PlayerSells Open Data
11 free datasets from PlayerSells: what X (Twitter) accounts sold for in completed escrow deals, and measurements by our own crawlers on X, Bluesky, Telegram, TikTok, YouTube, Kick and Twitch (follower count percentiles, X reach and engagement benchmarks, Telegram reach and subscriber change, Kick and Twitch viewing). Each one is a set of CSV files with a documented method, licensed under CC BY 4.0, archived on Zenodo and published on GitHub.
Collection DOI 10.5281/zenodo.23201221, source files on GitHub.
Which datasets are available?
Kick and Twitch share of viewing by category, week of 26 September 2026
Share of concurrent viewing that each streaming category took on Kick and on Twitch from 26 September to 2 October 2026, daily and for the whole week, measured with one method on both platforms. Gambling took 7.2% of Kick viewing and 1.7% of Twitch viewing on these readings.
- Measured
- 26 September 2026 00:00 UTC to 3 October 2026 00:00 UTC
- Files
- category_viewing_share.csv(18,181 rows)gambling_share_week.csv(5 rows)rank_comparison_30d.csv(6 rows)
- Analysis
- /learn/how-much-of-kick-is-gambling-one-week-measured-against-twitch
- Documentation
- Columns, method and limitations
How to cite: PlayerSells (2026). Kick and Twitch share of viewing by category, week of 26 September 2026. https://playersells.com/learn/how-much-of-kick-is-gambling-one-week-measured-against-twitch. Licensed under CC BY 4.0.
Kick and Twitch simulcasting, week of 28 September 2026
Streamers who broadcast on Kick and Twitch at the same time from 28 September to 4 October 2026, and how their concurrent viewers split between the two platforms. Twitch drew 83.3% of the combined viewers and Kick 16.7%.
- Measured
- 28 September 2026 00:00 UTC to 5 October 2026 00:00 UTC
- Files
- summary.csv(24 rows)kick_share_distribution.csv(4 rows)split_by_audience_size.csv(4 rows)top_channels_simulcasting.csv(5 rows)twitch_category_split.csv(190 rows)kick_category_simulcast_share.csv(12 rows)
- Analysis
- /learn/kick-and-twitch-simulcasting-measured-where-viewers-watch
- Documentation
- Columns, method and limitations
How to cite: PlayerSells (2026). Kick and Twitch simulcasting, week of 28 September 2026. https://playersells.com/learn/kick-and-twitch-simulcasting-measured-where-viewers-watch. Licensed under CC BY 4.0.
Telegram channel subscriber change by topic, August to October 2026
Which Telegram channels gained or lost subscribers between early August and early October 2026, by topic and starting size. 85.6% of 814 crypto channels lost subscribers, against 64.6% of all 14,874 channels in the panel.
- Measured
- First reading 1 to 7 August 2026, last reading 26 September to 2 October 2026
- Files
- subscriber_change_by_category.csv(7 rows)shrink_rate_by_size.csv(12 rows)
- Analysis
- /learn/crypto-telegram-channels-are-losing-subscribers-measured
- Documentation
- Columns, method and limitations
How to cite: PlayerSells (2026). Telegram channel subscriber change by topic, August to October 2026. https://playersells.com/learn/crypto-telegram-channels-are-losing-subscribers-measured. Licensed under CC BY 4.0.
Telegram channel reach per post by size and niche, October 2026 edition
How many views a Telegram channel's average post gets per subscriber, for every public broadcast channel with at least 1,000 subscribers whose page PlayerSells read in the 60 days to 3 October 2026. The typical channel with 1,000 to 5,000 subscribers reaches 27.5% of its subscribers per post; above 500,000 it is 3.1%.
- Measured
- 3 October 2026 (pages read 4 August to 3 October 2026)
- Files
- reach_by_size.csv(6 rows)reach_by_niche.csv(18 rows)
- Analysis
- /telegram-engagement-rate
- Documentation
- Columns, method and limitations
How to cite: PlayerSells (2026). Telegram channel reach per post by size and niche, October 2026 edition. https://playersells.com/telegram-engagement-rate. Licensed under CC BY 4.0.
X (Twitter) engagement rate benchmarks and posting pattern effects, October 2026
Engagement rate percentiles for 152,291 measured X accounts, broken down by follower band, and 45 posting pattern effects (hour, weekday, length, hashtags, links, media) measured within accounts with 95% intervals.
- Measured
- Trailing 30-day windows ending 6 and 7 October 2026 (see the files)
- Files
- engagement_rate_percentiles.csv(6 rows)engagement_rate_by_follower_band.csv(8 rows)posting_pattern_effects.csv(45 rows)
- Analysis
- /insights/engagement-rate-benchmarks/insights/insights/data.json
- Documentation
- Columns, method and limitations
How to cite: PlayerSells (2026). X (Twitter) engagement rate benchmarks and posting pattern effects, October 2026. https://playersells.com/insights/engagement-rate-benchmarks. Licensed under CC BY 4.0.
Views on X (Twitter) posts with a link against the same account's posts without one, 2026
How many views X (Twitter) posts with a link drew compared with the same account's posts without a link, measured on 6,447,046 original posts from 285,579 accounts published between 15 July and 7 October 2026. For the median account, text posts with a link drew 10.4% fewer views; for accounts with over 1 million followers, 30.4% fewer; for accounts under 1,000 followers, 16.6% more.
- Measured
- Original posts published 15 July to 7 October 2026, views read at least 48 hours after posting
- Files
- link_gap_overall.csv(2 rows)link_gap_by_follower_band.csv(5 rows)link_gap_by_category.csv(12 rows)link_gap_by_period.csv(3 rows)engagement_per_view_change.csv(4 rows)link_placement.csv(4 rows)
- Analysis
- /learn/does-x-still-penalize-links-we-measured-6-million-posts
- Documentation
- Columns, method and limitations
How to cite: PlayerSells (2026). Views on X (Twitter) posts with a link against the same account's posts without one, 2026. https://playersells.com/learn/does-x-still-penalize-links-we-measured-6-million-posts. Licensed under CC BY 4.0.
Prices X (Twitter) accounts sold for in completed escrow deals, 2026
What X (Twitter) accounts actually sold for in 187 completed PlayerSells escrow deals between 21 May and 19 October 2026: final prices paid, not asking prices. The median sale was $60 and half of all sales closed between $19 and $125. The median price per 1,000 followers fell from $35.48 at 1,000 to 2,500 followers to $6.93 at 20,000 and up. 69.5% of accounts sold at their asking price.
- Measured
- Completed sales 21 May to 19 October 2026, read 20 October 2026
- Files
- sale_prices_by_follower_band.csv(6 rows)discount_vs_asking_price.csv(6 rows)asking_vs_sale_price.csv(6 rows)sell_through_by_asking_price.csv(5 rows)sales_by_price_bracket.csv(6 rows)blue_check_by_follower_band.csv(4 rows)price_by_account_age.csv(3 rows)
- Analysis
- /learn/what-x-accounts-actually-sell-for-2026-escrow-data/pricing/x-account-prices-2026
- Documentation
- Columns, method and limitations
How to cite: PlayerSells (2026). Prices X (Twitter) accounts sold for in completed escrow deals, 2026. https://playersells.com/learn/what-x-accounts-actually-sell-for-2026-escrow-data. Licensed under CC BY 4.0.
YouTube public view gains before and after first-frame view counting, 2026
How fast 3,565 large YouTube channels gained public views in the 28 days after YouTube began counting views from the first frame of playback on 24 August 2026, against the 28 days before. The median channel gained views 12.5% faster, against -3.5% and +2.1% in comparisons made entirely before the change, and total views gained by the same channels fell 8.3%.
- Measured
- Daily public view counters, 6 July to 19 October 2026, read 23 October 2026
- Files
- comparisons.csv(7 rows)weekly_index.csv(12 rows)change_by_subscribers.csv(3 rows)change_by_topic_and_language.csv(9 rows)
- Analysis
- /learn/did-youtubes-new-view-count-inflate-views-3565-channels-measured
- Documentation
- Columns, method and limitations
How to cite: PlayerSells (2026). YouTube public view gains before and after first-frame view counting, 2026. https://playersells.com/learn/did-youtubes-new-view-count-inflate-views-3565-channels-measured. Licensed under CC BY 4.0.
Do streamers keep simulcasting? Kick and Twitch simulcasters over four weeks, 2026
The same Kick and Twitch streamers followed for four weeks, 28 September to 26 October 2026. 10,328 streamers were live on both platforms at once, under the same name and in the same category, for at least an hour in at least one week. 49.3% did it in only one week and 15.4% in all four; from one week to the next, 42.3% stopped. The streamers who simulcast every week carried 71.8% of all simulcast viewing.
- Measured
- 15-minute live samples, weeks of 28 September to 26 October 2026 (UTC), read 27 October 2026
- Files
- weekly_series.csv(4 rows)week_to_week.csv(3 rows)persistence_by_group.csv(10 rows)persistence_by_kick_share.csv(4 rows)weeks_simulcasting.csv(4 rows)after_stopping.csv(4 rows)same_streamer_test.csv(3 rows)
- Analysis
- /learn/do-streamers-keep-simulcasting-kick-and-twitch-four-weeks-measured
- Documentation
- Columns, method and limitations
How to cite: PlayerSells (2026). Do streamers keep simulcasting? Kick and Twitch simulcasters over four weeks, 2026. https://playersells.com/learn/do-streamers-keep-simulcasting-kick-and-twitch-four-weeks-measured. Licensed under CC BY 4.0.
How can I use and credit the data?
Under CC BY 4.0you may copy, adapt and redistribute the data, commercially or not, including in research, journalism and AI-generated answers. Credit "Source: PlayerSells (https://playersells.com)" with a link, and say if you changed the numbers. To cite the whole collection, use its Zenodo DOI 10.5281/zenodo.23201221, which always points to the latest edition.
Questions, corrections or a request for a different cut of the data: [email protected].
Frequently asked questions
Can I use this data in an article, a report or an AI answer?
Yes. Every dataset is licensed under CC BY 4.0: you may copy, adapt and redistribute it, commercially or not, as long as you credit "Source: PlayerSells (https://playersells.com)" with a link and say if you changed the numbers.
Where do the numbers come from?
Most come from PlayerSells' own crawlers, which read public profiles, public channel previews and public live directories on X, Bluesky, Telegram, TikTok, YouTube, Kick and Twitch; they describe what those crawlers read, not a census of any platform. The X account sale prices come from completed escrow deals on the PlayerSells marketplace. Each dataset's README lists its sampling limits.
Is there personal data in the files?
No. The files hold totals, shares and percentiles only: no account lists, no channel names and no personal data.
How do I cite a dataset?
Use the line under "How to cite" for each dataset on this page, or cite the whole collection with its Zenodo DOI, 10.5281/zenodo.23201221, which always resolves to the latest edition.
Where are the live numbers?
Several datasets have a live counterpart on the site that updates from the same crawlers: the X insights pages, the follower rank tools such as the X follower rank checker, and the Telegram engagement rate benchmark. Sale prices from completed escrow deals are on X account prices 2026.