anyone try new techniques that work in 2026?

It seems like everyone is still pushing getting backlinks from high DA sites, guest posting, etc.

Is anyone doing anything to get into llm training data?

Im about to give it a shot, knowing it might take 3-4 months to be even able to test the results

Do fake authors and build them up. Slay on Google News

Think long-term - digital PR, expert contributions, and structured content will compound over time, but testing impact will definitely take months, not weeks.

yes - structured content for sure. I just finished writing a python engine that takes my generic content and spits it out in the structures I need for what im up to.

LLM training data isn’t something you can game short-term, takes forever and no clear feedback loop.
What’s actually working is getting mentioned on high-trust sites (Reddit, niche forums, news) + branded searches.
Think entity building, not just links, Google + AI both lean on that now.

LLM training data is actually something I've been thinking about a lot lately. The idea is solid - if your content gets picked up into AI training sets, you essentially get cited by ChatGPT, Perplexity, Claude without any traditional SEO work.
The problem is nobody really knows what gets scraped and when. Common Crawl is the main dataset most LLMs train on, and they crawl everything but there's no timeline or guarantee your stuff makes the next training run.
What actually increases your chances - get your content on platforms that are definitely in training data already. GitHub, Reddit, Hacker News, Wikipedia citations, Stack Overflow. A page that gets referenced from those has way higher odds of ending up in the next dataset than a random blog post.
Also structured data helps. LLMs love clean, factual, well-structured content. If your page looks like something an AI would want to quote, it probably will get quoted.
3-4 months is optimistic honestly, training runs don't happen that frequently. But it's worth doing alongside normal SEO, not instead of it.

vakage - check your dm

Binkmaster said:

It seems like everyone is still pushing getting backlinks from high DA sites, guest posting, etc.

Is anyone doing anything to get into llm training data?

Im about to give it a shot, knowing it might take 3-4 months to be even able to test the results

Click to expand...

In my opinion, you should focus on the basic and core techniques of SEO rather than chasing new concepts that may take your time without delivering real results.

yeah traditional link building is getting super crowded so shifting to generative engine optimization is definitely the right move right now. trying to get baked into the actual base training data of models like gpt or claude takes months of waiting for their next big parameter update, but if you focus on rag instead you can get cited by perplexity or chatgpt search almost instantly. the best way tbh is just to pump out highly structured pages with original statistics and get your brand mentioned a lot on high trust sites like reddit and big publishers that the ai bots are actually scraping daily for real time answers.

What's working for me right now is AI-assisted content combined with strong internal linking and topical clustering. Forget trying to rank single pages, build out a whole topic properly and Google rewards the cluster. Also LinkedIn and Reddit content is getting picked up in AI summaries a lot more now so being active there helps with brand visibility even outside traditional SEO. The LLM training data angle is interesting, curious to see what you find in a few months.

LLM training data is the new gold rush. Most people are just spamming schema, but getting into the actual datasets usually means hitting the big web crawls like Common Crawl or staying active on Reddit/GitHub where models scrape the most. GL with the 4 month test, keep us posted if the "AI SEO" actually pays off!

Interesting angle—LLM visibility is still early, but it’s less about “getting into training data” and more about being cited on trusted, high-quality sources that models learn from. I’d focus on publishing original research, data-driven content, and getting mentions on authoritative sites. It’s a slow play, but worth testing alongside your regular SEO efforts.