> 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 can Xiaohongshu product research surface improvements worth testing?

> Search Xiaohongshu products, read buyer reviews for comparable SKUs, then check note discussions to turn complaints into product questions worth testing.

To find product improvements worth testing, search products with `xiaohongshu.app_v2.search_products`, read buyer reviews for comparable SKUs with `xiaohongshu.app_v2.get_product_reviews`, then read note discussions to see under which conditions each complaint appears. The result is a list of product questions, each tied to the original review or comment. The example below researches winter hats and complaints about fit, scratchy fabric or warmth.

## Select comparable products, then retrieve reviews

Send both requests to `POST https://api.beatapi.io/v1/social-data/call` with BeatAPI Bearer authentication. Start with one page of product results from `xiaohongshu.app_v2.search_products`:

```json
{"action":"xiaohongshu.app_v2.search_products","params":{"keyword":"冬季帽子","page":1,"search_id":"","source":"explore_feed"}}
```

Check the products, variants and sources in the search response. For the next search page, pass the returned search identifier and the next page number. Select products whose audience, material and style match the research. Record price, units sold and variants only when the response confirms them. Before requesting reviews, confirm the SKU identifier that `xiaohongshu.app_v2.get_product_reviews` expects as `sku_id`:

```json
{"action":"xiaohongshu.app_v2.get_product_reviews","params":{"sku_id":"<confirmed_sku_id>","page":0,"sort_strategy_type":1,"share_pics_only":0,"from_page":"score_page"}}
```

Review pages start at `page:0`. Set `sort_strategy_type` to 1 for the latest reviews or to 0 for general sorting. To read only reviews with photos, set `share_pics_only` to 1 and record that sampling condition. Request further pages based on the actual responses. Field paths for price, sales and review content are not yet registered, so keep the original responses and map the fields you actually receive.

## Turn review evidence into a product question

A review such as “It is warm walking, but my ears get cold when cycling” describes a usage context. For each review, record the original words, the variant, the usage conditions, the experience, the date and the source. When an issue recurs, report how many products and reviews you examined. Record packaging, delivery and product experience as separate issues.

Group buyer reviews and note comments under one product question, such as “ear flap length,” and keep the source of each item separate:

| What the source says                                          | What to check                                                                    |
| ------------------------------------------------------------- | -------------------------------------------------------------------------------- |
| A buyer says their ears got cold while cycling                | Check the reviewed SKU, flap length and riding conditions                        |
| A note commenter asks about fit under a helmet                | Record a pre-purchase question and test the relevant variant with a helmet       |
| A review for another variant says it feels warm while walking | Keep the variant and use case before deciding whether the reports are comparable |

Then use [note search](/guides/xiaohongshu-notes-research) to collect questions people ask before buying. Keep buyer reviews and note discussions as two separate samples, and tie each finding to its source. The deliverable is a concrete question to test next, such as “Do longer ear flaps work better for cycling?” Budget calls as product search pages plus review pages for selected SKUs; current prices are on the [pricing](https://beatapi.io/pricing) page.