Forget Google: Meta Just Became a Search Engine With 3 Billion Users

Okay, I need to talk about something that's been quietly breaking my brain for the past few weeks. And no, it's not the usual existential crisis about whether my product listings are optimised enough (though that's also happening, obviously, at 1AM, in my pants, eating cereal over my laptop like a functioning adult).

It's about where people start looking for products now. Because for twenty years, the answer was simple: Google. Someone wanted running shoes, a blender, a suspiciously cheap duvet — they typed it into one search box, and the entire ecommerce industry built itself around showing up in that box.

That era is ending. Not because Google is dying — it isn't, and anyone telling you otherwise is either selling you something or hasn't looked at an earnings report recently. What's happening is weirder and, honestly, harder to deal with. The search box has shattered into about six different surfaces, none of which play by the same rules, and one of them is Meta, which has apparently decided to become an answer engine while nobody was paying attention.

Where Your Customers Are Actually Starting Now

Here's the thing that messed with my head once I actually mapped it out: a single purchase intent — let's say someone wants a new office chair — can now begin on at least five completely different surfaces.

There's Google, which is itself becoming three things at once — the classic blue links, AI Overviews summarising products above the fold, and AI Mode offering full conversational recommendations. There are the standalone answer engines: ChatGPT, Perplexity, Gemini. There's TikTok, where a genuinely alarming percentage of younger shoppers now discover products without ever framing it as "searching." There's Amazon, which for a huge chunk of purchases is the first and last search that happens.

And now there's Meta AI, sitting inside Facebook, Instagram, WhatsApp, and Messenger, quietly answering product questions for people who never intended to search for anything at all.

These surfaces are all chasing the same customer at the same moment. What they don't share is a playbook. Each one surfaces answers through different machinery, pulls from different sources, and rewards completely different seller behaviour. Your customer is one person. The routes to reaching them are splintering like a phone screen dropped on concrete.

Why Meta AI Is the One That Should Worry You Most

Most of the AI search conversation has focused on model quality — which chatbot gives the smartest answer, which one hallucinates least, which one actually understands that "best running shoes for wide feet under £120" is a purchase intent and not a philosophy question. That's the wrong lens for Meta AI, and judging it on that axis is exactly why most sellers have written it off.

Meta's advantage isn't the model. It's distribution. By mid-2025, Meta AI had reportedly hit around a billion monthly users across Meta's apps. The company reports something like 3.5 billion people using its family of apps every single day. WhatsApp alone passed three billion monthly users. Instagram hit a comparable number. Facebook isn't cool anymore — I know, I know — but "not cool" and "irrelevant" are very different things when you're trying to sell products to actual humans. (Most of whom, it turns out, are also not cool. Welcome to commerce.)

Here's what makes this different from just another chatbot: someone sees your competitor's product on Instagram, and instead of leaving to Google it, they ask Meta AI whether it's any good, what the alternatives are, and where to buy it. A family planning a birthday party in a WhatsApp group asks Meta AI for gift recommendations without ever opening a browser. Someone in a Facebook group about home gyms asks Meta AI to summarise what people are actually saying about a particular rowing machine.

The first search — historically Google's most valuable asset, and the moment your entire funnel depends on — is being absorbed into the surface where the intent already lives. Your customer never leaves the app. They never see your listing. They never click your ad. They just... get an answer, and it may or may not include you.

The Discovery Problem That's Actually Three Problems Wearing a Trenchcoat

Here's where this gets properly uncomfortable for anyone selling online.

We've covered previously how different AI search engines cite largely different sources — there's surprisingly little overlap between what ChatGPT surfaces and what Perplexity recommends for the same query. The practical conclusion was that answer engine optimisation isn't one job but several, because each engine runs its own retrieval logic.

Meta AI extends that problem past the point where the familiar toolkit helps at all. The other engines — Google's AI Overviews, ChatGPT, Perplexity — mostly retrieve from the crawlable, linkable web. Your product pages, your blog content, your structured data. The stuff you've spent years building and optimising. Meta AI increasingly doesn't. Its Facebook AI Mode surfaces answers rooted in what people are actually posting — in Groups, in Reels, in comments, in reviews shared socially. The retrieval inputs aren't your website. They're public social content, creator posts, community engagement, and product mentions you may not even know exist.

That's not a fourth variation of the same SEO problem. It's a completely different substrate. It's like spending years perfecting your recipe for a baking competition and then being told the judges are actually scoring based on what strangers said about your cake on Reddit. (This analogy is already collapsing, but you get the point.)

The Scissors Gap (Or: Why Your Marketing Budget Is About to Get Very Uncomfortable)

Two things are happening simultaneously, and neither of them is fun.

Discovery — the earliest, most valuable moment where a customer forms a shortlist of products — is scattering across surfaces that share no common playbook. Google, the answer engines, TikTok, Meta's social layer. You now need to be visible across all of them just to have a shot at being considered.

Meanwhile, conversion still concentrates in a handful of destinations — your website, Amazon, maybe Shopify. The surfaces where people actually buy things haven't fragmented. Just the surfaces where they find things.

So the cost of being discovered is rising across an ever-widening front, while any single channel returns less. The two lines — what it takes to be found and what each channel delivers — are pulling apart like scissors. And you're the thing getting cut. (Sorry. That metaphor got darker than I intended.)

This gets sharper when you look at the money. Forecasts now have Meta overtaking Google in net worldwide ad revenue in 2026 — roughly $243 billion against $240 billion. If AI answers get monetised through paid recommendations and sponsored suggestions (and of course they will, because nothing on the internet stays free for long), the commercial value flows to whoever owns the moment the question gets asked. Not whoever has the best product page. Not whoever has the best SEO.

Meta already holds the audience, the ad infrastructure, the creator relationships, and the commerce signals. An answer engine inside its apps lets it own discovery, the question, the recommendation, and eventually the checkout — while keeping the machinery invisible. The customer sees helpful advice. You see a black box that may or may not be recommending your products and offers you absolutely no way to tell.

What Sellers Should Actually Do Now

Not panic. Panic is not a strategy, though it does show up at every marketing meeting I've ever attended, usually around slide seven.

Stop assuming the search starts on Google. For your specific product category, run the same queries across Google's AI answers, ChatGPT, Perplexity, and Meta AI inside Facebook and Instagram. Compare what comes back. Track whether your products appear, whether answers cite or link to your listings, and — on Meta specifically — whether public social content seems to shape the response.

Audit your social substrate. On Meta's surface, the retrieval inputs aren't your product pages. They're your Groups presence, your Reels, your creator relationships, your product mentions, your review language. If you've been treating social as "brand awareness" and search as "performance marketing," you've just discovered they're the same thing on this platform — which is either exciting or horrifying depending on how your social content looks right now.

Treat visibility as a portfolio, not a channel. Some surfaces reward the classic web-first playbook — structured data, technical SEO, authoritative content. One rewards public social content and community engagement you may not currently think of as "search" at all. The allocation question — where to spend finite budget as discovery fragments — is now a genuine strategic decision, not a technical one.

Get honest about what you can't measure. Across five surfaces, several of which surface mentions rather than links and one of which draws on private social behaviour, connecting visibility to revenue becomes guesswork dressed as data. Anyone selling you a single dashboard that claims to capture your presence across all of it is selling you a very expensive feeling of control.

The Bottom Line for Sellers

The search box didn't get smarter. It multiplied — into feeds, group chats, video apps, marketplaces, and messaging threads. And one of those copies is now sitting inside the apps where 3.5 billion people already spend their attention, answering product questions with signals you can't see, using inputs you may not control.

The old question — where do I rank? — is becoming less useful by the month. The question that matters now is: where does my customer's question begin? And if the answer is "inside a WhatsApp group while they're also watching Reels and haven't opened Google in three days," then your carefully optimised product listing is doing its best work in an empty room.

Which, honestly, is a pretty accurate description of how most of my marketing efforts feel most of the time. But at least now I know which empty rooms to worry about.

• Workers who adopt AI get busier, not freer. ActivTrak finds email and chat time more than doubles and business software use up 94%. UC Berkeley says they reclaim outsourced tasks and multitask across bots in every spare moment.

• Claude Cowork rolls out to mobile and web, letting users start a task at their desk and pick it up anywhere. It preps client meetings, drafts follow-ups unsent, and runs scheduled tasks with no device online. Beta starts on Max.

  

• ChatGPT for PowerPoint launches across all plans, letting users create slides, rewrite decks, and turn notes or spreadsheets into presentation-ready content in plain language while keeping output editable in PowerPoint.

• AI Overviews grow 71% on commercial-intent SERPs over six months, per Semrush. They now coexist with Google Ads twice as often as a year ago, and Finance queries jump 231%. Keywords with AI Overviews carry higher CPC.

• Meta launches Muse Image across its apps and previews Muse Video. The agentic model plans, searches, and self-refines, ranking No. 2 on Arena behind GPT Image 2. New Content Seal watermark survives cropping and screenshots.

• AI search reshapes paid media as 37% of consumers start with AI tools and 60% of searches end without a click. AI-referred traffic converts up to 4x higher, and engines cite premium publishers 3x more than brand-owned content.

🤖 Voice/Style/Tone AI Prompt Snippet Generator - Craft prompts to replicate any text’s voice/style/tone.

📈 Simple Analytics - Get insights about your business by chatting with your analytics in real-time using AI.

🔎 Just words - Optimize your product's copy for user growth.

📝 Fixkey - Improve your writing everywhere on macOS

🪄 Magic Documents - Organize, rename, summarize, and extract key data from any document with AI.

😍It was great, all I needed

🙂Good, not great

🙄It sucked