7 AI Search Tactics That Aren’t Worth the Budget (& 3 That Are)

Content marketing experts generally agree that B2B brands will benefit from appearing in AI citations. Case in point: A recent Forrester survey of nearly 18,000 business buyers found 94% of respondents used generative AI to make their most recent purchase.

There’s less agreement on how to secure this lucrative real estate. When putting together our latest eBook, “How scaling B2B brands win in AI Search,” Eleven looked into which factors most closely correlate to a brand’s appearance in Google’s AI Overviews, ChatGPT, Perplexity, and other LLM search results.  

We found that some of the purported tips and "tricks" making the rounds on search blogs are unlikely to drive placements and impressions. Here are seven search tactics that aren’t worth prioritising, followed by the big three that are.

>> Download the full report: How scaling B2B brands win in AI Search

These AI ‘tips’ are (mostly) myths

1. You should chunk your content for AI consumption.

Over the last year, some AI Search practitioners started suggesting that chopping content into bite-sized chunks would help LLMs find and cite brands. The hypothesis is that AI search systems prefer retrieving individual passages rather than full articles.

But Google explicitly advises against the practice. It strongly recommends writing for people, not machines. And while other LLMs may reward some structural good practices — like answering a question early on the page or organizing information under descriptive headings — there’s little more than anecdotal evidence to support content chunking at scale.  

In fact, studies testing whether “content chunking” helps AI retrievals have produced mixed results. While some found benefits, no single approach or chunk size consistently performed best. NVIDIA found that retrieving whole pages was more accurate on average, suggesting there’s little basis for reformatting an entire website around one supposedly ideal structure.

2. Your content needs to be a certain length.

The idea that word count can influence rankings predates AI search. You’ve likely heard at some point over the last decade that “longer is better” for traditional SEO — a school of thought that gained traction after several widely shared (and ultimately debunked) studies in the early 2010s found that top-ranking Google results tended to contain more words.

Now, the pendulum is swinging in the opposite direction, with some SEO practitioners arguing that, in line with the chunking theory, shorter pages are easier for LLMs to retrieve and cite. 

The data, however, has proven inconsistent. A recent Ahrefs analysis found that, while pages under 1,000 words earned slightly more AI Overview citations than longer pages, word count had virtually no relationship with citation position.

As part of our AI visibility research, Eleven ran a small citation correlation analysis across about 300 pages and 1,257 URLs from a B2B partner. We found that word count was one of the weakest predictors of how often ChatGPT, Perplexity, or Claude would cite an article.  

3. You need an llms.txt file to excel at AI search.

First proposed circa 2024, llms.txt is a Markdown file that provides AI tools with a curated list of important pages and information about a website. Supporters liken it to a sitemap for LLMs and suggest it can help AI systems find and understand a brand’s content.

The catch is that llms.txt has yet to become a universally adopted web standard. Google, in particular, has said it doesn’t use the file to inform its Search results, including its generative AI features.

Beyond that, having a file for the AI tools that do read it won't inherently solve other problems that may be preventing your content library from earning citations. For instance, an llms.txt file can’t make blocked pages crawlable, improve weak content, or earn backlinks for your page.

4. Your AI visibility score is an absolute measure of performance.

AI visibility platforms such as Peec and Profound track how often brands appear across a selected set of prompts and AI search engines — and it’s certainly tempting to rely heavily on this tech, given how black-box-y LLM citations can feel. 

But research shows these platforms’ results can vary widely based on which prompts the tracker selects, which AI platforms it monitors, how often it runs queries, and more. A 2026 study of Perplexity, OpenAI SearchGPT, and Gemini, for instance, found that identical queries produced different citations over time and that many apparent differences between domains fell within the normal noise of the systems.

At Eleven, we've seen similar inconsistencies, both when using existing platforms and when testing our own AI citation dashboards or measurement methods. That’s why we advise partners not to overindex on proprietary visibility scores or similar results.  

5. You can ‘hack’ AI mentions by promoting your brand across Reddit.

Reddit and other social communities began to dominate search engine results back in 2023 after Google updated its ranking system to surface “hidden gems” or useful first-hand content. As AI search engines appeared to follow suit, some marketers have pushed manufacturing Reddit mentions at scale.

That strategy is innately risky. For starters, Reddit’s spam policy prohibits mass-posting repetitive content for exposure or financial gain, automated promotion, and other forms of artificial engagement. Small forums — and their individual members — can be equally unforgiving of brands trying to game their communities.

Moreover, resulting visibility has proven unstable. In August 2026, Promptwatch found that Reddit's share of ChatGPT Search citations collapsed by 86% in a single day, falling from a steady average of around 2.8% to just 0.5% on 14 August" The reason for the decline is unclear, although some speculate it’s a reaction to brands’ overuse of the platform as a marketing channel. In any event, the shift demonstrates how quickly visibility built on a single third-party platform can disappear.

6. Adding schema will automatically improve LLMs citations or mentions.

Schema markup gives search engines machine-readable information about a page, such as its author, organisation, products, or ratings. Because it helps machines classify information, some GEO practitioners suggest that adding it anywhere and everywhere can improve a brand’s chances of appearing in AI-generated answers.

But while schema is a useful technical SEO practice, there’s little evidence that adding more of it innately increases AI citations. Indeed, LLMs were primarily trained on human-readable content like blogs and literature. There’s no reason to assume LLMs should ‘prefer’ schema over other content. Moreover, Google explicitly states that websites don’t need any special schema.org structured data to appear in its AI Overviews or AI Mode.

7. Internal linking is an AI search lever.

To be clear, internal linking plays a vital role in traditional search. It helps search engine bots crawl and identify your site’s new or most important pages more easily. Still, our correlation study found the practice has almost no direct effect on AI citations — a finding we felt worth flagging as some agencies have moved toward selling internal-link optimization as a dedicated part of their AEO services.

So What Works for AI Search? 

Eleven found that the pages LLMs ultimately cite are often the ones already earning impressions and clicks in Google — and that the other strongest predictors of success are referring domains and keyword breadth.

In other words, while the lingo may have changed, the fundamentals of search have largely stayed the same. That’s why we work with brands to prioritise these three tactics:

• Create expert-led content: Google's emphasis on Experience, Expertise, Authoritativeness, and Trustworthiness (E-E-A-T) hasn't waned. If anything, AI's ability to facilitate content creation (and encourage it at scale) has made it more important, since it's the most significant input LLMs can't fake. That's why Eleven works with carefully vetted and accredited subject matter experts to surface original insights, explain complex proprietary data, and craft genuinely helpful thought leadership for our partners.
  
  
• Establish brand authority: Our research reinforced newer studies that have found LLMs give outsized influence to outside influence. That’s why we combine high-quality onsite content with offsite distribution and digital PR. Through targeted outreach, we help brands earn authentic coverage, mentions, and long-lasting backlinks from authoritative domains, strengthening their visibility in traditional and AI search.
  
  
• Show up on the channels AI reads: Offsite authority extends beyond backlinks. Your brand should also appear on the platforms your target audience trusts and spends their time. That includes podcasts, YouTube channels, review sites, community forums, trade publications, and more. These mentions can expose your business to new audiences while giving LLMs more independent sources to draw from when deciding which brands to recommend. Eleven helps partners build this presence through digital PR and content designed to perform across audio, visual and written channels.

Working together, these three levers form the foundation of "Expertise, Everywhere," Eleven’s bespoke approach to content marketing in the wake of AI Search. You can learn more about this approach and how we put it into action by downloading “How scaling B2B brands win in AI Search” today.