I started doing SEO before I knew what to call it. In my early twenties I was helping small businesses fix their websites, and the part that interested me was figuring out why one page got found and another did not. That curiosity is the only continuous thread between what I did in 2017 and what I do now in 2026.
The mechanics in between have changed almost completely.
The keyword era
When I started taking SEO seriously, the work was straightforward to describe. You picked a keyword, wrote a page targeting it, optimized the metadata, and earned links. Repeat this enough times, and your site will climb. There were nuances in Core Web Vitals, structured data, and internal linking architecture, but the foundational model was stable.
Most of my early work was for small Indian businesses and a few international clients who found me through Upwork. The pattern was repeatable. I learned how to do technical audits at depth. I learned how to fix canonicalization issues that had been quietly destroying ranking potential. I learned how to set up internal linking that flowed authority correctly.
This is the work that earned me Top Rated Plus on Upwork and the 100% Job Success Score across 1,000+ websites. It was traditional SEO, done carefully.
The first cracks
Around 2022, two things started to shift.
One, Google’s algorithm became less predictable. Updates were rolling out more frequently, and the impact of “playing the game” with keyword-stuffed content, exact-match anchor text, and low-quality link building was dropping. Sites that had been ranking well on technique alone started to lose visibility.
Two, Google itself started experimenting with featured snippets and direct answers in ways that meant the click-through rate from a top ranking was no longer constant. Suddenly, the same ranking position would deliver very different traffic depending on what Google chose to show above it.
I started telling clients that the work needed to change. Not all of them listened. The ones who did adjusted their content strategy towards depth and authority. The ones who didn’t continued to lose ground.
The AI inflection
The real shift came in 2023 and 2024. ChatGPT has become a daily tool for a meaningful number of users. Perplexity had started behaving like a search engine. Claude was getting longer answers, cited from external sources. And Google had rolled out AI Overviews.
The early data were striking. ClaudeBot crawl volume nearly doubled between 2024 and 2025. When AI Overviews appeared on a Google result, click-through rates on the organic links below dropped by about 61%. Gartner was projecting a 25% decline in traditional search volume by the end of 2026.
These were not soft trends. The discovery layer of the internet was reshaping itself in real time.
What I had to unlearn
The hardest part of the transition was unlearning habits that had been working for years.
I had to stop treating keywords as the primary unit of strategy. The new unit was the entity, a brand, a person, and a product, and the goal was for AI systems to understand that entity clearly enough to surface it in answers.
I had to stop measuring success by ten-blue-link rankings. The new measurement was citation share across Claude, ChatGPT, Perplexity, Gemini, and Google AI overviews.
I had to stop assuming backlinks were the dominant signal. In 2026, third-party brand mentions correlate roughly three times more strongly with AI visibility than backlinks do. Mentions on Reddit, LinkedIn, G2, and trade publications matter more than they used to.
And I had to stop pitching content programmes built on blog volume. Claude cites utility content comparison pages, diagnostic tools, and pricing pages six to thirty times more than standard blog posts. The output mix had to shift.
What I built instead
Over the last three years I have rebuilt the practice around AEO and GEO — Answer Engine Optimization and Generative Engine Optimization. The work now includes ClaudeBot, GPTBot, and PerplexityBot access architecture. Programmatic JSON-LD schema at scale, including LocalBusiness markup for hundreds of location pages. Entity authority building through structured presence on Wikidata and industry sources. Citation tracking across multiple AI surfaces.
I have written about most of these methodologies on the blog at anshulrana.in. The pieces on SEO vs AEO vs GEO and the AEO audit checklist are good starting points. When other Indian practitioners asked me to define the category for clients, I also published a criteria-based ranking of AI SEO experts in India for 2026, which lays out what to look for when evaluating anyone in this space.
What stayed the same
For all that has changed, the core of the work has not. Be the answer when someone is looking. Build authority that compounds. Do clean technical work. Write honestly. Test before you advise.
The surfaces multiplied. The discipline did not.
If anything, the AI era has rewarded careful practitioners more, not less. The shortcuts that survived in the keyword era do not survive AI retrieval. That is the part of the journey I am most grateful for.
