Baidu SEO for foreign brands: does search still matter in a super-app China?
Short version: yes, Baidu SEO still matters for foreign brands — but far less than the agencies selling it want you to think, and for reasons that have almost nothing to do with 2015-era keyword tricks. Baidu is still China's biggest search engine, but "search" in China no longer lives in one box. It's scattered across super-apps, in-app search bars, and now AI answers. Baidu's real value today isn't traffic; it's the verification layer — the place a cautious buyer, a distributor, or increasingly an AI assistant goes to check that you're real. Treat it as a credibility utility, not a lead-gen channel, and you'll spend the right amount on it: some, not everything.
Every few months a Western CMO asks me the same question, usually with a hopeful lean: "Do we still need to do Baidu?" What they're really asking is whether they can skip it. The honest answer is that they're asking the wrong question. Baidu isn't a yes/no line item anymore. It's one input into a much messier picture of how Chinese buyers find and vet things, and the mistake is treating it either as the whole game (the way a 2015 agency deck did) or as irrelevant (the way a jaded operator does after seeing the traffic numbers).
Baidu still leads — but "search" doesn't mean what it used to
Let's start with the fact that gets over-quoted. Baidu is still the largest general search engine in China. Depending on whose tracker you read and whether you count desktop or mobile, its share of the classic search-engine market lands somewhere in the mid-40s to mid-60s percent, and it's stronger on mobile — comfortably the majority of mobile search — than on desktop, where Bing and others have crept in. The trend line, though, points the same direction across every source: slowly down.
It's slipping not because a rival search engine beat it, but because the very idea of a central search box eroded. A Chinese user looking for a product searches inside Taobao. A restaurant, inside Meituan. A skincare recommendation, inside Xiaohongshu. A short video, inside Douyin. A government service or a merchant, inside WeChat. None of those queries touch Baidu, and none of them show up in "search engine market share." So the raw stat undersells how much Baidu has actually lost as the front door. Discovery in China is now federated across a dozen walled gardens, and Baidu is one garden among them, not the courtyard they all open onto.
That's the reframe that matters: don't ask "how much traffic will Baidu send us." Ask "what job does Baidu still do that the super-apps can't." Because there is one, and it's the reason to bother.
What Baidu is still uniquely good for
Baidu's surviving edge is that it's the open, cross-cutting index. The walled gardens don't let you search across them; Baidu still crawls the open Chinese web and its own properties. That makes it the default place people land when the question isn't "what should I buy" but "is this thing legitimate." Three jobs in particular still run through it:
| Query intent | Where it actually happens | Does Baidu matter? |
|---|---|---|
| "What lipstick should I buy?" | Xiaohongshu, Douyin | Barely — this left Baidu years ago |
| "Cheapest price for X" | Taobao, JD, Pinduoduo | No |
| "Is [foreign brand] real / any good?" | Baidu + Zhihu + Baike | Yes — this is the verification query |
| "[Brand] official site / contact / B2B info" | Baidu | Yes — especially B2B and distributor due-diligence |
| "[Category] explained / how does X work" | Baidu + AI answer box | Increasingly, via generative results |
Notice the pattern. Baidu lost impulse and price discovery to the super-apps, and it was never going to win those back. What it kept is the skeptic's query and the professional's query — the moments where someone needs to confirm you exist, that you're not a scam, and that a third party corroborates your story. For a B2B brand or a considered-purchase category, that's not a small residual. It's often the single query that unblocks a deal.
The AI answer box changed the target
Here's the part most foreign brands haven't caught up to. Baidu now puts AI-generated answers at the top of a lot of results pages, drawing on its ERNIE models, which have grown into hundreds of millions of users and handle enormous daily query volume. DeepSeek and other Chinese models pushed the whole ecosystem the same way. Practically, that means the win is shifting from "rank #3 on the results page" to "be the source the AI answer cites." The Chinese search world is going through the same SEO-to-GEO (generative engine optimization) shift as everywhere else, only faster, because the platforms are more aggressive about surfacing AI answers.
What earns those citations isn't keyword density. It's structured, credible, simplified-Chinese content that an AI can lift with confidence: a clean Baidu Baike entry, corroborating answers on Zhihu, consistent brand facts across the open web, genuine third-party references. The gap between brands that have that scaffolding and brands that don't is widening every quarter. This is the same shift I've written about in what AI changes about cross-border marketing — Baidu is just where it bites first for the verification query.
The infrastructure question everyone gets wrong
You'll read that you "need an ICP license to rank on Baidu." That's overstated. An ICP filing is a government registration required to host a site on mainland Chinese servers, and Baidu has never confirmed it as a direct ranking factor — studies routinely find that fewer than half of top-ranking pages even display an ICP number. So no, you can't be refused a ranking purely for lacking one.
But the practical chain still points toward doing it. Without an ICP filing you can't host on the mainland, and without mainland hosting (or a genuinely China-optimized CDN) your site crawls slowly through the Great Firewall's inspection — and a slow, sometimes-timing-out site is a real ranking and conversion problem, ICP or not. Add that Baidu paid ads, WeChat/Alipay payment integration, and Tmall all effectively assume a local entity, and the license stops being a Baidu question and becomes a "are you actually operating in China" question. The 2025 tightening of real-name verification on user-generated content raised the compliance bar further, with the ICP holder on the hook. My rule of thumb: if China is a test, skip the license and use a fast CDN; if China is a commitment, get the entity and stop treating Baidu infrastructure as optional.
What a foreign brand should actually do
Right-size the effort to the job Baidu still does. For most brands that's a short, unglamorous list:
- Own your verification results. Make sure that when someone searches your Chinese brand name, page one is clean and true: a working Chinese site, a Baidu Baike entry, and a couple of credible third-party mentions. This is the highest-ROI Baidu work and most brands skip it.
- Write in simplified Chinese, for the machine and the skeptic. Translated marketing copy won't cut it. Structured, factual, genuinely informative content is what both Baidu's index and its AI answers reward.
- Feed the properties Baidu trusts. Baike for the encyclopedic facts, Zhihu for the considered "is it any good" answers — Baidu leans on both, and they surface in AI results. The Zhihu trust play and Baidu SEO are increasingly the same project.
- Fix hosting before you chase keywords. A fast, China-reachable site beats a heavily "optimized" one that loads in six seconds. Sort speed and reachability first.
- Don't buy a big Baidu SEO retainer for a consumer-impulse brand. If your category lives on Xiaohongshu and Douyin, Baidu is a hygiene layer, not a growth channel. Budget it that way.
Bottom line
Does search still matter in a super-app China? Yes — but its job shrank to something specific and durable: Baidu is where people and machines go to verify you, not to discover you. Discovery moved into the super-apps and isn't coming back. So stop asking whether to "do Baidu" as if it were a growth channel, and start treating it as the credibility utility it's become — own your branded results, feed the AI answer box with structured simplified-Chinese content, get the infrastructure right, and spend the saved budget where discovery actually happens. The brands overspending on Baidu SEO and the ones ignoring it entirely are making the same mistake from opposite ends: both are still arguing about a front door that's already been replaced by a dozen side entrances.
If you're trying to figure out how much of your China budget Baidu really deserves — and where the rest should go — that's the work I do, so reach out.