2025 was the year everyone rushed to slap “AI” onto their LinkedIn headline. 2026 will be the year founders and marketers collectively hit unsubscribe.
The hype cycle peaked, the inbox noise became unbearable, and “AI content expert” turned into the new “growth hacker”: a label almost no one trusts anymore.
For all the noise, most of these experts delivered… nothing. No compounding growth, no durable rankings, no meaningful revenue lifts. Just louder promises and lower-quality content published at scale.
And the market finally noticed.
It Wasn’t AI — It Was the Humans Using It
The industry spent the year firefighting the consequences of fast content. One mis-architected prompt quietly duplicated itself across hundreds of pages. One technical mistake pushed entire clusters under filters. One oversight required weeks of manual cleanup.
Teams were drowning. Too many tools. Too much noise. Too many “strategies” that turned into liabilities after the next core update.
The companies that survived weren’t the ones generating the most content. They were the ones validating every trend through real technical and SEO experience, filtering hype from signal, and placing long-term bets that wouldn’t collapse with the next algorithm tweak.
Our Story Looked Very Different
Much of the ecosystem was staring at crocodile jaw graphs in Search Console and watching impressions collapse (along with the realization that many of those impressions were just bots.)
On my projects, clicks went up across many pages. This happened even in a niche where our competitors aren’t cute little startups — they’re Atlassian, Monday.com, Asana, and other enterprise-grade giants. The kind of competitors that usually flatten smaller sites by simply existing.
I didn’t cheat, exploit loopholes, or spin up 10,000 robotic blog posts. I did what actually works in the long run. Here’s what made that growth possible.
What Actually Worked in 2025 (or What the Hype Machines Ignored)
1. Clustering and internal linking
We didn’t dump topics into a blog but architected clusters that drove authority. Used internal links to consolidate ranking power. Let me explain how.
If your site has 50 posts about "project management" but they don't reference each other, Google sees 50 weak signals. By clustering, you create a "knowledge graph" that mirrors how LLMs understand relationships. The goal is to maximize PageRank flow: preventing authority from getting stuck on your homepage and forcing it deep into your money pages.
So, first of all, we ran a crawl using Screaming Frog, specifically filtering for pages with < 3 unique in-links. Every new post must link back to at least 3 older posts, and 3 older posts must be updated to link to the new one immediately upon publishing.
Then, we designated one "Parent" page for every cluster and linked the Child posts back to the Parent with matching anchor text.
And of course, every topic passed a “why does this deserve to exist?” test.
2. A product Knowledge Base — launched before it became cool
One year ahead of the trend, we built a fully optimized KB. It quietly captured long-tail queries, fed chatbots with correct product answers, and introduced new users to the product before they even hit the website.
Marketing blogs are often too "fluffy" for the high-intent queries (like "how to integrate X with Y"). Knowledge Bases naturally align with the answer-engine evolution of search. Furthermore, KBs have high "dwell time" for users actually solving problems, a behavioral signal that correlates with ranking stability. We stopped treating docs as "support" and started treating them as "bottom-of-funnel acquisition.
It became one of our biggest SEO wins of the year.
3. Updates with real product insight, not AI filler
Google's core updates in 2025 focused heavily on information gain — a concept where content is ranked by how much new information it adds to the existing index. If an AI summarizes the top 10 results, information gain is zero. To rank, you must provide data that an LLM cannot possess because it isn't in its training set yet: proprietary data, new feature releases, or subjective human experience.
What we did is, before publishing, we pasted our draft into an LLM alongside the top-ranking competitor and asked: "What specific data point or opinion is present in my text that is missing from the competitor?"
If the answer was "nothing," the draft was rejected.
We linked every blog post's author byline to a verified LinkedIn profile and a specific "About the Author" page populated with verifiable credentials to satisfy E-E-A-T signals.
Finally, the product team was a goldmine: new features, screenshots, customer questions, and internal docs. We turned this into editorial-quality content that both algorithms and humans actually trust.
4. AI-powered localization into four languages
English SERPs are a Red Ocean — saturated and expensive. Non-English SERPs (DE, FR, ES, PT) are often 2-3 years behind in terms of competition intensity. By localizing, you increase your surface area of luck.
That wasn't a translate and pray approach. We published our content strategically into four geos and tracked gains.
Blind machine translation creates trust issues. The strategy relies on human-in-the-loop workflows where AI provides volume, and humans provide the "trust signals" (idioms, cultural nuance).
SEO matters in this case as well. Instead of relying on on-page tags which can break code, we managed localization via a dedicated XML sitemap that mapped every URL to its 4 language counterparts.
5. Community posting that fed the internet’s other brains
Google and LLMs started inserting Reddit and other niche communities' threads directly into their answers. By planting high-quality answers in these communities, you are essentially performing inception on future AI models, training them to associate your brand with specific solutions.
We published solution-driven content in business communities where real buyers spent their time. In our case, we posted on the Atlassian community and provided a complete solution even without linking to our site. Visibility > link placement.
6. Smart automation of internal research and content prep
Quality content requires connecting disparate dots (like a sales call from June, a product update from August, and a competitor change from October). Humans are bad at recalling vast archives of data instantly. AI isn't.
So, we configured a Gemini Gem with our specific brand voice guidelines ("Use active voice," "No Oxford comma," "Grade 8 reading level"). Every human-written draft was passed through this Gem for a final "compliance check" before formatting.
Then, we created a private source of truth chat in NotebookLM. We uploaded some proprietary information about the product that contains all details and pain points. When starting a new piece, we prompted: "Based on source documents 1-10, list the top 5 objections enterprise clients have regarding security."
But (and this is critical) humans did the actual editorial work. Because by mid-2025, even “high-quality AI writing” started to feel the same. Readers feel it, and their trust drops. The winning strategy became hybrid.
Summary
Overall, companies that leaned into AI-driven promotion of their alternative products were the ones that kept revenue stable or even grew.
If your product was hit and you're trying to figure out how to reposition, re-rank, or re-route traffic, send me a message. I’ll walk you through possible next steps — free of charge.
Blind AI optimism is over. Smart, validated, technical AI execution is our future. Join those who are at the forefront of change!