A few years ago, Google was the go-to for almost every online search. That’s changing. People now turn to AI tools like ChatGPT and Perplexity for quick, aggregated answers. Alternatively, they go to communities on Reddit, LinkedIn, TikTok, and Instagram for personal recommendations.

If your content strategy still revolves around ranking in SERPs for top-of-funnel keywords, you might be chasing a shrinking opportunity.

A year ago, if I wanted to understand a specific technology, say, “edge computing?” or “SQL partitioning?”—I’d go to Google, sift through results, and hope to find an article that wasn’t just fluff stuffed with keywords, and clearly explains the concepts to my understanding..

Now? My first stop is ChatGPT. I ask questions as I would to a teacher, refine my query, and get a direct, well-structured explanation—without ever clicking a website. Companies like PlanetScale or Fly.io that may have spent time and effort ranking for these search terms might never see me on their site because I no longer need to use search.

Just last week, while trying to show a friend a cool new tool I had found, I grabbed his laptop, opened a new tab, and to my surprise he set his browser was set to load the ChatGPT homepage instead of Google for every new tab. He told me he hadn’t used Google Search in months. And I know he’s not the only engineer with a setup like this.

Even when people do use Google, 60% of searches now end without a single click, according to an analysis by Rand Fishkin of SparkToro. Search for anything today on Google search, and here’s how the results typically appear:

• At the very top, AI-generated answers,
• Then, paid search results,
• Then, the People Also Ask section,
• And finally, organic search results—the content brands have spent time writing and optimizing—buried beneath all of that.

Google has taken a page from social media platforms like X by designing search to keep users inside its ecosystem. What used to be prime real estate for publishers is now increasingly being taken over by Google’s own services.

If your team has been discussing what it means to do content marketing in the age of AI, send them this piece.

While the examples I use here are software-engineering-focused because I’m a developer advocate, this shift in search behavior affects every industry. The way people find and engage with content is changing. Let’s talk about what that means, and what to do about it.

Top-of-Funnel SEO Is Losing Relevance: Google’s Changing Landscape and the Zero-Click Problem

Like I said, if you’re still pouring resources into ranking for broad informational keywords, it might be time to rethink your TOFU strategy.

In another brilliant research report from SparkToro’s Rand Fishkin, 332 million Google searches from 130,000 U.S. devices were analysed over 21 months (Jan 2023 – Sept 2024) and found that 51.7% of searches are informational.

This isn’t surprising.

Informational searches have always made up the largest share of queries, which is why TOFU content has been a core pillar of SEO-driven content strategies for brand awareness.

At the TOFU stage, users aren’t looking for a specific product or company. They’re in research mode, trying to understand a topic or problem. They want definitions, explanations, and high-level guides. At this stage, searches would be things like:

• “What is Zero Trust security?”
• “How does edge computing work?”
• “What is database sharding vs. partitioning?”

For years, brands have been using TOFU content to capture potential customers early, educate them, and subtly introduce their product or service. But that strategy is losing effectiveness, not because people are searching less, but because Google is making sure they don’t need to click away to get answers.

To reiterate, 60% of informational searches now result in zero clicks. That means for every 1,000 searches, only 360 clicks make it to a non-Google-owned, non-ad-paying website.

This isn’t accidental. It’s by design.

Users now expect a single, definitive answer rather than sifting through countless search results. AI tools like ChatGPT and Perplexity have trained them to value quick, summarized responses over deep dives. Google has recognized this shift and is quietly transforming from a search engine into an answer engine that keeps users on its platform instead of directing them to external sites.

In June 2024, Google introduced Search Generative Experience (SGE), a new feature that places AI-generated summaries at the top of search results. These summaries pull information from multiple sources and remove the need for users to visit external websites.

As if that isn’t bad enough for SEO efforts, there have been rumours of Google Search making plans to roll out an 'AI Mode' within Google Search that will look nearly identical to its Gemini AI chatbot. Instead of clicking on links, users will simply receive AI-generated answers right on the search page.

Over time, as AI-generated responses improve, even fewer users will browse through search results.

For the 40% of searches that do result in a click, the data shows they’re largely branded or navigational:

• 44% of Google searches are branded, meaning users already know the platform or website they want to visit and use search as a shortcut. The top searched terms include YouTube, Gmail, Amazon, ChatGPT, WhatsApp Web, and Instagram.
• 33.1% are navigational, e.g “AWS documentation” or “Fly.io pricing.”
• 14.5% have commercial intent, e.g “Tailscale vs. ZeroTier” or “SOC 2-compliant Zero Trust platforms.”
• Only 0.69% are transactional, indicating direct purchase intent.

What this tells us is clear. By the time someone searches on Google now, they’ve already been influenced elsewhere. The awareness and consideration happened on another platform, perhaps through social media, AI-driven recommendations, or community discussions.

Google is slowly evolving from discovery engine to an endpoint for users who have already made up their minds. If you want to be part of their decision-making process, you need to be where the awareness actually happens. That means looking beyond search engines and focusing on the platforms where your audience actively engages, learns, and discusses the topics you’re covering.

How to adapt to imminent search behaviour changes

Let’s get one thing straight, this is not a call to abandon content creation or burn your SEO strategy to the ground. Content still matters, probably more than ever.  But what you create and how you distribute it needs a rethink.

If you create content in the technology space like I do, you should recognize that your audience are usually first adopters. They’re most likely ahead of this behaviour change. If they’re adapting, we should too.

Consider all the recent moonshot/disruptive inventions (ha, when you’re doing an MBA, you’re bound to pick up a few phrases like moonshot and disruptive) like Uber, Airbnb, Telsa, Space X, and co. They were initially met with a lot of skepticism but ended up changing consumer behaviour across travel, transportation, and ownership significantly.

The Industrial Revolution did the same. Before coal, wood was the primary fuel source. Feudal lords controlled wealth through land and agriculture. Then steam engines replaced manual labor, factories rose, and cities grew. The invention of cars and trains made it possible to live far from work, creating the world of commuting we know today.

The digital age is undergoing its own upheaval. AI is reshaping how people find information, and assuming that content discovery and consumption habits will remain unchanged is wishful thinking at best.

                     Will ChatGPT now become the largest traffic referrer?

Here’s what I think we should be doing to stay on top of these habit changes:

1. Prioritise putting your content where your customers spend time organically

If SEO isn’t the powerhouse it used to be, what should companies do instead? Simple: double down on content discovery and distribution. Find out where your customers are and go there.

The thing is SEO or Search engines like Google and Bing have always been just one medium for content discovery. But because they dominated how people discovered information for so long, we ended up assigning them outsized importance, at the expense of other discovery channels.

However with all the data I’ve shared, you can agree with me that it’s no longer enough to publish content and wait for search engines to send traffic your way.

Marcus Sheridan, author of They Ask, You Answer (one of my favourite books on content marketing), on a social media examiner podcast episode, said that Google may see a 25-40% decline in usage over the next decade. More importantly, for publishers and businesses, there could be a 70% decline in website visitors from search results.

You have to put your content where people already are. Places like:

• Social media & professional networks (LinkedIn, Twitter/X, Reddit, Hacker News)
• Community platforms (Dev.to, Hashnode, Medium, FreeCodeCamp)
• Newsletters (both owned and through guest contributions)
• Video & short-form content (YouTube, TikTok, Instagram Reels)
• Conferences?

I like to put my money where my mouth is. That’s why content distribution and repurposing are at the top of the content strategy plan I wrote for my team in 2025. We’re approaching distribution as seriously as we approach production.

Here are two things we’ll be doing differently in our dev advocacy team:

a. Content Distribution Wednesdays:

Since Wednesday at 8 AM is one of the best times to post on Hacker News, I’ve challenged my team to distribute or repurpose content weekly.

Every Wednesday, we’d mark off time on our calendars to take a piece from our content barn (preferably something fresh and relevant to a current campaign) and repurpose or distribute it on Reddit, Hacker News, LinkedIn, or X, both on personal pages and company channels. I’m in the process of automating this reminder for slack.

b. Syndicating content to platforms with built-in distribution:

Since we’re still on the topic of being where your audience are, developer-focused publishing platforms like Dev.to, Medium, Hashnode, and FreeCodeCamp already have huge engaged developer followings with built in distribution.

We’d be serious about syndicating content to one of those platforms or even making one our primary content source. I’ve even toyed with the idea of syndicating our email newsletter to a corresponding LinkedIn Newsletter. A couple of people in my network have found massive subscriber success with it, and it all boils down to built in distribution.

2. Write for the machines?

Platforms like Perplexity already cite sources when surfacing AI-generated responses. If AI-driven discovery grows, companies may need to structure their content so that AI tools can accurately retrieve and reference their insights.

Lately, I’ve noticed utm referrals from ChatGPT showing up in our HubSpot marketing analytics reports.

Just two days ago, I found myself using ChatGPT for research I’d normally do on Google. I asked:

“What are the top 10 most popular SaaS apps that support IP allow lists and are widely used across enterprises?”

ChatGPT returned a list containing SalesForce, MicroSoft Azure, Okta, MongoDB Atlas, citing two pages from Bayansecurity.io docs titled “Secure Saas Applications with IP Allowlisting” and “Use IP Allowlisting to enforce zero trust policies for specific SaaS Applications integrated with Okta and Entra ID” as it’s source**.** The list also included tools like Palo Alto, Rubric, OpenVPN and Sailpoint, citing doc pages of the respective tools on setting up or managing IP allowlists.

But here’s where things got interesting: It ranked Workday as #2, citing a GeekFlare blog titled “Top 41 SaaS Companies Dominating the Market in 2025.” The problem is, that blog never mentioned IP allowlisting at all.

Curious, I ran the same search on Google. Instead of surfacing relevant documentation, Google gave me generic “Top 10 SaaS platforms for businesses” articles, completely missing the nuance of what I was actually asking.

So, how do LLMs rank or surface web sources?

I don't know definitively. But one thing is clear: their ranking mechanisms are different from Google’s. That said, keywords still matter, so keep surfacing them in titles and throughout your content.

When I asked for sources for my “What is Database Sharding and Partitioning?” query, ChatGPT returned results from PlanetScale, PingCap, Baeldung, and GeeksForGeeks. Meanwhile, Google’s top 5 results for the same query in chronological order was:  PlanetScale, Hazelcast, AWS, GeeksForGeeks, and SingleStore.

I don’t have all the answers yet, but one thing is clear: if we’re moving away from SEO dependency, we need to rethink how we create content.

Product-Led Content is what your content marketing efforts is missing.

We’ve established that top-of-funnel informational content—things like “What is serverless computing?”—is driving less traffic to websites. People are finding answers faster elsewhere, whether through AI chatbots, Google’s AI Overview, or directly from communities and forums.

This shift could actually be a blessing in disguise for content marketers. For years, SEO-driven strategies have forced teams to churn out broad, educational content that often had little to do with their product, all in an effort to capture as much traffic as possible. But traffic for its own sake has never been the goal. The real goal is attracting high-quality leads that convert into customers. If content doesn’t contribute to revenue through brand trust, affinity, or direct conversions, it’s just noise.

If AI can generate it, it’s probably not worth writing.

The kind of content that still works today is the kind of content that AI can’t generate and Google can’t summarize in a snippet. It’s Product-led content.

Product-led content is content that naturally integrates the product into valuable, educational, or thought-provoking narratives. It’s not about forced mentions or awkward sales pitches. It’s about making the product an organic part of a compelling story.

Product led content done right has the following capabilities:

• It’s tied to the product. The topic is closely aligned with a feature or problem your product solves, so mentioning it feels seamless rather than forced. By the time a reader finishes the article, they’ve learned something valuable and seen how the product fits into the solution. Even if they don’t sign up immediately, they leave with a positive impression. Fly.io does this brilliantly—every blog post is an engaging, technical deep dive that subtly integrates their platform without feeling like a pitch.
• It’s value-driven. The content delivers insights, workflows, or technical knowledge that the audience can apply, whether or not they use the product. Developers, for example, are naturally skeptical of marketing fluff but love deep technical content. Show them how you solved a hard problem. Share your mistakes, your trade-offs, and what you learned. Teach them something they can take away, independent of your product. That’s what earns trust.
• It’s behind-the-scenes and personal. People love to peek behind the curtain. Share how your team tackled a complex problem, the internal debates that happened along the way, and the trade-offs you made. The best way to structure this is with “How We Built It” stories that walk through real engineering challenges and decisions. Even announcement blog posts can follow this structure, making them engaging rather than self-promotional.
• It’s collaborative. The best product-led content doesn’t come from marketing alone. It comes from working closely with engineers, designers, and product teams who actually build and use the product every day. Without this collaboration, marketing risks producing shallow content, while engineering risks building features that never get talked about. A regular content brainstorming session with product and engineering ensures that what gets written is both technically accurate and deeply useful.

At work, we evaluate our content ideas through the simple grading system below to make sure we’re prioritizing the right kinds of stories:

• 0: There’s no natural way to mention the product without forcing it.
• 1: The product is relevant but not essential to the narrative.
• 2: The product is useful and integrated but not the only solution.
• 3: The entire post hinges on the product—it’s central to the story.

If search-driven traffic is shrinking, the answer isn’t to panic. The answer is to double down on content that drives real conversations.

Final thoughts and more pondering

Now that I’ve said we should focus on creating content that get people taking, maybe we should all also be thinking of retaining those engagements. How do we keep those conversations happening within our own ecosystem?

A few things have been on my mind:

Should we bring back comment sections on blogs?

At some point, most companies decided that managing blog comments wasn’t worth it. Spam, moderation, and low engagement led to a gradual shift away from them. But should we rethink that? In the past, comment sections didn’t just encourage discussion—they kept readers coming back.

People who left comments often returned to check replies, fostering deeper conversations within the company’s platform. Today, those discussions have moved to Hacker News, Reddit, and X. But what if we could reclaim some of that engagement? What if companies focused on building relationships inside their own ecosystem instead of relying entirely on external platforms?

Should companies create AI-powered search agents trained on their own content?

As AI-driven search habits become more common, it may be time for companies to take control of that experience. Instead of waiting for Perplexity or ChatGPT to surface their content, companies could train AI chatbots on their documentation, blog posts, and knowledge base. Users could query these AI agents for installation guides, troubleshooting steps, or best practices and get direct, authoritative answers from the source. This would not only improve user experience but also align with how people now expect to consume information by getting a single, definitive answer instead of sifting through multiple sources.

I don’t have all the answers, but I’d love to hear your thoughts about everything I’ve written.