> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.beatapi.io/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.beatapi.io/_mcp/server.

# How do I research one brand across Weibo, Douyin and Xiaohongshu in a usable table?

> Query the same research question separately, retain platform conditions, remove namesakes and group sourced wording and usage situations.

**Send the same research question to each platform's own search action, then put the sourced results in one working table that you define.** Search Weibo with `weibo.web_v2.fetch_realtime_search`, Douyin with `douyin.search.fetch_general_search_v2` and Xiaohongshu with `xiaohongshu.app_v2.search_notes`. Keep each platform's original response and query conditions, and group rows by the actual content object and the user question.

For example, for a new headset, record the brand, model and confirmed aliases, then read one initial batch from each platform for connection, comfort and calling questions. Record unclear models, namesakes and retellings of the launch announcement as separate rows.

## What still depends on the platform when using one key?

The BeatAPI key and Bearer authentication are the same for all three platforms. Before choosing an action, try one real question on each platform and check three things: whether the action returns the objects and level of detail you need, which filter and continuation inputs the action uses, and whether the result supplies sources you can hand off. Account inventories, video search and comments are different tasks with different actions. For a launch brief that needs a WeChat announcement, find the article with [WeChat Search](/guides/wechat-topic-search) and read the article body if needed; use the matching platform workflows for videos and notes. Give your working sheet the columns you need, and keep each action's original fields and cursor contract. Estimate the planned requests with [pricing](https://beatapi.io/pricing).

## How do you query the three platforms?

Send each JSON below as a separate body to `POST https://api.beatapi.io/v1/social-data/call` with BeatAPI Bearer authentication. Replace `<brand_and_model>` with the brand or model term you recorded for the round.

Weibo, first batch:

```json
{
  "action": "weibo.web_v2.fetch_realtime_search",
  "params": {"query": "<brand_and_model>", "page": 1}
}
```

Douyin, latest videos within a week (the first `cursor` is 0):

```json
{
  "action": "douyin.search.fetch_general_search_v2",
  "params": {"keyword": "<brand_and_model>", "cursor": 0, "sort_type": "2", "publish_time": "7", "content_type": "1", "search_id": "", "backtrace": ""}
}
```

Xiaohongshu, latest notes within a week:

```json
{
  "action": "xiaohongshu.app_v2.search_notes",
  "params": {"keyword": "<brand_and_model>", "page": 1, "sort_type": "time_descending", "time_filter": "一周内"}
}
```

One batch per platform means three queries. Record each brand term separately and reuse responses you already saved. To read further pages, use each action's own continuation state; the three actions do not share cursors.

## How do you align time scope?

| Platform    | Example scope                                 | What to retain                                                   |
| ----------- | --------------------------------------------- | ---------------------------------------------------------------- |
| Weibo       | Keyword search without a separate date filter | Actual publication times when available and manual window review |
| Douyin      | Latest, last-week videos                      | Filters, actual objects and collection time                      |
| Xiaohongshu | Latest, last-week notes                       | Filters, actual objects and collection time                      |

The three rows above are observations under different platform conditions. List posts with missing publication times separately; collecting all three platforms on the same day does not make their time coverage identical.

## How do you define common columns?

The columns below are working columns that you define; the three actions do not return shared response keys:

| Column                              | Evidence                                                        |
| ----------------------------------- | --------------------------------------------------------------- |
| Platform, term, conditions and time | Actual request records                                          |
| Original source and stable identity | Actual output and confirmed object                              |
| Wording and context                 | Text, description or comments actually read, with material type |
| Product and conditions              | Stated model, device or environment, leaving gaps open          |
| Topic, interpretation and next step | Reviewed classification with reasons                            |
| Request ID and original location    | Traceable sources for further reading                           |

Map each platform's fields separately instead of copying a shared `author`, `metrics` or `page.next_cursor` projection. The [Xiaohongshu search guide](/guides/xiaohongshu-notes-research) records the verified note path for `xiaohongshu.app_v2.search_notes`; map identifiers and text for Weibo and Douyin from their actual results.

## How do you handle overlap and interaction?

When the same object reappears on one platform, keep one content object with several observations. When matching wording appears on two platforms, record a possible retelling relationship, and do not merge authors or users until the relationship is confirmed.

Group questions across platforms, but keep each platform's engagement definitions. A connection issue that appears on two platforms is reported with supporting excerpts and each platform's sample scope, not with likes, favorites and reposts summed into a global heat score.

Select material by relevance, concrete situation and follow-up need first. Then use engagement within each platform to decide what to read further, so that unrelated popular posts do not displace specific low-engagement questions.

For example, a Weibo post says calls keep cutting out, a Douyin video reposts a screenshot of the Weibo post, and a Xiaohongshu user says switching to a laptop forces Bluetooth to reconnect. Save all three sources, link the Weibo post and the Douyin video as the same original report, and keep the Xiaohongshu note as a separate experience. Check the headset model, OS version and trigger before deciding whether the three sources belong in the same follow-up item.

## Which comments come next?

Read comments for the selected objects with [Weibo comments](/faq/weibo-comment-research), [Douyin comments](/guides/douyin-video-comments) and [Xiaohongshu comments](/guides/xiaohongshu-note-comments). Start with one batch and add reply context where a comment is ambiguous. Comment requests are budgeted separately from search requests.

Example instruction for the assistant:

> Research this headset on Weibo, Douyin and Xiaohongshu, initially one batch each. Preserve conditions and sources, remove namesakes and group actual connection, comfort and calling excerpts. Keep engagement platform-specific, list missing times/context and explain which comments deserve more reading.

Plan terms, platforms and actual pages with current [pricing](https://beatapi.io/pricing). For regular observation, continue with the [brand-discussion review](/guides/cross-platform-brand-review).