What’s inside
The row-level export is bounded to the top 15,000 creatives by confirmed multi-day longevity, so the file stays a sane download size. The aggregate State of Native Advertising report is computed over the full 689,322 creatives observed in the window. By design, no thumbnail / image / video URL of any creative is included in any row — the dataset is FACTS about ads, never the copyrighted creative itself.
Schema
| Field | Type | Description |
|---|---|---|
| creative_id | string | Stable per-network creative id. |
| network | string | Source native network slug (taboola | outbrain | mgid | revcontent). |
| network_label | string | Human-readable network name. |
| advertiser | string|null | Observed advertiser / brand name. |
| advertiser_slug | string|null | URL slug for the advertiser page. |
| headline | string|null | Ad headline TEXT (the words, not the image). |
| vertical | string | Editorial vertical slug. |
| vertical_label | string | Human-readable vertical. |
| countries | string[] | ISO-3166 country codes where the creative was observed. |
| geo_count | number | Distinct country count. |
| ad_type | string|null | Observed ad format (text | image | video …). |
| first_seen | date | First observation date (YYYY-MM-DD). |
| last_seen | date | Most recent observation date (YYYY-MM-DD). |
| days_active | number|null | Active days within the 30-day window. |
| days_running | number | First-to-last observed span in days. |
| verdict | string | Longevity verdict: scaling | evergreen | active | burned-out | new (same logic as the live pages). |
| lp_host | string|null | Landing-page hostname the creative clicks through to. |
| detected_tracker | string|null | Detected tracking platform/network on the landing page, when identified. |
| variant_count | number | Count of detected sibling/variant creatives. |
| page | url | Link to the creative's page on mediabuyer.site. |
Sample record
One line of the JSONL export looks like this (a real record shape):
{"creative_id":"…","network":"taboola","network_label":"Taboola",
"advertiser":"…","advertiser_slug":"…","headline":"…",
"vertical":"general","vertical_label":"General",
"countries":["US","CA","GB"],"geo_count":3,"ad_type":"text",
"first_seen":"2026-04-12","last_seen":"2026-06-11",
"days_active":60,"days_running":60,"verdict":"evergreen",
"lp_host":"example.com","detected_tracker":null,"variant_count":5,
"page":"https://mediabuyer.site/spy/taboola/…"}Prefer to read the first lines directly? Open the JSONL file or the CSV.
By the numbers
A few figures this snapshot supports (full report linked below):
- Median confirmed creative lifespan: 10 days (across creatives observed running at least one day).
- Longest-lived network by median: RevContent at 25 days median.
- Share of creatives currently Scaling (fresh + live): 2.8%.
Read the full State of Native Advertising report →
License & how to cite
CC BY 4.0 (attribution). These are DERIVED FACTS about third-party ads, not the ads themselves. No raw creative images are included or redistributed. Facts are free to reuse with attribution; the underlying creatives remain their advertisers' copyright.
When you use these figures, please credit mediabuyer.site and link back to this page or the specific creative page. Suggested citation:
“MediaBuyer Native Ads — Open Derived Dataset, mediabuyer.site (observed Apr 11 – Jun 11, 2026).”
Methodology
Every field is derived from repeated public observation of served native-ad creatives — no spend, CTR, or ROI is claimed or inferred. The verdict (Scaling / Evergreen / Burned-out / Active / New) uses the same longevity logic as the live creative pages. Dates and the observation window are derived from the data, not from the build clock, so the export is reproducible from its inputs. Full method and definitions: methodology · data dictionary · limitations.