If you store OLX adverts in your own system, you need to know when their status changes on OLX — for example when an advert becomes active after moderation, expires, or is deleted. This article shows how to stay up to date by asking OLX only for adverts that changed, instead of downloading your whole advert set over and over.

### 1. Common approach: fetch everything [#1-common-approach-fetch-everything]

Many integrations sync statuses like this: every few minutes they page through `GET /api/partner/adverts`, fetch every advert (including inactive ones), and compare each one with their own data.

That works, but almost all of that work is wasted. Most of your adverts have not changed since the last check, and every run still downloads them all again.

**Illustrative example**

Imagine you have 5000 adverts and get 100 adverts per page. One full check needs 50 requests. If you repeat that every 5 minutes, that is 600 requests per hour — even if only 20 of your adverts changed in that time. These numbers are examples only, to illustrate the problem.

Fetching everything means:

- long syncs that load your servers
- a large amount of data to download, store, and compare
- a higher risk of hitting request limits (see [FAQ](/docs/faq))
- late detection of changes, because a full check of a large advert set takes time

It also loads OLX unnecessarily, because the same unchanged adverts are served again and again.

### 2. Better approach: advert change feed [#2-better-approach-advert-change-feed]

Instead of asking “give me all my adverts”, ask: “which of my adverts changed since the last check?”. That is what the advert change feed answers:

```
GET /api/partner/status/adverts
```

The feed returns only identifiers of adverts that changed, starting with the oldest change. For each identifier you fetch the current advert with `GET /api/partner/adverts/{advertId}` and update its status in your system. If nothing changed, you get an empty list — one small request and you are done.

It is like a mailbox: you do not re-read every email to see what is new — you look only at new messages.

### 3. Before and after [#3-before-and-after]

Same example as above (illustrative data: 5000 adverts checked every 5 minutes, 20 adverts changed in an hour):

|                      | Fetch everything           | Advert change feed                                 |
| -------------------- | -------------------------- | -------------------------------------------------- |
| Requests per hour    | 600                        | 12 feed requests + 20 advert requests              |
| What you download    | All 5000 adverts, 12 times | A short list of identifiers and 20 changed adverts |
| When nothing changed | Full download anyway       | One request with an empty list                     |
| Work on your side    | Compare every advert       | Update only changed adverts                        |

### 4. How the change feed works [#4-how-the-change-feed-works]

The feed request takes these parameters:

| Parameter | Required | Description                                                                                                                                                                                                                     |
| --------- | -------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `since`   | Yes      | Moment from which you want to read changes, as an ISO-8601 datetime with timezone, for example `2026-05-11T10:00:00Z`. You receive changes recorded strictly after that moment. Send the same value on every following request. |
| `cursor`  | No       | The `nextCursor` value from the previous response. It tells the feed where the previous read ended. Omit this parameter only when you start reading from the `since` moment.                                                    |
| `limit`   | No       | Maximum number of identifiers on one page. Default 100; values above 1000 are treated as 1000.                                                                                                                                  |

Every response contains the same three fields:

- `data` — identifiers of adverts that changed, starting with the oldest change. The list is empty when nothing changed.
- `hasMore` — `true` when more changes wait on the next page.
- `nextCursor` — your bookmark. It is always present, even when `data` is empty.

Full contract is in the [API Reference](/api-reference).

### 5. Synchronization step by step [#5-synchronization-step-by-step]

**Step 1: Start reading the feed**

You already have your adverts in the system, for example from a one-time full download. For `since`, use a moment just before that download started, so you do not miss any change. Send the request without a cursor:

<SiteCode
    lang="bash"
    code={`curl -H "Authorization: Bearer <token>" \
  -H "Version: 2.0" \
  "https://{SITE_HOST}/api/partner/status/adverts?since=2026-05-11T10:00:00Z"`}
/>

```json
{
    "data": ["1024578390", "1024578412"],
    "hasMore": true,
    "nextCursor": "MXwyMDI2LTA1LTExVDEwOjA0OjEyLjMzMVp8MTAyNDU3ODQxMg"
}
```

**Step 2: Read remaining pages**

While `hasMore` is `true`, send the same `since` together with the received `nextCursor` as `cursor`:

<SiteCode
    lang="bash"
    code={`curl -H "Authorization: Bearer <token>" \
  -H "Version: 2.0" \
  "https://{SITE_HOST}/api/partner/status/adverts?since=2026-05-11T10:00:00Z&cursor=MXwyMDI2LTA1LTExVDEwOjA0OjEyLjMzMVp8MTAyNDU3ODQxMg"`}
/>

```json
{
    "data": ["1024578455"],
    "hasMore": false,
    "nextCursor": "MXwyMDI2LTA1LTExVDEwOjA1OjAxLjEwMlp8MTAyNDU3ODQ1NQ"
}
```

**Step 3: Fetch only changed adverts**

For each identifier in `data`, fetch the current advert and update its status in your system:

<SiteCode
    lang="bash"
    code={`curl -H "Authorization: Bearer <token>" \
  -H "Version: 2.0" \
  "https://{SITE_HOST}/api/partner/adverts/1024578390"`}
/>

```json
{
    "data": {
        "id": 1024578390,
        "status": "active"
    }
}
```

**Step 4: Check for changes regularly**

When `hasMore` is `false`, your data is up to date. Store the last `nextCursor`. On the next check, send it as `cursor` (together with the same `since`). If nothing changed in the meantime, you get an empty list — still store the new `nextCursor`:

```json
{
    "data": [],
    "hasMore": false,
    "nextCursor": "MXwyMDI2LTA1LTExVDEwOjA1OjAxLjEwMlp8MTAyNDU3ODQ1NQ"
}
```

**Full loop in pseudocode**

```
since  = load("since")    # the moment your full download started
cursor = load("cursor")   # empty on the very first run

every few minutes:
    repeat:
        page = GET /api/partner/status/adverts?since=since&cursor=cursor
        for each advertId in page.data:
            ad = GET /api/partner/adverts/advertId
            save the status of advertId in your system as ad.status
        cursor = page.nextCursor
        save("cursor", cursor)
    until page.hasMore is false
```

### 6. Practical tips [#6-practical-tips]

- **Save the bookmark after every page.** Store `since` and the last `nextCursor` somewhere that survives a restart, for example in a database. After a pause or failure you resume exactly where you stopped — you miss nothing and you process nothing twice.
- **Use the cursor as-is.** Do not decode, shorten, or invent cursor values — always send back the exact value you received.
- **An empty list is good news.** It means nothing changed. Still store `nextCursor` from that response.
- **Changes appear with a small delay.** A change becomes visible in the feed a few seconds after it happens, so a very fresh change may only show up on the next check.
- **Pick a sensible page size.** A smaller `limit` means smaller responses; a larger one helps catch up faster after a longer pause.
- **Error handling.** A `400` response with code `validation_error` means the request is wrong — the `message` field points to which parameter. Fix the request instead of sending it again unchanged. If the cursor is rejected, start over by sending only `since` with the moment of the last successful sync.

### 7. Key takeaways [#7-key-takeaways]

- Do not download the whole advert set just to find a few that changed.
- Ask the change feed which adverts changed, then fetch only those adverts.
- Store `nextCursor` after every page — that is your bookmark.
- An empty list means no changes — one small request instead of a full download.

<SiteOnly site="olxpl">

The same approach applies to orders — see [Order Management](/docs/order-management) and the order change feed.

</SiteOnly>

More object types will follow later.
