TikTok New User First Week Activity Statistics 2026
No official number exists for TikTok new user first week activity. TikTok publishes no Day 7 retention rate, no first-session length and no count of videos watched before the For You feed settles.
What does exist: controlled test-account audits from 2025, Appfigures download estimates through August 2026, and Adjust’s cross-app retention medians. This brief covers feed personalization, feed resets, teen and political test results, install volume and retention benchmarks.
TikTok New User First Week Activity Statistics
How Fast Does a Fresh For You Feed Learn a New User?
Northeastern University researchers created brand-new Android accounts, searched one topic and watched the first 25 results. Over the next 200 For You videos, cooking and sports betting each reached 89 videos, or 44.5%.
Fitness reached 76 videos (38%). Untouched baseline accounts saw 17 cooking videos (8.5%) and three each for fitness and sports betting (1.5%).
| Topic | Seeded account: topic videos in next 200 | Seeded share | Baseline: topic videos in 200 | Baseline share |
|---|---|---|---|---|
| Cooking | 89 | 44.5% | 17 | 8.5% |
| Fitness | 76 | 38% | 3 | 1.5% |
| Sports betting | 89 | 44.5% | 3 | 1.5% |
Source: Kaplan et al., Northeastern University, arXiv:2605.10690 (May 2026), Phase 1 run 17 March 2025
Each account pressed Skip on the nickname and interest screens at sign-up and refused contact syncing. No likes, no follows.
The bots watched topic videos to the end, a signal TikTok itself lists and the behaviour behind these TikTok rewatch rate benchmarks.
Everything else got 0.2 to 2 seconds before a swipe, the pattern tracked across the app in scroll-past rate data.
‘Not Interested’ Cut Topic Videos to 4.75% of a Trained Feed
Cloned accounts that kept watching cooking after the feed learned it saw cooking in 30.5% of their feed. Skipping cooking videos cut that to 16%, a 47.5% decrease.
Marking those videos Not Interested dropped the share to 4.75%, an 84.4% decrease. The authors found that level statistically no different from a non-personalized baseline.
The control sits inside the Share menu or behind a long press. The paper notes it may not be known to all users.
Source: Kaplan et al., arXiv:2605.10690, preliminary cooking test with two accounts per condition, spring 2025
How Often Did Feeds Relapse?
Each topic ran five times. Once accounts stopped signalling disinterest and watched the topic again, skip-only feeds snapped back more often than Not Interested feeds.
| Topic | Relapsed after skipping stopped | Relapsed after Not Interested stopped |
|---|---|---|
| Cooking | 5 of 5 runs | 2 of 5 runs |
| Fitness | 4 of 5 runs | 2 of 5 runs |
| Sports betting | 2 of 5 runs | 1 of 5 runs |
Source: Kaplan et al., arXiv:2605.10690, Phases 2 and 3 run 27 March to 22 April 2025
The bots only watched, skipped and tapped Not Interested. The authors state real users who like, comment or follow may not see the same results.
Teen Test Accounts: TikTok New User First Week Activity in France and the UK
France: Amnesty International, October 2025
| What appeared | Accounts | Time from first scroll |
|---|---|---|
| Sad or disillusioned videos, before any preference signalled | 3 of 3 | Within 5 minutes |
| Feed almost entirely mental-health videos, up to half depressive | 3 of 3 | 15 to 20 minutes |
| Videos expressing suicidal thoughts | 2 of 3 | Within 45 minutes |
| Videos romanticizing suicide | Test accounts (count not stated) | 3 to 4 hours |
Source: Amnesty International, “Dragged into the Rabbit Hole”, published 21 October 2025
Two accounts were registered as 13-year-old girls and one as a 13-year-old boy. Researchers let sad videos play and swiped past the rest, the same watch-and-skip logic Northeastern used.
Every milestone falls inside one first day. Separate automated tests with the Algorithmic Transparency Institute found the share of recommended sad or depressive content more than doubled depending on watch history.
UK: Global Witness, October 2025
Global Witness set up seven accounts as 13-year-olds on factory-reset phones, all with Restricted Mode on. Three ran in March and April 2025; four ran after the Online Safety Act’s child-safety duties took effect on 25 July 2025.
For three accounts, sexualised searches appeared the first time the user tapped the search bar. That panel suggests around 10 searches.
All seven reached pornographic content a small number of clicks after sign-up. For one account it sat two clicks away.
Search suggestions matter here, since plenty of younger people search inside the app, as TikTok search usage statistics show.
Global Witness first flagged the problem in January 2025, and TikTok said it removed several recommendations globally. After the 2025 tests, TikTok said it acted on more than 90 pieces of content.
Source: Global Witness, investigation published 3 October 2025, tests run March to summer 2025
Fresh German Accounts Got 78% Pro-AfD Party Content on TikTok
Before Germany’s February 2025 federal election, Global Witness opened new accounts on factory-reset devices. Each followed only the four main parties and their leaders, then viewed those eight accounts’ five latest posts.
Party-political posts recommended from accounts the testers did not follow leaned heavily one way on both platforms tested.
| Platform | AfD-supportive posts | Recommended party-political posts | Share |
|---|---|---|---|
| TikTok | 28 | 36 | 78% |
| X | 14 | 22 | 64% |
Source: Global Witness, published February 2025, updated 27 March 2025
AfD polled around 20% at the time, according to Global Witness. TikTok said the method did not reflect accounts that engage with a variety of topics.
How Many TikTok New User First Weeks Start Each Month?
Appfigures estimated 40 million TikTok downloads in August 2026, second worldwide behind ChatGPT’s 53 million. Its analysts described TikTok’s downloads as fairly stable, even after the US split.
| Month | Estimated TikTok downloads | Global download rank |
|---|---|---|
| February 2026 | 40 million | 2nd |
| May 2026 | 43 million | 2nd |
| August 2026 | 40 million | 2nd |
Source: Appfigures Intelligence, monthly top-apps reports for February, May and August 2026 (App Store and Google Play combined)
February 2026 was TikTok’s first full month under new US ownership, after an Oracle-led consortium took control in late January.
Downloads count estimated app installs, not new accounts or active users. Where TikTok’s existing audience sits is broken out in TikTok users by country.
TikTok New User First Week Retention Against Adjust App Benchmarks
TikTok has never released cohort retention. The nearest verified yardstick is Adjust’s cross-category median, published in April 2024: 26% of installers return on Day 1 and 13% on Day 7.
| Checkpoint after install | All apps | iOS | Android |
|---|---|---|---|
| Day 1 | 26% | 27% | 24% |
| Day 7 | 13% | 14% | 11% |
| Day 14 | 10% | 11% | 8% |
| Day 30 | 7% | 8% | 6% |
Source: Adjust, median retention across all verticals, published 16 April 2024
These are medians across all app categories, not TikTok figures. Games and social apps led Adjust’s verticals at 29% Day 1 retention.
Day 7 Retention by Region
| Region | Day 1 | Day 7 | Day 30 |
|---|---|---|---|
| APAC | 26% | 12% | 6% |
| Europe | 24% | 11% | 6% |
| LATAM | 26% | 11% | 5% |
| MENA | 24% | 10% | 4% |
| North America | 23% | 10% | 5% |
Source: Adjust, median retention by region across all verticals, published 16 April 2024
Social apps alone ran from 29% Day 1 retention in Europe to 23% in MENA. Daily use among people who stay is tracked in TikTok watch time per user.
FAQs
How long does it take TikTok’s algorithm to learn a new account?
TikTok publishes no official timeline. In Northeastern University’s March 2025 test, fresh accounts seeded with one topic saw it in 38% to 44.5% of their next 200 For You videos.
Does “Not Interested” work on a new TikTok feed?
Yes, in controlled tests. It cut cooking from 30.5% to 4.75% of a trained feed in spring 2025, though feeds still relapsed in some runs after accounts resumed watching.
What is a good Day 7 retention rate for an app?
Adjust’s April 2024 cross-category median was 13% on Day 7, with 14% on iOS and 11% on Android. TikTok has never published its own Day 7 figure.
Reddit users say TikTok boosts new accounts in their first week. Is there data?
No published data confirms it. TikTok releases no first-week reach figures, and the 2025 test-account audits covered here measured what new accounts were shown, not how far their own posts travelled.
How many people download TikTok each month?
Appfigures estimated 40 million TikTok downloads in August 2026, ranking second behind ChatGPT’s 53 million. Downloads measure app installs, not new accounts or active users.
Sources
https://arxiv.org/html/2605.10690
https://appfigures.com/resources/insights/most-downloaded-highest-earning-apps-august-2026
https://globalwitness.org/en/campaigns/digital-threats/tiktok-directs-13-year-olds-to-porn/
