. This goes right back to old fashioned public relations. Getting your brand mentioned in publications along with what you do helps build context for LLMs and makes it more likely that you’ll show up in the replies for your audience. Even things as small as getting mentioned in Reddit posts can help LLMs associate those queries with your brand.

The way you maintain “position” in LLMs is also different. As mentioned previously, instead of trying to rank for a keyword, you’re hoping to get a mention in a response. If that mention doesn’t fulfill the user’s request and the user asks a follow-up question, it could signal to the LLM that the content it has from you isn’t sufficient and should be replaced with another data source. When optimizing for LLMs, you will need to ensure that you’re covering the breadth of knowledge for a question so the answer given to the user is sufficient without them needing to ask another query.