> 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 search the newest X posts instead of top results?

> Set search_type to Latest in twitter.web.fetch_search_timeline to search recent X posts instead of the default Top ordering, and keep Top and Latest cursor chains separate.

To search the newest X posts, call `twitter.web.fetch_search_timeline` with your search phrase in `keyword` and set `search_type:"Latest"` explicitly. Omitting `search_type` defaults to `"Top"`, which returns X's top search results. After a product launch, for instance, run a `Latest` search for new issue reports and a separate `Top` search for high-visibility discussions.

## Which search type fits my task?

| Task                                       | `search_type`                        | Keep in the deliverable                                                           |
| ------------------------------------------ | ------------------------------------ | --------------------------------------------------------------------------------- |
| Find recently discussed brand problems     | `Latest`                             | Original post ID, text, publication time, collection time, and follow-up evidence |
| Review what X's top search results discuss | `Top`                                | Original query and sort, response position, and supporting text                   |
| Compare both views                         | Separate `Top` and `Latest` searches | Independent manifests, never one mixed cursor chain                               |

`Top` is not a local likes-descending sort, and `Latest` is not a local timestamp sort. Preserve the order returned in each response; if you apply a custom local ranking later, label that ranking rule explicitly rather than describing custom rankings as official popularity.

## Save requests before deciding which ones to execute

The examples below use the phrase `weather forecast`. Replace `weather forecast` with your actual query and save the `Latest` request body as `request-latest.json`:

```json
{"action":"twitter.web.fetch_search_timeline","params":{"keyword":"weather forecast","search_type":"Latest"}}
```

Only when you also need the top-result view, save the `Top` request body separately as `request-top.json`:

```json
{"action":"twitter.web.fetch_search_timeline","params":{"keyword":"weather forecast","search_type":"Top"}}
```

Check current pricing with the free Inspect call before approving requests. One `Latest` page currently costs an estimated `$0.003`; fetching one page of each sort mode costs an estimated `$0.006`. Writing the JSON files locally makes no API call, whereas the `curl` command below executes a new billed request. Avoid running both sort modes just for presentation or turning a one-page budget into automatic pagination.

```bash
mkdir x-latest-run-1 &&
curl -sS https://api.beatapi.io/v1/social-data/call \
  -H "Authorization: Bearer $BEATAPI_API_KEY" \
  -H 'Content-Type: application/json' \
  --data-binary @request-latest.json \
  -D x-latest-run-1/headers.txt \
  -o x-latest-run-1/response.json
```

Record the actual collection time and HTTP status next to `x-latest-run-1`, then verify `status`, `action`, `request_id`, and `data.timeline` in `response.json` because a completed HTTP request does not guarantee a succeeded result. If `data.timeline` is empty, follow the [empty-search diagnosis](/faq/x-empty-search-results) before changing the query or `search_type` and paying for another call.

## Can I switch from Top to Latest while following a cursor?

No. A `data.next_cursor` returned by a `Top` search cannot continue a `Latest` search. To fetch the next page, send the exact same `keyword` and `search_type` while adding only the returned cursor. Changing either `search_type` or `keyword` starts a new search from page one without a cursor, with its own cost record, manifest, and output folder. See the [search CSV guide](/guides/x-search-posts) for collection and saved-response export.

Do not rerun collection just to repair a local CSV. Export the saved responses locally into two separate folders (`top-review` and `latest-review`). When the same `tweet_id` appears in both exports, both `Top` and `Latest` surfaced the same post rather than two distinct posts; match across exports by `tweet_id` and keep each observation's collection time and counters without letting one export overwrite the other.

## What should the final business brief contain?

For recent issues, list each post's original ID or link, a brief supporting quote, the issue category, the reason to follow up, and the source timestamp; a result count alone is not a brief. For `Top` observations, attach the selected posts as evidence. Differences between a `Top` page and a `Latest` page do not establish a platform-wide sentiment shift. Never convert `null` views into zero; see the [view-coverage report](/faq/x-missing-views).

Contract reference: `data:twitter.web.fetch_search_timeline`.