How Search Changed — A Timeline

1990s 📂 Keyword matching — the search engine as a card catalogue ↓

Early search engines (AltaVista, Lycos, Yahoo) worked exactly like a library card catalogue. You typed words, the engine looked for those exact words on web pages, and gave you a list of pages that contained them. No understanding, no context — just matching strings of letters.

Your search experience

You type: "running shoes flat feet"

Engine finds pages that contain those exact words

You get a list of hundreds of links — most of them irrelevant

You click around yourself until you find something useful

💡 The engine had no idea what "flat feet" meant. It just found the words. A page about shoe factories in China that happened to mention "flat" and "feet" could show up at the top.

1998 🔗 Google launches PageRank — links become votes ↓

Google's big idea was simple: if lots of other websites link to a page, it's probably trustworthy. Like being the most cited author in a library — it doesn't mean you wrote the best book, but it's a good signal. This made results much better than random keyword matching.

Your search experience

You type: "best running shoes flat feet"

Google shows the pages that other running websites link to most

Runner's World and REI rise to the top because they're well-cited

You click one, read the article, maybe find your answer

💡 Still a list of 10 blue links. Still required you to click through and read. The engine knew which pages were popular — but not what you actually needed.

2007 🖼️ Universal Search — images, news, and videos appear in results ↓

Google started mixing in images, news, videos, and maps alongside the regular links. Results became richer, but still required you to click and explore.

Your search experience

Now you also see images of shoes and a YouTube video review

More options — but still no direct answer

💡 More formats, same principle: Google points you somewhere. You do the reading.

2012 🧠 Knowledge Graph — Google learns about things, not just words ↓

Google built a giant database of facts about the world — people, places, brands, products — and the connections between them. Now when you searched for a brand, Google actually "knew" what it was, not just what pages mentioned it.

Your search experience

Search for "Brooks Running" → a panel appears on the right with brand facts

Google knows Brooks is a shoe brand, not a person named Brooks

First time a small answer appears without clicking anything

💡 The engine started understanding things. "Running shoes" wasn't just two words — it was a product category with brands, features, and relationships.

2013 💬 Hummingbird — Google starts understanding full sentences ↓

Before Hummingbird, you had to "talk like a search engine" — short keywords, no grammar. After it, you could type a proper question. Google now understood the meaning of a full sentence, not just individual words.

Your search experience

Before: you'd type "running shoes flat feet"

After: you can ask "what are the best running shoes if I have flat feet?"

Google understands the question, not just the words in it

💡 A big shift for users. A huge shift for websites — they now had to answer questions properly, not just stuff keywords into pages.

~2014 📦 Featured Snippets — the first "answer boxes" ↓

Google started pulling one short answer directly out of a web page and showing it at the very top of results — no click needed. This was the first time search seriously started answering instead of pointing.

Your search experience

You search "best shoes for flat feet"

A box at the top says: "People with flat feet need motion control or stability shoes. Top picks include Brooks Adrenaline, ASICS Kayano…"

You got the answer without clicking anything

💡 Zero-click searches started here. Great for users. Terrible for websites that lost traffic even though their content was being used.

2018 🔬 BERT — Google reads context, not just keywords ↓

BERT is an AI model that reads words in context — understanding that the same word means different things in different sentences. "Running" in a fitness article is completely different from "running" in a political one. Google finally got this.

Your search experience

You search "do I need extra support for flat feet when running?"

Google understands "support" means structural shoe support — not customer support, not emotional support

Results are suddenly much more accurate for complex questions

💡 This is when keyword-stuffing stopped working. Google now understood what a page was about, not just what words appeared on it.

2022 🤖 ChatGPT launches — search becomes a conversation ↓

OpenAI released ChatGPT and 100 million people signed up in 2 months. For the first time, people could just ask a question in plain English and get a direct, written answer — no links, no clicking, no reading through 10 pages. The world realized: this is how search should feel.

Your search experience

You type: "I have flat feet and I want to start running, what shoes should I get and why?"

ChatGPT writes you a 3-paragraph answer explaining overpronation, motion control, and 3 specific models

You follow up: "Which one is best for trails?" — it remembers what you asked

Zero links. Zero clicking. Just a conversation.

💡 Google's 25-year head start suddenly felt irrelevant. The question wasn't "what page has the answer?" — it was "can the AI just tell me?"

2023 🔭 Perplexity launches — AI search with sources shown ↓

Perplexity combined ChatGPT-style answers with live web search and transparent citations. Instead of just giving you an answer, it showed you exactly which websites it pulled from — making it more trustworthy than a pure AI chatbot.

Your search experience

You ask your shoe question

Perplexity writes a direct answer AND shows you: [1] Runner's World [2] REI Blog [3] Podiatry Today

You can verify if you want, but most people don't

💡 The first real signal that traditional Google-style search had a serious competitor. And that websites could lose traffic even while being cited as sources.

2024 🌐 Google AI Overviews — the answer box goes mainstream ↓

Google added AI-generated summaries at the very top of search results for more than half of all searches. A paragraph of generated text, with a few source links, now sits above all the traditional results. 2 billion people worldwide see this every day.

Your search experience

You search "best running shoes for flat feet" on Google

Before you even see any links, a blue box says: "People with flat feet tend to overpronate. Motion control shoes like Brooks Adrenaline GTS or ASICS Gel-Kayano provide the stability needed…"

The answer is right there. The traditional links are pushed way down

Only 1% of users click the source links inside the AI summary

💡 Click-through rates on #1 organic results dropped 34.5%. Your content can be the source of the answer while your website gets zero visitors. That's the new reality.

2025 ✨ AI Mode — the whole search experience becomes a conversation ↓

Google launched AI Mode — where the entire search interface is a chat. No more 10 blue links as the default. You ask a complex question, Google reasons through it across multiple sources at once (this is called "query fan-out"), writes you an answer, and you can keep the conversation going.

Your search experience right now

You ask: "I have flat feet, I'm a beginner runner, I want to train for a 10k in 3 months, what shoes AND training plan should I follow?"

Google breaks this into 8 sub-questions internally and searches all of them at once

You get: a shoe recommendation, a 12-week training plan, a weekly mileage table, and an injury-prevention tip — all in one response

You follow up: "Which shoes are available at stores near me?" — it remembers your flat feet and your location

Fewer than 4% of AI Mode users click any external link

💡 This is where GEO lives. The question is no longer "how do I rank #1?" — it's "how do I get included in the AI's answer?"

Soon 🤝 Agentic search — the AI doesn't just answer, it acts ↓

The next stage isn't just answering your question — it's completing the task. AI agents will search, compare, decide, and buy on your behalf. You give the goal, the AI does the work.

Your search experience — coming soon

You say: "Find me the best shoes for flat feet under €120, check if they're available in my size, and order them"

The AI searches, compares, finds your size at a local store, checks your saved payment method, and completes the purchase

You never opened a browser, visited a website, or clicked anything

💡 For brands, this is the most radical shift yet. The AI becomes the customer. You're no longer marketing to humans — you're marketing to algorithms that shop on their behalf.

2030? 👓 Embodied & ambient search — AI in your glasses, your home, your world ↓

Search stops being something you do and becomes something that happens around you. Your glasses see what you see. Your assistant anticipates what you need before you ask. The "search box" disappears entirely.

Your search experience — 2030

You walk past a running shop window. Your glasses recognize the shoe in the display

Without you asking anything, your AI says: "That's the Brooks Adrenaline GTS 25. Good fit for your flat feet. It's €15 cheaper on their website right now."

You say "order it" and it's done

💡 Search has become invisible. Your AI knows you, watches the world with you, and acts on your behalf. The concept of "searching" becomes as dated as going to a physical library.

How search went from a list of linksto an AI that just answers

How search went from a list of links
to an AI that just answers

Online search evolution 1990 - 2030