Making Sense of SEO in 2025
What’s Changing, What Isn’t, and What to Do About It
There are times when working in SEO feels like chasing a moving target, and lately, it’s been moving faster than ever. Between the rollout of AI features, shifts in search behaviour, and the constant pressure to prove performance, it can be hard to tell whether you’re staying ahead or just trying to keep up.
In July this year (2025), I attended Google’s Search Central Live: Deep Dive Asia Pacific 2025, a three-day event held in Bangkok, packed with insights from the Search Relations team, technical engineers, and regional marketing leads. I didn’t go in expecting breakthroughs. I went in hoping to make sense of the noise and feel a little more grounded in the work.
What I took away was a deeper understanding of how search is evolving, what still matters (and what doesn’t), and how to adapt. Much of what I heard aligned with my current approach, but there were moments that made me rethink how I frame SEO for clients, students, and stakeholders alike.
Below are five takeaways that stood out, not because they were new, but because they helped me make sense of where we are now, and where we’re going next.
1. AI Features Are Built on Search, Not Separate From It
There’s been a lot of hype about SEO needing to reinvent itself for AI. And if you’re anything like me, you’ve probably wondered: do we really need a whole new playbook? AIO? GEO? Something else entirely?
That question was one of the first addressed at Search Central Live and the answer was surprisingly reassuring. In Gary Illyes’ words:
“If you know how search works, you will understand how AI features work.”
That one statement confirmed what I’d been sensing for a while: AI Overviews are not a separate ecosystem. They are features - like featured snippets or image packs - that draw from the same index, same crawl infrastructure, and same search signals as traditional blue links. Googlebot still crawls, robots.txt is still respected, and AI Mode still relies on the same index we optimise for today.
That said, the way content is served within AI-powered experiences is changing. One key concept discussed was query fanout, where Google issues multiple, related versions of a user’s query behind the scenes, then reconstructs a response using the most relevant results across those variants. This expands the scope of what your content might appear for but also demands clarity, structure, and intent alignment.
You don’t need to overhaul your strategy for AI, but you do need to ensure it’s built for how search works today. Over-reliance on keyword optimisation, without aligning to user intent or content quality, may struggle to keep up. For those already focused on creating useful, well-structured content and maintaining technical best practice, it’s more of a gentle shift than a major change.
As Gary pointed out, “Past success on Search may not guarantee future success.” It’s a useful reminder that staying visible means staying proactive, not holding onto what once worked.
One particularly challenging area right now is measurement. Daniel Waisberg spoke candidly about the difficulty in quantifying performance from AI surfaces. Impressions may not be visible. Clicks may not happen. Zero-click experiences are increasing. And traditional metrics often don’t capture the true value being delivered.
“Each person measures in a different way,” Daniel said.
“If you use clicks, the problem is deeper. What happens on Search is not what happens on your website. That’s the challenge; connecting the data in a meaningful way.”
He also pointed out that current AI features aren’t stable enough for Google to provide consistent metrics: results vary too much from day to day. They’re working toward better reporting, but right now, we need to accept a level of ambiguity and focus instead on how well our content connects with real people.
There was also discussion around LLM.txt, the proposed standard for AI crawlers. Google won’t be supporting it. Their concern is that it can be easily manipulated or spammed. Instead, they’re exploring improvements to robots.txt to allow publishers to set more nuanced controls, not just for access, but also for usage. That could be a space to watch in the months ahead.
Ultimately, AI features are just that - features. Like image packs or featured snippets, they exist within a much broader search experience. And search itself remains what it’s always been: complex, constantly evolving, and never “solved.”
SEO isn’t becoming something new, it’s deepening. And if your strategy is built on helping users, not just targeting terms, you’re already aligned with where search is heading.
2. Search Behaviour Has Changed, And So Must Our Content Planning
I’ve known for a while that search behaviour was shifting. But what stood out to me during this event was the scale, speed, and scope of that change, and how far it’s moved beyond what most strategies are built for.
Search queries are getting longer, more conversational, and more complex. Five or more words per query is becoming the norm, not the exception. But it’s not just length, it’s expectation. People assume that Google can handle nuance, anticipate their follow-up questions, and deliver not just a single answer, but context, guidance, and clarity. And Google is rising to meet that expectation.
They’re making search more effortless than ever, but also more layered. Users can now search with natural language, visual input (via Lens), spoken prompts, and persistent follow-up suggestions. Multimodal search isn’t some abstract future concept - it’s here. And it’s reshaping what it means to be visible in organic search.
That means we’re no longer just writing for standalone queries. We’re contributing to a broader discovery process, one where our content needs to predict, support, and evolve with the user’s next move. Google’s features are designed to accommodate that, issuing query suggestions, surfacing deeper answers, and stitching together insights from a variety of sources. That’s the level of depth we need to plan for.
The rise of visual content is also playing a much bigger role than some of us may realise. In Southeast Asia alone, over 290 million people are YouTube viewers, and more than 40% of shoppers rely on video to make informed buying decisions. It’s clear that video is no longer just complementary to the search experience, it is the search experience for many users. And Google has fully integrated it into SERPs, rewarding content that can engage and retain attention.
The implications for SEO are significant. We need to diversify the types of content we produce. We need to include formats like video, images, and content designed for voice-based queries where appropriate. We need to consider where content appears on the page, how quickly it engages, and whether it supports the user’s next action, not just the one that brought them in.
We also need to stop thinking of the customer journey as a sequence of clean steps. The funnel metaphor is being replaced by what one speaker aptly described as “a mess, not a funnel.” People move fluidly between platforms, questions, and formats, often circling back or jumping ahead. That means our content needs to do more than guide discovery. It also needs to support comparison and decision-making, and reinforce the value of those decisions once they’re made. That last part is especially important; users often keep searching after converting, looking for confirmation that they’ve chosen well. Content that sets expectations, highlights benefits, or deepens trust plays a crucial role in keeping them engaged beyond the click.
I came away with a stronger sense that the role of content isn’t just to answer queries, but to support curiosity. We need to guide, inform, and enable users throughout their journey, not just show up at the point of conversion. That means predicting questions, embracing a mix of formats, and recognising that discovery doesn’t just happen in Google’s blue links anymore. It happens in videos, in AI-assisted snapshots, in social feeds, and in follow-up prompts.
We can’t afford to be one-dimensional in how we plan content. Because the way people search isn’t one-dimensional either.
3. You Can Use AI, But It Can’t Replace You
I’ll admit I went into this event half expecting Google to draw a harder line on AI-generated content. With so much low-effort material flooding the web, it would have been easy for them to say, “Don’t use it.” But the message from Search Central Live was far more pragmatic: AI can support SEO performance, as long as quality remains the priority.
Google made it clear that they’re not trying to distinguish AI-written from human-written content. Their systems are trained to evaluate signals of quality - effort, originality, accuracy, and skill - regardless of who or what created the content. That nuance really stood out to me. It reframed the conversation away from the how and back to the what: is the content genuinely helpful, trustworthy, and built for people?
This mindset matters, especially as AI becomes a more regular part of how we work. The speakers encouraged us to embrace AI as a tool that can improve workflows, simplify management, and support content creation at scale. Used thoughtfully, it can reduce friction across teams and make it easier to adapt content across channels. That’s particularly relevant for those of us managing SEO in fast-paced or resource-limited environments.
One speaker also shared that when AI-generated content is carefully reviewed and tailored to the user, engagement can match - or even outperform - traditional human-led processes. Custom models trained on brand-specific data offer even greater potential.
But that doesn’t mean we can hand off the responsibility. AI still requires human oversight. It still hallucinates. If we’re generating briefs, articles, or site content, we need to be sure we’re not adding to the growing volume of generic material that lacks substance. It’s not the tool that determines quality, it’s how we guide it, refine it, and review it.
That goes for imagery too. You’re free to use AI-generated visuals, but the same question applies: does this genuinely add value for the user? If the answer’s no, it may be worth leaving it out, or at least ensuring it’s not indexed.
This takeaway helped clarify something I’ve been wrestling with: AI can absolutely make life easier, but it’s our job to ensure the end result still reflects relevance, expertise, and intent. When used well, AI becomes an assistant, not an excuse to cut corners.
4. Indexing Is Complex, And May Be Your Greatest Barrier
This wasn’t a brand-new topic for me, but over the course of the event, I picked up more precise language. It helped deepen my understanding of how Google processes web pages it has crawled and gave me fresh ways to explain these concepts to clients, students, and mentorees.
Gary Illyes summed it up well:
“Indexing is way more complex than we give it credit.”
Most people - even SEOs - tend to treat indexing as a single step: a page gets crawled, then indexed. But that overlooks the many invisible processes happening behind the scenes. Parsing, structuring, tokenising, mapping; and that’s before you even get to ranking.
One of the most important takeaways was just how unforgiving this system is. HTML might appear fine in a browser, but that doesn’t mean a search engine can understand it. Browsers are designed to be flexible and display content as best they can, but crawlers need structure. They attempt to interpret messy HTML into something usable by generating a Document Object Model (DOM), identifying key sections like navigation, headers, and main content. If those areas aren’t clearly defined, or if key content is buried in the wrong section, or hidden behind JavaScript, it can undermine your visibility from the start.
That structure also affects how weight is assigned to your words. Google doesn’t treat every part of the page equally. Content in prominent areas carries more influence. If something’s important, it needs to appear in the right place.
Another detail I found useful was the reminder that Google uses the content itself to determine the page’s language, not just html tags. That means mixing multiple languages on the same page can cause confusion and affect how the page is processed. It’s something I’ll be more mindful of when reviewing international or multilingual sites.
From there, the discussion moved into tokenisation and why raw HTML can’t just be “indexed.” It needs to be broken down into tokens: the individual words and short phrases that form the searchable parts of your content. These tokens are then cross-referenced with a posting list, a fundamental structure that maps each token to the URLs that contain it. This allows Google to retrieve relevant pages at speed and scale.
What stood out here was how central content remains to this process. If there’s nothing meaningful to tokenise, there’s nothing to serve. The quality and clarity of your content are what enable search engines to match your page with relevant queries.
Finally, we looked at the synonym system, something that’s often invisible but incredibly powerful. It’s designed to bridge the gap between the language people use in queries and the language used in content. But it doesn’t behave like a thesaurus. These synonyms are context-dependent and generated from the surrounding words in the query. They don’t always “read well” to a human, and that’s okay. Google isn’t aiming for elegance, it’s optimising for relevance.
This deeper technical understanding didn’t change what I believe about SEO, but it gave me stronger language and more precise tools to explain why it works the way it does. I’m walking away with a renewed appreciation for what happens behind the scenes and a clearer sense of how to communicate that to others.
5. SEO Isn’t Dead! (But It Might Be Due For A Rebrand)
If you’ve worked in SEO for any length of time, you’ve likely encountered the same alarmist headline on repeat: “SEO is dead.” It surfaces with every core update, every rise of a new platform, and now, with the accelerating presence of AI in the search landscape. And I’ll be honest, over the past few years, I’ve caught myself wondering: Is there still a place for this work?
Trying to keep up has been exhausting at times. The way we talk about search has changed. The tools we use are changing. Even the way people find information - across Google, social platforms, AI tools, and niche apps - is evolving. I’ve asked myself questions that I know many others are grappling with too. Am I being replaced by automation? Will search engines become irrelevant altogether?
What I appreciated about this event was that it didn’t shy away from those tensions. Instead, it offered something far more useful: perspective. Gary Illyes said it with a wry smile, but it landed:
“SEO tends to die a lot. It’s died every month since 1997. And yet, here we are.”
And more importantly, here’s why we’re still here. As Liz Reid, VP of Search, reminded us:
“Search is never a solved problem. Old challenges evolve and new challenges are constantly popping up - because the internet and the world are always changing.”
It’s easy to feel overwhelmed when you’re focused on rankings and results. But what I took away was a much-needed shift in mindset: we’re no longer optimising just for position. We’re designing for presence.
In the past, SEO visibility was often tied to a single ranking or blue link. Today, it’s about showing up across formats and surfaces; from AI overviews and image packs to Discover, Maps, and YouTube. This means shifting our focus from isolated tactics to integrated strategies that align with how people actually navigate information.
That shift feels bigger than just adapting to new tools. It feels like an evolution in the role itself; from SEO as a channel to SEO as a connective layer across content, product, and digital experience. We need to see beyond search engines and think more holistically about organic discovery. It’s time to broaden the conversation and, perhaps, rebrand what this work truly represents.
The event helped me step back from the noise and recentre on the value we bring; helping people find what they need, and helping businesses show up with clarity, context, and purpose.
Just as importantly, it reminded me I’m not the only one asking these questions. Hearing others reflect on the same challenges, as well as their solutions, was one of the most valuable parts of the event.
Conclusion
The more I reflect on those three days in Bangkok, the more I come back to this: the fundamentals of SEO are still here, but they’re sitting within a much broader, faster-moving environment. It’s not about throwing out what we know. It’s about sharpening it, stretching it, and applying it in ways that make sense for how people search today.
Each takeaway from the event added another layer to that picture; from the infrastructure behind AI features, to the rise of multimodal search, to the very language Google uses to index and serve results. Some ideas challenged my assumptions. Others helped me articulate things I’d already been sensing. But all of them pointed in the same direction: we need to build for adaptability, not certainty.
That’s not always easy, especially when you’re managing SEO as one part of a much broader digital role. But it’s also what makes this work so rewarding. Search isn’t static, and neither are we.
What this event offered wasn’t just reassurance, it was practical clarity. A reminder that presence matters more than position. That quality is still at the core. And that while technology is changing fast, our job is still very human: to help people find what they need, when they need it, in a way that genuinely serves them.
And for anyone else feeling the pace, or wondering if the role of SEO still holds its place - you’re not alone. We’re all adapting. We’re all learning. And we’re all still showing up.
