> 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 choose recent Xiaohongshu videos or image posts for content research?

> Choose time, engagement order and content format for a specific research question, then screen and interpret the results before handing over a source list.

To search Xiaohongshu notes from the past week, call `xiaohongshu.app_v2.search_notes` with `time_filter:"一周内"`. Set `sort_type:"collect_descending"` to order by saves or `"comment_descending"` to order by comments. Pass `note_type:"视频笔记"` for videos or `"普通笔记"` for photo posts. Start each sort order on page one.

For a beginner’s camping coffee tutorial, search ordered by saves to find useful examples, then review the selected notes. To see where users get stuck, run a separate search ordered by comments and read the discussions under the selected notes.

## Choose the question before the filters

| What you need                           | First search                                         | What to check next                                                                                                |
| --------------------------------------- | ---------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------- |
| Useful checklist or tutorial structures | Last week, ordered by saves, in your intended format | Does the post answer a beginner's question with usable steps? Saves do not prove use or purchases.                |
| Material that draws discussion          | Same keyword and time window, ordered by comments    | Choose relevant posts, then read comments and replies; totals do not reveal questions, agreement or disagreement. |
| Recently published angles               | Newest first                                         | Look for changes in questions and framing; recent publication does not establish a recent surge.                  |

If you make videos, study video topics and narration first; for image posts, study how the information is arranged. Keep a separate query record for each format when comparing formats. Positions from different search conditions do not form one ranking.

## Turn a page into a useful handover

Check each title and excerpt for relevance to beginner camping coffee before using engagement as a signal. A popular cafe check-in can be excluded; a modestly engaged campsite-equipment explanation can still be useful. The two cases are screening examples, not findings from a real search.

Fill `selected` and `selection_reason` in the [content research table](/guides/xiaohongshu-notes-research). For each source, explain why the source matters, what to study and what remains unverified. Retrieve detail descriptions only where excerpts are insufficient. To explain what readers ask, continue to [comment research](/guides/xiaohongshu-note-comments). A retrieved description does not mean a person has read the complete text.

For separate saves and comments samples, keep each query's filters, time and source, then deduplicate candidates by note ID. Run the exporter once per search; do not mix both manifests into one run. A post that appears in both samples is still one post, not two independent pieces of evidence.

## What should I change when the result is unhelpful?

* Mostly irrelevant results: narrow the keyword or research question before buying more pages.
* Few relevant results: check whether time and format together are too restrictive. Change one condition at a time in a new run and record the change; an empty result does not establish that no discussion exists.
* The first item does not have the highest count: keep the item's search position and compare exported saves or comments within the retrieved sample. Keep missing values missing, and do not claim a platform-wide maximum.

The `sort_type` ordering is a platform search request, not a strict numerical guarantee: the saved comments-ordered test page was not strictly decreasing by comment count. A local sort still ranks only the retrieved sample. Posts published in the same week can have different ages and accumulation periods; totals do not measure growth speed or content quality.

## Which parameters set time, order and format?

The one-page example below requests camping-coffee videos from the last week, ordered by saves. First Inspect `data:xiaohongshu.app_v2.search_notes` for current inputs, readiness and pricing. Each page is another paid query; one saves page plus one comments page means two queries. Do not fetch details or more pages automatically.

```bash
curl -sS https://api.beatapi.io/v1/social-data/call \
  -H "Authorization: Bearer $BEATAPI_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{"action":"xiaohongshu.app_v2.search_notes","params":{"keyword":"露营咖啡","page":1,"time_filter":"一周内","sort_type":"collect_descending","note_type":"视频笔记"}}' \
  -o xiaohongshu-search-page-1.json
```

| Selection                    | Parameter value                                                                  |
| ---------------------------- | -------------------------------------------------------------------------------- |
| Last day / week / six months | `time_filter: 一天内` / `一周内` / `半年内`                                               |
| Any publication time         | `time_filter: 不限`                                                                |
| Saves / comments / likes     | `sort_type: collect_descending` / `comment_descending` / `popularity_descending` |
| Newest                       | `sort_type: time_descending`                                                     |
| Videos / image notes / both  | `note_type: 视频笔记` / `普通笔记` / `不限`                                                |

Keep the Chinese literal values in an English client; do not translate `视频笔记` into `video`. Start at page 1. For a further page, keep the first response's `search_id`, `search_session_id` and actual next-page information. Changing the keyword, ordering or a filter starts a new session at page 1. The [search guide](/guides/xiaohongshu-notes-research) covers saving and exporting.

Sources: on 2026-11-16, one comments-ordered page was requested and 20 notes were exported, with the list at `data.data.items[].note`. The week/video/saves combination was checked against the backend input contract only. Page size and complete results are not guaranteed.

## Compare one search topic across Douyin, Xiaohongshu and Channels

For a topic such as rental storage, choose a Xiaohongshu note sort order from the table above, use [Douyin video search](/guides/douyin-python-research) for videos ordered by likes, and use [WeChat video search](/guides/wechat-topic-search) for popular or recent work. Record each platform’s actual time window and page scope, then select relevant works within each platform's sample. Keep the original sort order, available metrics, observation time and selection reasons. Combine themes and quotations across platforms, but keep each platform’s counts and ranks separate.