What Actually Works in AI SEO (And What Doesn't)

If you want quick wins with AI and SEO, pumping out dozens of niche-targeted blog posts daily with AI can spike your traffic, especially if you target ultra-specific searches (think “top legal software for dentists in Houston”) and always feature your brand. Big brands with strong domains are even more insulated from risk, but this approach is short-sighted. It degrades your brand and offers little value to real people. Sustainable growth requires discipline, consistency, and content people actually want to cite. Short-term spammy tactics may work (for now), but genuine reputation, recognizable authority, and clear brand signals matter a great deal for both Google and LLM results, much more than any “hack.”

Does Mass AI Content Actually Work?

Yes, it can temporarily boost your site’s visibility. Here is the basic playbook people are using:

• Identify dozens or hundreds of ultra-specific, often low-volume commercial queries (sometimes by simply copying what LLMs and competitors already surface).
• Use AI to generate brief posts or listicles for each, always featuring your own brand as the top solution.
• Mix in big recognizable names (Salesforce, HubSpot) plus uninspiring competitors, so the list looks objective on the surface.
• Repeat daily. Some sites churn out 30 or 50 posts per day using this formula.
• The speed and volume make it difficult for competitors or Google to catch up, at least until something changes.

So, yes, these sites can achieve 10x traffic increases from LLM appearances and sometimes from Google, at least in the short term. But there is more nuance behind the scenes.

Short-term tactics like mass AI content might deliver a burst of clicks, but they rarely bring lasting trust or brand value.

How Long Does This Last?

Not long. Every time a loophole appears, even the biggest sites only stay ahead for a while. Google’s history is full of these cycles. One year it was auto-generated city pages, then “best X for Y” lists, then UGC spam, and so on. Now we see it play out again, short win, quick shutdown, repeat.

With a big enough domain, you might get away with it longer. But even established brands see eroding engagement and sometimes brand blowback when visitors notice generic content that always ranks its maker as the “top” pick.

Over time, these strategies almost always lead to more headaches:

• Brand trust declines as customers notice repetitive or self-serving lists
• Google’s algorithms and AI overviews evolve to ignore suspicious signals
• Genuine sources, news, government, editorial, become harder to displace
• Churned pages often drop as fast as they rose

What Are Big Brands Actually Doing?

Most don’t risk their main site. If you are an established company with real customers, you care about brand reputation. Plastering 40 self-congratulatory AI-written lists each day across your site is obvious to anyone who checks.

Some tuck “questionable” articles away in obscure subfolders, or use clever naming (“columns,” “community corner”) and bury these sections at the bottom of the blog. Guest authors or questionable pieces are stashed there to keep general users from stumbling onto low-value pages, but still benefit from the extra links and citations. It’s pragmatic, but at some point, everything indexed is still your brand.

Even if you bury weak content in subfolders, the risk is there, if the page is indexed, you own it in the eyes of search and users.

You also see variance between strategies:

• Mass AI content on “satellite” or throwaway domains, not the homepage
• Major brands using their domain power carefully, for genuine comparison or review content (with a real human touch)
• A growing number of companies focusing on long-tail commercial intent searches, but tempering AI automation with light human editing or sensibility checks

What About LLM Optimization?

This is the current frontier. LLMs like ChatGPT, Perplexity, and Google AI Overviews pull their “answers” from sources across the web, with more weight given to:

• Editorial sites (mainstream news, niche blogs, large magazine publishers)
• Specialized or official resources (government, university, respected directories)
• Recognizable brands (the ones LLMs already “know” and can confidently reference)
• Pages that echo consistent, verifiable phrasing about a brand (“Top Seattle bike repair shops – SeattleBikes.com” on multiple trusted sites)

The more consistently you describe your brand and expertise across the web, the more likely LLMs are to “understand” who you are and cite you.

One overlooked tactic: use the exact wording you want to be associated with everywhere, website, LinkedIn, local listings, PR bios, directories, even in your press releases. If you sometimes call yourself a “dental CRM consultant” and other times a “practice management software consultant”, pick one title and run with it everywhere so the association becomes clear.

Where Do LLMs Pull From?

Their biases differ:

• Google gives disproportionate space to Reddit, YouTube, and familiar UGC sources
• Perplexity likes YouTube and editorial content, sometimes official orgs
• ChatGPT leans toward editorial websites, avoids much YouTube
• Some LLMs give more attention to sites not blocking AI crawlers (which is another turbulent area with Cloudflare’s recent changes)

I tested a prompt in several AI search tools recently and saw strikingly different results for niche “best of” queries. In one tool, our brand was listed at number two, while in another, we didn’t appear at all. That kind of variance is common, especially outside top-tier commercial categories.

Here is a simple illustration:

LLM / AI Tool	Sources Favored	Example Search (‘best time tracking app for architects’)
Google AI Overview	Reddit, YouTube, News Magazines	Lists big SaaS brands, one Reddit thread, and a random blog with matching review
Perplexity	YouTube, government and editorial	Mix of 3 SaaS, 2 personal productivity blogs, and a YouTube demo
ChatGPT (web)	Editorial, major publishers	Avoids video, favors magazine-style “top 10” posts from sites like CNET and PCMag

Is Structured Data or LLM.txt Important?

Not in the way most people think. There’s a myth going around that you must load your site with schema markup (even niche ones) or slap an “llm.txt” file on your web server to win AI citations. But that’s not how it works.

Structured data is useful for search engines, but there is no proof that more schema, or any llm.txt, will make LLMs or Google favor you as a citation source.

Schema helps Google better categorize your content and may help with featured snippets, but LLMs decide what to cite using much broader context and trust signals. “Llm.txt” was never a real ranking factor. In fact, all the top content marketing and SEO tools do not use it, and none of the LLM providers say they look for it.

If there’s no point, and you risk leaking sensitive details in your “llm.txt”, why waste time?

Building Long-Term Trust and Visibility (That Actually Pays Off)

The tactics that keep working, whether search is ruled by links, LLMs, or something else, aren’t glamorous:

• Be ruthlessly consistent in your brand description. Use the same tagline, expertise line, and summary everywhere, online and off.
• Get cited or mentioned in places AI and Google trust, reputable editorial sites, big industry blogs, media outlets, official directories, and popular social channels.
• Ensure your commercials pages (the ones that convert) directly answer niche buyer intent, like “best budget 3D printers for teachers”, and highlight why you are a strong pick for that need. Supply proof, testimonials, or comparisons rather than fluff listing.
• If you run guest content or swing at new search verticals, bury risky or off-brand posts in clearly marked, secondary sections to limit damage to your main brand.
• Review what LLMs cite or use for your target searches. Reverse engineer the types of sources they favor, then get mentioned, recommended, or linked there. Sometimes, getting cited once on a mid-tier review roundup is enough to get you into answer boxes and LLM outputs.

How to Handle Sourcing for LLMs and AI Overviews

I checked this week for a friend in B2B SaaS. Their brand did not appear for “best payroll platforms for East Coast startups” in Google’s AI Overview or Perplexity. Why? Every LLM result favorited sites that had written more than one list, drew on a blend of customer reviews plus editorial picks, and described platforms (not just individual features) using the exact keywords startups might search for.

After we structured a page to answer exactly that query, using language already showing up in AI overviews and making sure our brand was always described the same way everywhere, the page crept into both Google’s featured box and LLM lists within a month, with minimal backlinks. It worked, but only because we looked closely at what the models already trusted and built directly for that behavior.

AI-driven search is unpredictable, but consistent language and citations from sites the models already trust will go further than a hundred recycled best-of lists.

Does This Mean SEO Is Dead?

No, not at all. Fundamental skills still matter. Less people are chasing the same scraps of basic SEO because the easy wins have shifted to social (short video and UGC). But if you can find undertargeted, transactional keywords, gaps that have real buyer intent but little competition, you can still build lasting funnels that make money for years. AI tools have not (yet) killed the game for anyone willing to look beneath the surface.

If anything, the rise of AI and LLMs has made doing the basics (real citations, consistency, genuine product pages, authentic testimonials, and expertise signals) more important. Hype cycles come and go, but Google, and now the models behind AI search, always return to verifiable trust and clear authority when the dust settles.

Practical Steps: Where to Focus Your Efforts

The basic “to do” list for anyone building visibility now is pretty simple:

1. Pick a single, clear brand or expertise line and use it everywhere
2. Create dedicated product and service pages answering specific, high-intent commercial queries
3. Secure mentions and links from reputable editorial or media sources, even small ones count if they’re trustworthy
4. Check what shows up in AI overviews and top LLM outputs for your vertical, and model your messaging and structure accordingly
5. Only use AI mass content on non-critical or peripheral sites. Treat your main brand asset as a long-term investment, not an experimental sandbox

Examples (Not Just Theory)

• A regional accounting software provider can build “best accounting tools for restaurant chains in Ohio” pages using actual insights, case studies, and side-by-side comparisons, then get mentioned in Ohio-based business journals and trade groups rather than just reposting endless best-of lists
• A coaching brand sticks with “Career Coach for Tech Leaders” everywhere (site, LinkedIn, profile bios, directory listings) and requests reviews using that exact language. Eventually, this term becomes synonymous with their actual brand, not just generic coaching noise
• An e-commerce logistics company asks trade journalists to include them in annual fulfillment platform roundups, even if they are listed second or third. The resulting link from an established trade mag lands them in LLM answers (and brings targeted organic traffic), instead of endless weak directory links

These approaches are common sense, but they work far longer and with fewer risks than any fleeting loophole.

Key Myths Worth Ignoring

• You do not need endless structured data or to maintain “llm.txt” files. There is no evidence these matter for AI rankings or LLM citations.
• Changing your site’s headline format to match LLM prompt structure (turning everything into questions) is weak at best and a waste at worst.
• Mass guest posting or focusing on third-tier directories never beats actual editorial mentions and natural citations (from sources people trust).
• If a supposed “hack” sounds too easy, it usually has a short shelf life. Think meta keywords, everyone rushed to use them, then Google simply ignored them.

Final Thoughts (Without a Summary Heading, as requested)

Much of the advice you see in modern SEO boils down to looking for short-term exploits. But the practical reality is unchanged: trust, recognizability (both “in real life” and digitally), and authoritative content win for both humans and AI. There are no long-term “cheats,” just cycles of hacks that dry up. If you focus on the basic principles, clarity, consistency, and genuine real-world mentions, your brand will stand up through the next round of algorithm and AI changes.

🤖 Summarize with ChatGPT 🔍 Summarize with Perplexity 💡 Summarize with Claude