Industry search demand

570 candidate industries expanded into ten search roots each, queried against Baidu and 360 autocomplete: 569 came back with words, 73174 distinct words in total.

The numbers

What one record is

One row per industry that returned words. Columns: the industry name, the number of deduplicated search words, how many of those are decision-type, research-type, negative-type and learning-type, the decision ratio (decision words divided by all words), the number of roots used, the collection source, which pool the industry name came from, and the date that row was collected.

Records

The point of the table is not how big an industry is. It is what kind of question the industry's own name drags out of an autocomplete box: a decision, a reason to keep reading, a complaint, or nothing but a definition.

Columns

• industry - the industry name, written in Chinese in the file, the way it was queried.
• words_dedup - distinct suggested words after deduplication. This is the size of the question, not the size of the market.
• decision, research, negative, learn - how those words split by intent.
• decision_ratio - decision words divided by all words. This is the column the ranking is built on.
• roots - how many of the ten roots returned anything.
• src, pool_src, collected_at - where the words came from, which pool the industry name came from, and when that row was collected.

Sample rows

Boundaries

• The roots have to be words people actually type. A formal industry name such as a statistical classification understates real attention, which is why names are merged before the run.
• Word count is not market size. Two industries can pull the same number of words and mean opposite things, which is exactly what the decision ratio is for.
• The collection covers the Chinese-language side only. The English side was measured separately and is kept as background, not published as a ranking.

Download

• industry-demand.csv text/csv — 569 rows
• industry-demand.json application/json — 569 rows, columnar, plus the collector's intent scheme

Files are served from this domain and stay at these URLs. Licence: CC BY-NC 4.0 — attribution required, commercial use needs written permission.

Method & definition

• Source. InfoGap own keyword collector: every candidate industry name is expanded into ten search roots (five decision, three research, two negative), queried against Baidu autocomplete (www.baidu.com/sugrec) and 360 autocomplete (sug.so.360.cn/suggest), and the suggestions are deduplicated into distinct words. No third-party dataset is redistributed.
• Definition. One row per industry that returned words. Columns: the industry name, the number of deduplicated search words, how many of those are decision-type, research-type, negative-type and learning-type, the decision ratio (decision words divided by all words), the number of roots used, the collection source, which pool the industry name came from, and the date that row was collected.
• Last updated. 2026-10-04
• Update frequency. daily (collector runs 02:00 Asia/Shanghai)
• Time range. see temporalCoverage in this page's schema.org Dataset block.
• Licence. CC BY-NC 4.0 — attribution: InfoGap (infogap.store), link to this page. Commercial use requires written permission.

Where to go next

• Data pages: https://infogap.store/data/p/industry-sides/
• Method pages: https://infogap.store/method/how-to-check-our-numbers/
• Hub: https://infogap.store/data/
• Machine-readable: registry.json
• For AI engines: llms.txt — what to cite, and with which definition.