NexGenData

Posted on Jul 6 • Originally published at thenextgennexus.com

China Social Media Trends API: Tracking Weibo, Douyin, and Baidu in Real Time

#apac #api #baidu #china

TL;DR. Mainland Chinese consumer attention concentrates on three platforms with almost no overlap to Western analogs: Weibo (~600M MAU, the closest functional equivalent to early Twitter for trending public discourse), Douyin (the domestic version of TikTok with ~750M DAU and a distinct content profile), and Baidu (the dominant Chinese search engine — Google has minimal China share). For consumer-brand research, China consumer-equity due diligence, and competitive intelligence on China-market entrants, you need real-time access to what's trending on these three platforms. Each exposes some public surface area but none publishes a comprehensive English-language API. This guide covers what each platform actually reveals about Chinese consumer attention, the structured trend data you can pull, and a working Python pipeline using the China Trends Tracker actor.

Why these three platforms

The Chinese internet operates on a parallel stack to the global Western one — Google, Facebook, and X have negligible mainland presence. For a non-Chinese researcher trying to understand what consumers are paying attention to in China, the three platforms above are the canonical sources, supplemented by RED / Xiaohongshu for lifestyle and Bilibili for younger-skewing video.

Weibo (微博) — public posts and trending topics. The closest cultural analog is "what's trending on X/Twitter today." The Weibo Hot Search (微博热搜) list updates throughout the day and is heavily watched as a real-time signal of public attention, including on consumer-brand topics.
Douyin (抖音) — the dominant short-video platform. Distinct from international TikTok (different content moderation, different creator economy, no overlap of creator accounts). Trending videos and hashtag challenges concentrate consumer-product attention faster than any other surface.
Baidu (百度) — search interest. Baidu Index (百度指数) publishes search-volume data with weekly and daily granularity for keywords. There is no public Baidu Index API; the index is browser-only.

What trend data is actually useful

For consumer-research and brand-monitoring use cases the structured fields that matter:

Trending topic with rank — Weibo Hot Search position over time. Rank movement is more informative than absolute rank.
Topic category — Weibo flags topics as entertainment, current affairs, sports, etc. Filter heavily for any business research.
Engagement signal — Weibo post count for a topic; Douyin video count and aggregate views per hashtag.
Time series — same topic across multiple snapshots to compute lifetime, peak, and decay.
Associated brands / entities — named-entity extraction over the trend text and top-engagement posts.

For Baidu Index specifically: the canonical signal is search-volume index for a keyword, with overall and Daily-Active-User splits. Higher noise than Weibo trending but lower bot contamination.

Working Python example

The China Trends Tracker actor consolidates Weibo Hot Search, Douyin trending hashtags, and Baidu search-interest signals into a single normalized JSON feed. Curl:


    curl -X POST "https://api.apify.com/v2/acts/nexgendata~china-trends-tracker/run-sync-get-dataset-items?token=$APIFY_TOKEN" \
      -H "Content-Type: application/json" \
      -d '{"platforms": ["weibo","douyin","baidu"], "category": "consumer", "maxResults": 200}'

Python — brand mention monitoring:


    import os, json, urllib.request

    APIFY_TOKEN = os.environ["APIFY_TOKEN"]
    ACTOR = "nexgendata~china-trends-tracker"

    brand_terms = ["蜜雪冰城","瑞幸","喜茶","茅台","小米","比亚迪","蔚来","SHEIN"]

    payload = json.dumps({
        "platforms": ["weibo","douyin","baidu"],
        "keywords": brand_terms,
        "includeAssociatedPosts": True,
    }).encode("utf-8")

    url = f"https://api.apify.com/v2/acts/{ACTOR}/run-sync-get-dataset-items?token={APIFY_TOKEN}"
    req = urllib.request.Request(url, data=payload, method="POST",
                                  headers={"Content-Type": "application/json"})

    with urllib.request.urlopen(req, timeout=600) as r:
        trends = json.loads(r.read())

    for t in trends:
        print(f"{t['platform']} | rank={t.get('rank','-')} | {t['topic']}")
        print(f"  engagement={t.get('engagement','-')} | sentiment={t.get('sentiment','-')}")

Returned shape:


    {
      "platform": "weibo",
      "snapshotAt": "2026-05-29T11:00:00+08:00",
      "rank": 7,
      "topic": "蜜雪冰城新品",
      "topicEn": "Mixue new product launch",
      "category": "BRAND_CONSUMER",
      "engagement": 4280000,
      "engagementType": "discussion_count",
      "associatedEntities": ["蜜雪冰城","Mixue Bingcheng"],
      "topPostUrl": "https://weibo.com/...",
      "snapshotUrl": "https://s.weibo.com/top/summary"
    }

Cost comparison: China social listening tools

Source	Coverage	Sentiment / NER	Annual cost
Baidu Index (web)	Search interest, by keyword	No, requires Baidu login	Free, no API
Weibo official open platform	Limited, app-only API for approved partners	No	Partner program required
Sina Weibo public hot search	Top trending topics	No	Free, web only
Brandwatch China module	Multi-platform aggregation	Yes, with English UI	Enterprise quote, $10k+ year
Talkwalker China module	Multi-platform aggregation	Yes	Enterprise quote
Local CN social-listening vendors (CMNet, etc.)	Deep CN coverage	Yes	Variable, often $5k–$30k/yr
China Trends Tracker actor	Weibo + Douyin + Baidu signals	Basic NER, sentiment optional	PPE per snapshot

For enterprise-scale brand surveillance with compliance-grade audit trails, Brandwatch and Talkwalker remain the right choice — they integrate with marketing-team workflows, retain historical archives, and handle multi-region monitoring. For research-team and investment-team use cases that just need "what's trending in China today on the platforms my consumers actually use," the lighter PPE-priced approach is materially better fit.

What this enables

Consumer-brand due diligence for VC and PE deals in China consumer — pre-deal sentiment baseline plus post-deal tracking
Equity research on China consumer names — Mixue, Luckin, BYD, Xiaomi, NIO. Real-time attention signals as a feature for fundamental analysis
Competitive intelligence — track non-Chinese brands' Chinese-market attention (Apple, Tesla, Nike all have heavy CN signal but no equivalent US public data)
Crisis monitoring for global brands operating in China — Weibo trending is typically the first signal of brewing reputational issues

Sampling cadence and storage

Weibo Hot Search rotates quickly — the average lifetime of a trending topic is hours, not days. For meaningful trend research you want 15-30 minute snapshots stored as a time series, then aggregated nightly. Daily-only snapshots miss the majority of relevant topics. The actor exposes batch and scheduled modes; for serious time-series use, schedule snapshots at 15-minute granularity.

Bot contamination and trend authenticity

One realistic caveat for any Chinese social-trend research: paid promotion and coordinated inauthentic activity are well-documented features of the Weibo and Douyin trend ecosystems. Brand-paid trending topics (the platforms run formal trending-buy ad products), MCN-coordinated creator promotions, and outright bot amplification all show up in the raw trend feed. For investment-grade research, the trend rank alone is insufficient — you need engagement-velocity profile (organic trends accelerate gradually; paid trends spike then collapse), creator-account quality scoring, and cross-platform corroboration (a trend that appears organically on Weibo, Douyin, and Baidu simultaneously is materially more credible than a single-platform spike). The actor exposes the raw feed; downstream credibility scoring is the analyst's responsibility.

Censorship and trend disappearance

A trend that appears on the Weibo Hot Search and then disappears within minutes is itself a signal — sensitive topics get scrubbed faster than organic decay would explain. For research workflows the practical handling is to sample at high frequency (15-minute intervals minimum) and to retain the snapshot history independently of the platform. The disappearance window is occasionally a useful research feature; conversely, a trend that survives multiple snapshots is meaningfully more durable. The actor's time-series mode captures appearance and disappearance timestamps for downstream analysis.

What this does not cover

WeChat (微信) — the substantial majority of WeChat content is private (1:1 messaging and closed Moments). There is no scrapable public Weibo-equivalent surface.
RED / Xiaohongshu (小红书) — lifestyle content, distinct platform with its own dynamics. Worth a separate feed for any beauty / fashion / lifestyle research.
Toutiao (今日头条) — news-aggregator with editorial curation, structurally different from Weibo trending.

For consumer-equity research linking attention data to share prices, pair this feed with A-share consumer-sector data via Eastmoney for mainland-listed names and Chinese ADRs for the US-listed consumer cohort.

Get started: Pull your first China trends batch free at the China Trends Tracker actor page.

DEV Community