Nerd Nite: An A-Z of Search Behaviour

Nerd Nite: An A-Z of Search Behaviour

Event Recap

On Wednesday evening (2nd September 2026), I spoke at Nerd Nite Christchurch, held at A Rolling Stone. Nerd Nite is a bit different from the usual industry events I speak at. It's a room full of curious people from every background imaginable, not marketers, not SEOs, just people who like learning weird, interesting things over a drink. So my brief to myself was simple: take everything I know about search behaviour, strip out the jargon, and find the bits that would actually land with someone who has never thought about "organic search" in their life.

The talk was called Google Knows You Better Than Your Mum: The A-Z of How We Search, and it was exactly what it sounds like: a tour through the data, psychology, and quirks behind how we all search, with a couple of live demonstrations along the way.

I've spent close to twenty years working in search, and I still don't think most people, including plenty in my own industry, appreciate quite how much has shifted in the last eighteen months, or how much of it is happening quietly, without anyone opting in. So rather than a slide deck full of client-facing SEO advice, this was a chance to just show people the shape of it.

More Places to Search, But Google Still Wins

I started with the question everyone in my industry loves to argue about right now: is AI actually replacing search? Short answer: not yet, not by a long way.

Here's the scoreboard, as it stands:

  • Traditional search engines still account for roughly 80% of all searches (the rest split between commerce platforms, social networks, and AI). [SparkToro, March 2026]
  • Google holds 89% of the New Zealand market, gradually declining, but the platform picking up the difference isn't ChatGPT. It's Bing. [StatCounter, July 2026]
  • Google processes roughly 210x more searches than ChatGPT worldwide.
  • ChatGPT's New Zealand market share sits at just 0.25%.

Bing's own story is a nice example of a number that looks bigger than it is: it crossed 1 billion monthly active users this year, which sounds enormous until you realise it's still only handling 13-14 billion searches a month, against Google's 200 billion-plus. I'd put most of that billion down to Microsoft Edge defaulting to it, something I see constantly in enterprise environments running the Microsoft suite, rather than people genuinely choosing Bing.

None of this means AI isn't growing. It absolutely is, just from a much smaller base. ChatGPT passed 900 million weekly users in February 2026, then crossed 1 billion monthly app users by June, the fastest app in history to hit that milestone. Google's own Gemini, second-highest AI market share here in New Zealand, passed 1 billion monthly users just last month. And 63% of those users are now talking to it rather than typing. Gemini's own data shows busy parents are 43% more likely to use voice for everyday tasks, which, if you've ever tried typing one-handed while holding a toddler, makes complete sense.

Beyond Google: TikTok, Reddit, and the Rise of "Alternative" Search

One of the clearest signals that search is fragmenting, not just shifting to AI, is where people are actually going for answers. Google's own Head of Search, Liz Reid, said as much last year: people are increasingly going to short-form video, forums, and user-generated content, rather than traditional websites.

Reddit's growth backs this up nicely: 80 million weekly searches in Q4 2025, up from 60 million the year before. Its AI Q&A feature, Reddit Answers, grew even faster, from roughly 1 million queries to 15 million in the same window. TikTok's telling a similar story. Adobe found nearly half of US consumers (49%) used TikTok as a search engine in 2026, up from 41% in 2024, drawn especially to its short-form, tutorial-heavy format.

Travel is where this shows up most vividly, and it's worth breaking down properly, because the two halves of the story pull in different directions:

  • TikTok claims 66% of its users say the app is their most helpful source of travel inspiration, and that people are 2.6x more likely to book after searching there.
  • But when it comes to actually planning travel, a separate survey of 1,000+ US consumers found Google is still the default first stop, at 57%, with ChatGPT trailing at just 18%.
  • Meanwhile, traffic from AI tools to travel sites specifically is up 194% year-over-year, and over 2,000% since Adobe began tracking it less than two years ago.

So: TikTok wins the daydreaming. Google still wins the actual planning. And AI is the fastest-growing channel of the three, even if it's starting from the smallest base. AI-referred travellers are also 21% more engaged and spend 70% longer per visit; they're just not converting into bookings quite as well yet.

We Don't Trust It, and We Use It More Anyway

This is the contradiction that sits at the heart of the whole talk. 70% of consumers say they're using AI tools for search more than they were a year ago. The two biggest reasons people give for switching away from traditional search: 40% are tired of clicking through link after link, and 37% are worn down by ads and clutter. Fair enough, really.

And once people do switch, 60% say AI gives them clearer answers than a normal search. But trust doesn't follow far behind:

  • 80% feel confident AI is giving them unbiased information, yet 85% still double-check it somewhere else anyway.
  • A separate, Yelp-commissioned survey backs this up almost exactly: 65% of Americans have used AI search, but only 15% trust it "a lot."
  • 63% still double-check elsewhere, and 51% said AI results feel like a "walled garden" that's hard to verify.

Trust in AI search really depends on who you ask, and what they're actually searching for.

Health is where this trust gap gets high-stakes. A friend recently told me he doesn't use Google at all anymore, only Claude, especially for anything embarrassing or health-related. That same day, he'd asked it why his sneeze smelled funny. It's a funny story, but it's also a real, concrete example of a much bigger pattern: health is one of the topics people bring to AI disproportionately more than the time they actually spend on it, right alongside paperwork, money, and legal issues.

Microsoft's own analysis of 37.5 million Copilot conversations found health is the dominant mobile topic, consistently, across every hour and month observed. A follow-up study of over half a million health-related conversations dug into what that actually looks like:

  • People ask about symptoms and condition management twice as often on mobile as desktop.
  • Emotional wellbeing conversations are 75% more common on mobile.
  • Roughly 1 in 7 of those symptom conversations aren't even about the person asking. It's a child, an ageing parent, a partner.
  • They spike specifically in the evening and overnight, exactly when a doctor isn't reachable.

Google's still the first platform people turn to for health questions, and AI Overviews now appear on over 82% of health-related searches. But ChatGPT has become the second most common destination:

  • More than half of US consumers turn to it to better understand a diagnosis after seeing a doctor.
  • 41% seek a second opinion from it.
  • 20% say they'd follow ChatGPT's advice over their own doctor's.

The part that actually stopped me: 26% of US consumers trust ChatGPT's advice more than their own doctor's. A quarter of people. I don't think that's really about AI being brilliant. I think it says a lot more about how much people in the US trust the healthcare system they're dealing with in the first place. That's my own reading of it, not something the data proves outright, but it's hard to look at that number and not wonder what it's really measuring.

Shopping: More Research, Not Less

Product search follows the same Google-first pattern. For research, 58% start with Google, ChatGPT ranks second or third at 19%. When it comes to actually deciding what to buy:

  • 39% turn to Google first
  • 15% to Reddit (just ahead of AI tools at 14%)
  • 11% each to review sites and to friends & family

Reddit's seen a 40% year-on-year rise in shopping conversations, and 84% of its shoppers say they feel more confident in what they've bought after researching there.

What surprised me most in this section was how much more verification AI seems to trigger, not less. A study combining 450+ screen-recorded AI shopping sessions with survey responses from 600 US consumers found:

  • Only 46% fully trusted AI's shopping recommendations.
  • 95% took extra steps to feel confident before finishing a session.
  • The share of shoppers visiting a retailer's own site doubled, from 20% before using AI to 50% after.

AI was supposed to simplify shopping. Instead, for a lot of people, it's added a step.

Zero-Click Search Is Real, and It's Growing

Not every search type is seeing more engagement, though. New SparkToro research found that 68.01% of Google searches in the US ended without a single click in the first four months of 2026, up from 60.45% in 2024. AI Overviews appear to be the main driver. They now show up on more than 20% of Google searches, and click-through rates drop by nearly 60% when they do.

What I found most interesting here wasn't the click-drop itself, but a study from researchers at the Indian School of Business and Carnegie Mellon University, who ran a genuine randomised experiment and found no measurable difference in user satisfaction between people who saw AI Overviews and people who didn't. In other words: AI Overviews aren't making the experience better. They're just capturing clicks that would otherwise have gone to a website.

The Way We Ask Has Changed

For years, "Googling" was a verb, and we were all conditioned into a particular clipped, robotic way of typing into a search box. Liz Reid actually has a name for that style: she calls it "keywordese." And she says people are dropping it. Instead of typing keywords, we're starting to just say what we actually want.

The data backs this up from several angles. Google's own AI Mode report found the average AI Mode query is now three times longer than a traditional search. A separate, independently measured SparkToro clickstream analysis confirmed six-to-nine-word searches are growing fastest.

The words themselves have changed too. The most common words in longer AI Mode searches are "information," "identify," "find," "explain," "summarise." But the one that stands out to me is much smaller: the most common first word is now, often, simply "I." When you start a search with "I," as in "I need," "I want," "I'm trying to," you're not typing a command anymore. You're describing your own situation. You're talking to it, not typing at it.

And it's not only length that's changed. What people are actually trying to do has shifted too. One study of 1.1 million ChatGPT conversations found the three dominant topics, Practical Guidance, Seeking Information, and Writing, make up roughly 77% of all usage. "Seeking Information" specifically, the closest substitute for old-fashioned web search, grew from 14% to 24% of conversations in a single year. And almost half of everything people ask ChatGPT for is Asking (49%), seeking advice, rather than Doing (task completion) or Expressing (no clear goal at all).

Search at Work, and Search to Learn

Flip to work specifically, and the pattern reverses: Doing makes up nearly 56% of work-related queries, against 35% Asking and 9% Expressing. At work, people mostly want ChatGPT to actually complete something. A huge chunk of that is writing: about 40% of work-related messages are writing-related, though about two-thirds of those are edits, translations, or summaries rather than new content from scratch. ChatGPT has become the team's copywriter.

This varies meaningfully by industry too:

  • Computer-related roles: 57% of usage is work-related
  • Management and business: 50%
  • Engineering and science: 48%
  • Other professional roles: 44%
  • Non-professional roles: 40%

On learning specifically, one in ten ChatGPT messages (10.2%) are requests for tutoring or teaching, and within the "Practical Guidance" category specifically, that's 36%, the single largest topic in there. AI Mode data shows the top school subjects people study for are maths, Spanish, history, English, and biology. It's not only school work, either. People are increasingly using it for proper qualifications. Professional exams, licenses, that sort of thing. A proper mix.

A Quick Detour: What New Zealand Actually Searched For

One of my favourite moments in preparing this talk was pulling Google's Year in Search data specifically for New Zealand, rather than relying on the US and global lists most of this research is built around. What do Lorde and a Labubu have in common? They were both in New Zealand's top trending searches for 2025, alongside family fugitive Tom Phillips, the All Blacks versus France, and a fair few weather warnings.

That last detail is worth dwelling on: four of New Zealand's top ten local search moments in 2025 were weather warnings. Google's own local team reads this as a real rise in people using search as a real-time safety tool during severe weather, which tracks, given how many Cyclone and storm events we've had. Two health conditions, COPD and osteoporosis treatment, also made New Zealand's overall Top 10, which Google's NZ comms team interprets as Kiwis increasingly taking health matters into their own hands, a nice, locally-grounded echo of the health section above.

Why Do Two People Get Different Answers to the Same Question?

This was the centrepiece of the talk, and the part I most wanted the room to experience, not just hear about. I asked everyone to search the exact same phrase, "What are the best noise-cancelling headphones?", on their own phones, then compare with their neighbour. Results varied, sometimes quite a lot. We did it a second time with "Recommend a good coffee shop near me," and unsurprisingly, given the location-dependence, results diverged even more.

There are two quite separate reasons this happens, and I think it's worth being precise about the difference.

The first is pure unpredictability. A study by SparkToro had 600 volunteers run 12 identical prompts through ChatGPT, Claude, and Google's AI, each prompt run 60-100 times per platform, across categories from chef's knives to cancer care hospitals. They found less than a 1-in-100 chance of getting the same list of brands twice, even from the exact same AI, asked the exact same question, repeatedly. This isn't about you specifically. It's just how these systems work.

The second is genuine personalisation: the platform using what it already knows about you to shape the answer. In sworn antitrust testimony, Google described a system called "Glue": a huge internal record of user activity logging every query, language, location, device, what was clicked or hovered over, and how long someone stayed. Google says this system measures "popularity as measured by user intent and feedback systems." That data trains Google's ranking models and reshapes results accordingly, not a new feature but one that's been considerably amplified by AI.

Robby Stein, Google's VP of Product for Search, said earlier this year that AI Overviews only appear where the system has learned users engage with them and find them useful. Which means your click, right now, is quietly shaping what someone else sees later. And in many cases, Google actually runs additional queries under the hood, ones you never see, to expand what you typed and predict what you were really looking for.

Jesse Dwyer of Perplexity put the personalisation mechanism plainly in an interview earlier this year: two people can ask the exact same question and get different answers because some AI tools load personal memory into the context window. There's no longer one shared, universal set of results: the answer is assembled dynamically, on the spot, based on whatever the platform already knows about the person asking.

Here's the tension I wanted to leave the room with. A recent YouGov survey of US respondents found 68% of people who don't use AI for search say they're not comfortable with it using their personal data to tailor answers. And even among people who already use AI, only 31% are comfortable with it, and only if they can control or turn it off. We just proved, live, in that room, that it happens. And most of us, when asked directly, say we don't want it to.

So, Does Google Know You Better Than Your Mum?

Probably. But it's not just Google anymore. Whichever platform you're using (Google, ChatGPT, TikTok, Perplexity), they're all doing some version of the same thing. Watching what you click. Remembering what you've searched before. Building up a picture of you, one query at a time.

We don't fully trust any of them. But we use them more every year, for bigger and more personal things: health, money, what to buy, where to eat, who to believe. Most of us, if you ask us straight, say we're not comfortable with any of that shaping our answers. We say that. And then we keep searching anyway.

A huge thank you to Nerd Nite Christchurch for having me, and to everyone in the room who put their hand up, ran a live search on their phone mid-talk, and generally made the interactive bits actually work. It's a different kind of audience from the industry events I usually speak at, and I loved it.

 

Research Referenced

Every statistic in this talk was individually fact-checked against its original, primary source before use, where a source updated or superseded an earlier figure, I've noted that below. Grouped roughly by theme, in the order they appear above.

Search vs. AI market share

  • SparkToro / Datos, "Search Happens Everywhere" analysis — Search Engine Land
  • Rand Fishkin (SparkToro), Google vs. ChatGPT search volume — Search Engine Land
  • Similarweb data, Google vs. AI chatbot visitor traffic — Search Engine Journal
  • StatCounter, New Zealand and global search engine market share (July 2026)
  • Microsoft, Bing 1 billion MAU disclosure (Q3 FY2026 earnings) — Search Engine Journal

Platform growth

  • OpenAI, ChatGPT 900M WAU / $110B funding round disclosure — Search Engine Land
  • OpenAI, ChatGPT crossing 1 billion monthly app users (June 2026)
  • Google I/O 2026, Gemini app 1 billion MAU / 63% voice usage — Search Engine Journal
  • Google, AI Mode usage report (19 May 2026) — Search Engine Journal

Platform diversity — Reddit, TikTok

Trust and adoption

Health search

  • Microsoft, Copilot conversation analysis (37.5M conversations) — Search Engine Journal
  • Microsoft, Copilot health & wellbeing deep-dive (Jan 2026, reported via Axios, March 2026)
  • Tebra, consumer and provider survey (Feb 2026)
  • Google, AI Overview health citation analysis — Search Engine Journal

Shopping and product research

Zero-click search

Query language and structure

How people use ChatGPT

New Zealand Year in Search

  • RNZ, "Google's top New Zealand searches of 2025 revealed" — RNZ
  • Google Trends, Year in Search 2025 (New Zealand) — Google Trends

Personalisation and answer variability

 

Back to blog

About The Author

I've been where you are — staring at a website that should be working harder for your business, wondering why other companies seem to show up everywhere while you're nowhere to be found.

I've spent over a decade figuring out why that happens and how to fix it, and I've helped 120+ businesses solve exactly this problem. Not because I have some secret formula, but because I take the time to understand what's actually going wrong with your specific situation.

No cookie-cutter solutions. No overwhelming you with everything that could be better. Just focusing on what will actually make the biggest difference for your business right now.

Find out more