When creating content for SEO you’re catering to an algorithm with a wider audience with different interests and needs. AI or LLMs are creating more personalized results on the same or similar queries based on what they know about the user. This is why one person sees one set of results and sources in an output than a different user on the same query.
To begin surfacing in an LLM (ChatGPT, Claude, Perplexity, etc…), in addition to third party signals that are actually meaningful, your content and website experience needs to focus on the different customer personas and make sure these needs of these personas are met.
This does not mean you should create a million satellite pages or stuff uneccesary content into articles, it means you should focus on your customer base and meet their needs where you can. From there the LLM learning systems will know who you cater to and be able to surface your brand or your information when it is relevant to the user.
- Traditional search relied on finding the most factual information and generating a webpage, video, shopping ads, and rich results.
- RAG relies on the search result to ground an output, but also knows what is needed for a solution by combining what it knows is needed.
Our reliance on putting every detail on a single page is no longer as important as RAG knows what products and services can meet a user’s needs by combining the elements it has learned from how to guides, product specs, and with the help of vectors and entities. It’s why it can say what goes into baking sourdough bread or creating a specific hairstyle and cater the output to someone who is new to cooking or has short hair vs. long.
An example could be me asking a group of people “Why don’t most kitchens have mirrors in them?” This sounds like a weird question for us as people right? For us yes, for an LLM not so much as it is a logical input and will need a reasonable output.
Depending on what the person is interested in for hobbies (they do), interests (topically), and brand loyalty (shopping history and brand engagement), LLMs will likely customize the output.
- Home cooks and chefs may get responses that there is extra light being reflected and it can make it less safe to cut and chop food.
- Photographers may have responses that talk about messing up light in the room and disrupting shots as part of the response, and information on how to account for it.
- Plant parents may have information on the added ambient light hurting the plants, or how the refracted beams may focus in on a plant that cannot handle direct light and hurt it.
The plant part would not be relevant for the home cook, and while home chefs are likely posting photos to Instagram and videos to TikTok, photography may not be in their known interests or entity, so it may not show up for each. Photographers that need a clear screen for editing or that do product photography in their home would need to offset the additional light caused by the reflection, so the output could have talking points about this.
So what do you do here if your company sells mirrors or renovates kitchens with mirrored/reflective surfaces? Simple, create content for your users and not for every single use case.
Side note: I am making the above up, I have zero idea if people are actually searching for this. The topic and idea came into my head while I was on my run this morning.
Here’s a few ways this could be done on a website if this is your industry and you want to sell reflective surface products for kitchens.
- Product pages can include specs and selling points about the reflective surfaces like how it could be shiny but control ambient light, or how it can enhance or diffuse light to control the amount available.
- Creating tutorials that are helpful vs. a one off situation with kitchen renovations can help here too. Examples inside the articles for placements and structuring could include creating extra ambient light that helps task lights be more effective, or how you can use the additional light to generate more ambient light and reduce the need for electricity during the day. Then these can be turned into video content on YouTube and social media that can further be used to know about the products.
- Content about accessories and decor items to create different moods or atmospheres help meet the needs of a space. Drapes and what the materials do to the light, colors and how they absorb or reflect, and different surfaces and materials. Each of these show authority and provide solutions while feeding what the products and mix-and-match accessories can do for them to help people based on their life and work needs.
When these types of content combine, LLMs can use entities and vectors along with RAG to identify products and the stores or service providers that sell them, even if it doesn’t say it right on the page. On top of this they can identify the ones that are in the price range, provide the right brand experience, and cater to the individual rather than a person searching for a query once they know enough about the user’s individual history.
There is more that goes into it including the users chat history and trackable shopping history, as well as third party trust builders like actual content on third party websites and the information the brand shares that is helpful vs. self promotional. But this is a very likely way we’ll be seeing LLMs and even AI Overviews generate outputs in the somewhat near future. It has already started, and if these systems continue to grow, we can expect these systems to create this level of output. That gives small brands a huge chance to compete as long as they don’t chase silver bullets and focus on meeting their customer needs.
