Taobao Keyword Suggestions Scraper: the Chinese words shoppers type into 淘宝

Give it a seed term and you get back Taobao's own autocomplete list: up to 100 suggestions per query, in the order Taobao ranks them, with the seed and the parent query on every row. Turn expansion on and it re-asks about each suggestion it found, which is the only way to go deeper because Taobao's suggestion endpoint has no pagination.

The honest limits, up front. The index is Chinese. 蓝牙耳机 returns 100 suggestions and "bluetooth earphones" returns none, so this only pays off if your seeds are in Chinese. And there is no search volume here. Taobao publishes a demand count on one narrower list, it answers for roughly half of head terms and almost no long-tail ones, and demandCount is null everywhere else.

Input	Chinese seed keywords
Output	One row per unique keyword
Ceiling	100 suggestions per query, 100,000 keywords per run
Account needed	None, and nothing to install
Price	$0.085 per 1,000 keywords, flat on every plan

🔤 What Taobao Keyword Suggestions Scraper does

At depth 0 it asks Taobao about your seeds and nothing else, which gives you up to 100 keywords per seed. Short, broad seeds return the most.

At depth 1 it takes everything that came back and asks about each of those too, which is roughly a hundredfold more keywords. Depth 2 does it again and gets very large very quickly, which is why Maximum keywords is a hard cap across the whole run and why it is worth setting deliberately.

A keyword that shows up under two different parents is returned once and billed once. The run keeps track of what it has already seen, so you are not paying for the same term twice.

Two optional extras ride alongside. includeDemandData adds one request per seed against Taobao's Tmall list, which is the only variant carrying a numeric demand count, a category id and a relevance score. includeRefinementTags attaches Taobao's own attribute and brand vocabulary for the parent query, the style, material and brand facets with their "hot" markers.

📥 What you give it

{
  "seedKeywords": ["蓝牙耳机", "连衣裙"],
  "expansionDepth": 0,
  "maxKeywords": 1000,
  "includeDemandData": true,
  "includeRefinementTags": false
}

Field	Default	What it is
seedKeywords	box starts at 蓝牙耳机 and 连衣裙	Terms to expand, one per line, in Chinese. Each seed returns up to 100 suggestions.
expansionDepth	box starts at 0	0 asks only about your seeds. 1 also asks about every suggestion that came back. 2 goes one level deeper again.
maxKeywords	box starts at 1000	Hard cap on unique keywords across the whole run, 1 to 100,000.
includeDemandData	box starts on	One extra request per seed for the Tmall demand count, category ids and relevance score. Only asked for your seeds, never for expanded keywords.
includeRefinementTags	box starts off	Adds Taobao's refinement vocabulary for the parent query. Off because it repeats the same list on every row of a query.
maxConcurrency	box starts at 15	Under Advanced. How many suggestion requests run at once, 1 to 30.
proxyConfiguration	leave it alone	Optional. Only if you want the run to go out through a particular network of your own.

None of these carry a schema default on purpose, so an empty API call writes one sample row rather than starting a run you did not ask for.

📤 What you get back

A real row from a recent run:

{
  "keyword": "连衣裙女夏",
  "parentQuery": "连衣裙",
  "seedKeyword": "连衣裙",
  "depth": 0,
  "rank": 1,
  "source": "suggest",
  "demandCount": null,
  "relevanceScore": null,
  "categoryIds": [],
  "scrapedAt": "2026-09-20T19:22:10.804Z"
}

Field	What it is
keyword	The suggestion, exactly as Taobao returns it. Nothing is translated.
parentQuery	The query that produced this keyword. At depth 0 it is the seed itself.
seedKeyword	The root seed the chain started from, so you can group a large expansion back to its origin.
depth	0 for a direct suggestion, 1 or 2 for one found by re-asking.
rank	Where it sat in Taobao's list, 1 to 100. Taobao's own ordering, not a score we computed.
source	suggest for the main list, tmall-b2c for the narrower one that carries demand data.
demandCount, relevanceScore	Only on tmall-b2c rows. null on everything else, which is most rows.
categoryIds	Taobao category ids, and also only on tmall-b2c rows. Empty on a suggest row, as in the example above.
refinementTags	Only when you asked for them. Present on about 80% of short head terms and 13% of expanded long-tail terms.

🧾 Reading the output

Row	How to spot it	Billed
A keyword	It has a keyword and no _diagnostic	yes
The sample row	_sample: true	no
A diagnostic	_diagnostic: true and an errorCode	no

There is no charged field on these rows. Filter on _diagnostic being absent and you have the keywords.

Code	What it means
NO_SUGGESTIONS	Taobao had nothing for that query. Usually an English seed, or one that is too long or too specific.
QUERY_FAILED	That query could not be read. The rest of the run carries on.

If every request in a run fails, the run ends as FAILED rather than reporting a quiet success with an empty dataset.

▶️ How to run it

1. Open Taobao Keyword Suggestions Scraper and click Try for free.
2. Put your Chinese seeds into Seed keywords, one per line. Replace the examples.
3. Leave Expansion depth at 0 for a first run. You will get up to 100 keywords per seed.
4. Set Maximum keywords to something you are happy to pay for before raising the depth.
5. Click Start, then download the dataset as JSON, CSV or Excel, or read it from the Apify API.

💰 How much does it cost?

$0.085 per 1,000 keywords. Flat on every Apify plan, with no volume tiers.

One billed keyword per unique term. A term that appears under two parents is billed once, the sample row is not billed, and diagnostic rows are not billed.

Depth is the number to watch, not the seed count. Ten seeds at depth 0 is at most 1,000 keywords. The same ten at depth 1 is a far bigger run, which is what Maximum keywords is there to bound.

💡 What people use it for

• Finding the Chinese phrasing for a product before writing listings or ads, where the obvious translation is not what shoppers actually type.
• Building a long-tail keyword set for one category by seeding a head term and expanding once.
• Reading rank as Taobao's own ordering to see which variants it puts first.
• Pulling categoryIds off the Tmall rows to work out which category a term belongs to.
• Comparing the vocabulary around two competing product types by seeding both and diffing.

🚧 What it does not do

• No search volume. demandCount comes only from the narrower Tmall list and is null on most rows. Do not plan a volume analysis around it.
• English seeds return almost nothing. The index is Chinese and there is no translation step here.
• Keywords, not products. No prices, no images, no listings. If prices are what you need, a keyword feed is the wrong purchase however cheap it is.
• No history. Every row is a snapshot of what Taobao suggested at that moment.
• No pagination. 100 per query is Taobao's ceiling, so depth is the only way deeper.
• categoryIds is sparse. On one 5,000-row run it was filled on about 0.5% of rows, because the demand list is only asked for your seeds.
• A big run can come back short. Taobao can refuse a query, which becomes an uncharged diagnostic row. Only delivered keyword rows are billed.

🧭 Which Taobao scraper do you need?

If you want	Use
The words shoppers type	This one
Products from a chosen category	Taobao Category Products Scraper
Products from Taobao's own recommendation feed	Taobao Product Feed Scraper
Wholesale listings and prices on 1688	1688 Wholesale Scraper
Export suppliers and MOQ on Alibaba.com	Alibaba.com Scraper

❓ Questions people ask

Can I seed in English? You can, but Taobao will mostly return nothing and you will get NO_SUGGESTIONS rows. Translate your seeds first.

How many keywords will one seed give me? Up to 100 at depth 0. Broad terms hit that ceiling, narrow ones return far fewer.

Why is demandCount null on nearly every row? Because it exists only on Taobao's Tmall list, that list answers for roughly half of head terms, and it is only asked for your seeds.

How big does depth 1 get? Roughly a hundred times depth 0, so set Maximum keywords before you turn it on.

Can I schedule it? Yes. keyword plus parentQuery is a sensible key for diffing one run against the next.

Is this legal? Autocomplete suggestions are public and carry no personal data. Apify's write-up on scraping and the law is a good starting point, and we are not lawyers.

🆘 If something breaks

Open the Issues tab on the actor page. Send the seeds you used, the depth, and the run ID. If you got NO_SUGGESTIONS on a Chinese seed that works in Taobao's own search box, that is worth reporting.