Why Content Sites Should Stop Optimizing for AI Discovery (and What To Do Instead)

AI search has rewritten organic discovery. If your business model monetizes information, optimizing for AI discovery is futile and destructive. Generative Engine Optimization (GEO) or AI Engine Optimization (AEO), in this context means handing your intellectual property to systems that redistribute it with little to no return.

I do not imply that optimizing content for AI discovery is meaningless for every website, what I mean is that it’s not for everyone.

TL;DR

• Content-only sites cannot win the AI discovery game.

• AI citations mostly send high-intent users to brands that sell, not to publishers.

• Google has telegraphed the “cut out the middlemen” direction for years, from snippets and PAA to AI Overviews.

• Your options: diversify audiences off Google or become a real brand with products or services.

• Treat “recovery” as a business model problem, not a technical SEO problem.

Who Actually Benefits From AI Search Optimization?

Optimizing for AI discovery makes sense only for real brands that sell products or services.

Why? Because the value of raw information has collapsed. AI systems are now trained to provide users with direct answers, sparing them the need to click through to a website. For businesses selling products or services, that’s fine. AI cannot fulfill demand for the product itself. A chatbot can’t generate and ship your household products, consult on your Salesforce integration, or deliver a bag of pet food. In those cases, AI visibility becomes a new brand channel.

But for content sites whose only product is information, like news, reviews, tutorials, niche blogs, etc., the opposite is true. AI is cannibalizing your value proposition. 

Every optimization effort you make simply helps the chatbot replace you faster.

Here is a simple unit-economics check: Let’s use the formula for RPM: 

RPM_after_AI = RPM_ads × %clickthrough_from_AI + RPM_affiliate × %affiliate_intent_from_AI

For information pages, %clickthrough_from_AI tends toward zero because the answer is already given. %affiliate_intent_from_AI accrues mostly to brands or marketplaces. If both collapse, your RPM_after_AI trends to zero. If you need a model, start with your current page RPM, then simulate a 50–80% drop in informational clickouts and measure whether any remaining traffic covers fixed costs.

Content-only vs product/service brands

With these said, I must add that there are some instances where AI citations can be net positive for publishers, and these are:

• Narrative control: policy positions, official clarifications, corrections.

• Original data: proprietary studies, field tests, datasets, or calculators that cannot be replaced by generic text.

• Compliance criticals: docs users must verify at the source.

If you do not publish any of the above, AI citations rarely drive durable value.

Why High “AI Traffic Quality” Is Misleading

You’ve probably heard that traffic from AI citations is “high quality.” That’s partly true: users who break out of AI and visit a website tend to have strong commercial intent. But here’s the catch: that traffic overwhelmingly goes to brands that sell something.

For content sites, the bitter truth is users don’t need you anymore. They wanted an answer, AI gave it to them, and their journey is over. Even when some spillover traffic comes through, it’s scraps. That model cannot sustain a content-only business.

Independent data shows the open web is losing clicks fast. In July 2024, SparkToro estimated that only ~360-374 clicks per 1,000 Google searches in the US/EU went to external websites, the rest were zero-click outcomes. In 2025, Similarweb reported that publisher traffic fell materially after AI Overviews, with zero-click behavior rising and AI referrals growing but not offsetting the losses. These are structural shifts, not seasonal noise.

Google Has Telegraphed This Shift for Years

I’ve followed Google’s moves closely for nearly a decade. The writing has been on the wall long before AI overviews:

• Featured snippets and People Also Ask boxes: Google began answering questions directly in the SERPs, reducing the need for clicks.

• Vertical integrations: Flights, hotels, shopping, reviews, entire industries that once relied on middleman content sites were swallowed by Google surfaces.

• Messaging: “Write content for users, not search engines.” On the surface, it sounds noble. In reality, it delegitimizes the very existence of content sites, whose sole purpose is to rank and capture search traffic.

Google doesn’t dislike all content. It dislikes middlemen. If you stand between a user and their end goal (booking, buying, hiring), you’re disposable. If you’re the end provider, you’re safe… for now.

Google’s own comms and updates reinforce the direction. Core updates in 2024 explicitly targeted content “made just to perform well on Search,” and the legacy Helpful Content system was folded into core ranking signals. The message is durable: de-emphasize intermediaries and reward original, useful outputs or end providers.

For context, multiple analyses in 2024-2025 documented steep organic declines for news and magazine sites after AI Overviews, alongside rapid growth in AI referrals. The aggregate picture: more zero-clicks, growing AI-driven visits, net negative for most publishers so far.

3 Problems With the “Niche Site” Model

The niche site business model exploited this dynamic until Google shut it down. What began as enthusiasts sharing real experiences became a gold rush of low-effort reviews and long-form filler churned out by freelancers who never touched the products.

• Reviews became untrustworthy.

• Articles became bloated for keyword coverage rather than genuine insight.

• Sites scaled by quantity, not quality.

If you think about it outside the internet, many of these practices would have been considered fraud. Imagine a magazine filled entirely with ads disguised as editorial reviews. No wonder Google moved aggressively with the Helpful Content Update and follow-up algorithm shifts.

Why Your SEO “Recovery” Isn’t Working

I’ve audited dozens of content sites hit by recent updates. The pattern is always the same:

• Homepage shows nothing resembling a business, only a list of blog posts.

• No products, no services, no unique offering, just content designed to intercept searchers.

• Endless cycles of humanizing and rewriting content, pruning, and link building with no recovery.

The diagnosis is simple: your business model is the problem. Not your technical SEO. Not your backlink profile. The model itself is dying.

If you’re still uncertain whether or not to optimize your content for AI discovery, answer these three questions:

Do you sell a product or service?

If yes, optimize brand facts for AI (catalog, pricing policies, service scope, locations, proof). If no, skip AI optimization.

Is your information proprietary, time-sensitive, or paywalled? 

If yes, consider blocking or metering AI access. If no, assume AI will substitute your content.

Can you convert AI-referred visits profitably?

If yes, proceed with structured brand data. If no, redirect effort to audience building off Google.

Two Paths Forward for Content Site Owners

If you own a content-only site, you have two real options: diversify or transform.

1. Diversify Traffic Sources

Relying on Google, a platform actively cutting you out, is madness. If your content is truly unique or valuable, your audience still exists. They’re just not coming from SERPs anymore.

Where are they? On social media. I’ve seen colleagues drive massive audiences through Pinterest, Facebook, X, LinkedIn, even TikTok. Social is the last frontier for pure content discovery.

And here’s a radical step: block AI crawlers from your site or make them pay. If your business is content, why give it away for free to systems that replace you?

The easiest way to block AI crawlers from having visibility to your content is by adding specific directives in your robots.txt file. Honoring directives in the robots file is voluntary but respected bots generally do. If information is your product, start here and consider firewall rules for enforcement. Google itself notes robots.txt cannot force compliance, so treat this as policy not security.

Robots.txt examples:

# OpenAI

User-agent: GPTBot

Disallow: /

# Google-Extended (AI training)

User-agent: Google-Extended

Disallow: /

# Anthropic

User-agent: ClaudeBot

Disallow: /

# Perplexity (search bot, not for training per docs)

User-agent: PerplexityBot

Disallow: /

# Common Crawl

User-agent: CCBot

Disallow: /

If you allow some and block others, state your policy on a public page. Some publishers allow PerplexityBot to encourage attribution and link-outs while blocking training-oriented bots. Choose based on your model, and remember you can change stance later.

2. Become a Real Brand

The better path is transformation. Shift from being a content library to being a business.

Case in point: a friend ran a pet blog that pulled 200K monthly visits for years. After repeated traffic declines, down to 40K, then to a mere trickle, he tried everything: content pruning, backlink campaigns, technical fixes. Nothing worked.

The turnaround came when he pivoted. We researched pet products, launched a Shopify store, and rebuilt the site around commerce. Blog articles were clustered around product categories. Internal linking was strategic. Company pages and About Us were rewritten to position the site as an actual brand, not a faceless blog. Here are what we changed in 6 weeks:

• Re-mapped articles into product-led clusters, each pointing to a collection and 2-3 conversion pages.

• Rewrote About, Contact, Shipping, Returns, and Category intros to read like a real retailer, not a blog.

• Killed posts not aligned to inventory or services, redirected where topical overlap existed.

• Replaced generic “best X” listicles with “how to pick” buyer’s guides tied to SKUs.

• Set a social posting cadence that recycled existing evergreen posts and short video explainers.

The result: ~80K visits/month. Lower than the peak, but infinitely more valuable because traffic now aligned with sales. The business became sustainable.

The same could be done for services. Had he been qualified, offering pet training or veterinary teleconsults would have worked just as well. The key was moving beyond information into something real.

What This Means for the Future of Organic Discovery

For content site owners, the era of “ranking for information” is closing. AI and Google have made information free, abundant, and instantly accessible.

• If you sell only information, your moat is gone. Diversify traffic or block AI.

• If you sell products or services, focus on AI brand optimization: feed AIs accurate, authoritative data about your offerings, features, and differentiators so they recommend you.

• If you run a hybrid, restructure so that informational content supports the business, not the other way around.

Organic discovery hasn’t died. It’s just shifted from information to brands.

How to measure the pivot

• Create an “AI referrals” channel grouping (examples can be chat.openai.com, perplexity.ai, copilot.microsoft.com, gemini.google.com).

• Monitor % of sessions that touch product/service pages and revenue per session for aligned clusters.

• Track branded search share vs non-brand informational; the goal is a higher branded share even if total sessions fall.

• Compare social-referred saves/shares vs SERP clicks for your top evergreen assets to validate audience portability.

Market data shows AI referrals are growing quickly year-over-year, but they do not yet offset zero-click losses for most publishers. Plan accordingly.

Key Takeaways

• Content sites monetizing pure information cannot win in AI search.

• AI optimization makes sense only for product/service brands.

• Google has always targeted middlemen and will continue to.

• The niche site model collapsed because it scaled fraud over trust.

• True recovery isn’t about audits but becoming a real business.

• Social platforms and brand building are the only sustainable paths.

If your homepage is just a blogroll, the jig is up. Stop waiting for a “content site recovery.” Either diversify your traffic sources or transform into a business. Anything else is just boiling slowly while Google and AI turn up the heat.