Research brief · Updated 22 November 2026

How culture actually shifts, and who moves it

Viral content rarely changes a society on its own. Lasting change comes when three things line up: pressure that unsettles the old way, small groups who already live the new way, and trusted networks that carry it across a tipping point. This brief gathers the best evidence we know on each, with its limits.

5.81B

social media user identities worldwide, about seven in ten people.

DataReportal, Oct 20265
70%

of people hesitate to trust anyone with different values or backgrounds.

Edelman Trust Barometer 2026, n = 33,9384
~25%

committed minority that tipped a group convention in a controlled experiment.

Centola et al., Science, 20181
50–65%

the threshold reported for an everyday habit with real costs, such as reusable cups.

Reviewed in Özden, 20236
The mechanics

Three forces have to line up

Frank Geels' Multi-Level Perspective, the most used model of how whole systems change, describes change as an alignment across three levels.3 Network research adds the fourth ingredient: trusted ties that carry a new practice across a threshold.1

Landscape

Pressure from above

Slow, large shifts that no single actor controls: demographics, climate shocks, economic crises, pervasive AI. They weaken the old way of doing things.

Niches

Alternatives from below

Protected spaces where lead users and innovators test new practices. When they mature into a workable design, they can replace what is there.

Networks

Trust that carries it across

A new practice spreads through people who already trust each other. Below a critical mass it stays a niche; above it, it becomes normal.

Why surface campaigns fade

Sohail Inayatullah's Causal Layered Analysis looks at four layers beneath every signal. A campaign that only touches the headline tends to fade when the systems, worldviews and myths underneath stay the same.

When incentives change and the deeper story is reframed, a niche practice can cross over and settle as a new norm.9 That is why every signal on this site is paired with a solution or next step.

How we use these layers in our method →

  1. HeadlineThe visible event or trend people talk about.
  2. SystemsLaws, algorithms and incentives that produce it.
  3. WorldviewsBeliefs that keep those systems in place.
  4. Myths & metaphorsThe deep stories that anchor identity.
Interactive

How big must a committed minority be?

The famous "25% rule" comes from a real experiment, but it was a naming game with a ten-cent reward. Move the slider to see how the same minority fares in four very different settings.

25%

Arbitrary convention

Lab naming game. Largest failed minority 21%; 24.3% predicted for large groups.1

Likely tips

Workplace norm

Organisational studies suggest ~30%; 31% gave full consensus in the lab.1

Could tip

Across social networks

Meta-analysis: average 24% (models) and 27% (empirical), range 10–43%.9

Could tip

Entrenched habit with costs

Reusable cups instead of single-use: 50–65%. Identity and infrastructure add friction.6

Holds

In the lab, minorities of 25–27% moved 72–100% of the group to the new convention. Real habits with costs have not moved yet.

Habits are entrenched

Eating meat, driving or flying involve identity and infrastructure that a naming game does not.

Switching often costs money

Lab subjects were paid to coordinate. In real life, early adopters of electric cars or plant-based diets often pay more.

Networks are many-to-many

Norms form through families, feeds and broadcasters at once, not in one-on-one pairs.

Takeaway: treat ~25% as a lower bound for small groups with aligned incentives. For costly habits, plan for 50% or more, or change the defaults so the new norm becomes the easy path.

Values research

Values move with security, in both directions

The World Values Survey and European Values Study have tracked values since 1981. The 2023 Inglehart–Welzel map plots 111 societies on two axes.2

When security and personal agency grow, societies drift towards secular-rational and self-expression values. When they shrink, the drift reverses. National averages also hide large differences inside countries.

Two dimensions of cultural change

What pushes a society along each axis

Inglehart–Welzel cultural map axes Horizontal axis from survival values to self-expression values. Vertical axis from traditional values to secular-rational values. Rising security moves societies up and to the right; insecurity moves them back. Survival Self-expression Traditional Secular-rational More security & agency Shocks & insecurity

Traditional → secular-rational correlates 0.65 with the industrial share of the economy. Source: WVS/EVS2

The 2026 baseline

More connected than ever, and moving fast between platforms

Where attention goes shapes which ideas get a hearing. In 2026 it is shifting to short video, professional networks and AI assistants.

19h 18m

a week on social and video platforms per online adult, across 6.66 platforms a month.

GWI Q2 2026 via DataReportal5
1B+

monthly users of generative AI. ChatGPT alone had 900M weekly users in Feb 2026.

DataReportal; Wikipedia1128
16%

of under-35s already use AI chatbots as a source of news.

Reuters Institute DNR 20267
>1/3

of Netflix viewing hours (H1 2026) were non-English titles, up from under 10% a decade ago.

Netflix data via Resourcera27

Who gained and lost reach

Change in advertising reach, Oct 2025 → Oct 2026

  • Growth
  • Decline
  • Pinterest+76.8%
  • TikTok+17.4%
  • LinkedIn+16.6%
  • Facebook−1.1%
  • X−22.3%

Ad reach measures addressable audiences, not unique users. DataReportal via SearchLab12

Show as table
PlatformChangeReach Oct 2026
Pinterest+76.8%621M
TikTok+17.4%2.33B
LinkedIn+16.6%1.55B
Facebook−1.1%2.32B
X−22.3%433M

Where audiences actually engage

Organic brand engagement rate, 2025

  • TikTok3.70%
  • Instagram0.48%
  • Facebook0.15%
  • X~0.11%

Socialinsider benchmarks via SearchLab12. TikTok also turned attention into $64.3B of TikTok Shop sales in 2025 (Momentum Works estimate, +94%).

Show as table
PlatformEngagement
TikTok3.70%
Instagram0.48%
Facebook0.15%
X0.10–0.12%
The trust paradox

Reach is not the same as influence

Edelman describes 2026 as a move "from a crisis of grievance to a crisis of insularity".4 In that climate, top-down messages bounce off. What still works is a trusted person vouching for something.

62%

of people who trust a food or lifestyle influencer would consider trusting a company they now distrust if that voice vouched for it. For finance influencers: 57%.

Edelman Trust Barometer 20264
<40%

of people globally trust the news they see. Social and video beat news sites as a source in 30 of 48 markets.

Reuters Institute DNR 20267
27%

get news weekly from individual news creators. In Kenya it is 58%.

Reuters Institute DNR 20267
32%

expect the next generation to be better off, with 13-point drops in a year in India and China.

Edelman Trust Barometer 20264

The biggest accounts carry a lot of estimated noise

TikTok followers, Nov 2026, split by estimated authentic and fake share

  • Estimated authentic
  • Estimated fake or bot
  • Khaby Lame163.0M
  • Charli D'Amelio160.3M
  • MrBeast145.9M
  • Bella Poarch91.5M
  • Addison Rae87.6M

Follower counts: Wikipedia list26. Fake shares are an automated vendor estimate (Collabstr)8; platforms do not publish audited bot rates.

Show as table
AccountFollowersEst. fake
Khaby Lame163.0M28.72%
Charli D'Amelio160.3M26.97%
MrBeast145.9M26.97%
Bella Poarch91.5M26.17%
Addison Rae87.6M26.18%
What this means for us

We list voices for early sight and trust, not follower counts

Mega-creators are excellent at fast, global awareness. Changing a costly habit or a civic norm usually travels through people an audience already trusts on that topic: a local journalist, a practitioner, a community leader.

That is why each voice in our directory passes five tests, including early sight of change, bridging fields and credible, open sourcing.

See the five tests

A working model

Four tiers of influence, each moving something different

Our synthesis of platform data, regional creator markets and foresight research. The bars are our qualitative judgement of each tier's typical reach and the depth of trust it can mobilise.

Tier 1Format pioneers

Global mega-creators

ReachTrust
Moves
Slang, aesthetics, entertainment formats and fast fads across borders.
Struggles with
Complex civic or economic behaviour; high estimated bot share.

e.g. Khaby Lame, MrBeast, Charli D'Amelio, BTS, Willie Salim, Kimberly Loaiza26

Tier 2Adoption engines

Local creator pods

ReachTrust
Moves
Everyday adoption: fintech, farming tech, thrift fashion. One anchor creator, 3–10 micro, 15–30 nano creators and WhatsApp Status networks.
Watch for
Disclosed AI virtual hosts joining pods.

e.g. Mr Macaroni, Taaooma, Elsa Majimbo, Azziad Nasenya (Influencer Marketing Africa)29

Tier 3Accountability anchors

News creators & investigative media

ReachTrust
Moves
Political awareness and early warning grounded in field reporting.
Watch for
Creator-only news diets are more partisan than mixed diets.7

e.g. Rappler, Balkan Insight, Agência Pública, TechCabal, Down To Earth (all on our live radar)

Tier 4System shifters

Foresight voices & bridge builders

ReachTrust
Moves
How problems are framed, and the institutional blueprints that follow.
Example
Jonathan Haidt's framing of youth social media harm is widely credited with shaping Australia's under-16 law.13

e.g. Jonathan Haidt, Audrey Tang, Kate Raworth, Amy Webb, Ian Bremmer

Names are examples of each tier, not endorsements. Inclusion here does not mean a person is listed in, or has contributed to, our directory.

Live case studies, 2024–2026

Three shifts happening now, and the friction they meet

Each shows a different mix of ideas, shocks, rules and money. Each also shows that change is never one-way.

Australia's under-16 social media law spreads worldwide

4.7M under-16 accounts removed on 10 platformsA$99M max penalty after Sept 2026 amendment12+ countries advancing similar rules

Catalysts

Jonathan Haidt's research and books, whistleblower Frances Haugen, and state leaders in South Australia and New South Wales.13

Mechanism

Online Safety Amendment (Social Media Minimum Age) Act 2024, in force 10 December 2025, enforced by the eSafety Commissioner.15

Outcomes

Accounts of about 2.5M young people aged 8–15 removed; early studies report 9–40% drops in under-16 use or account ownership.1433 Canada, Brazil, France, Spain, the UK and Indonesia are among those following.

Friction

Teens move to smaller apps or browse logged out. Reviewers caution that success means less total screen time and better mental health, not just deleted accounts.16

Lesson: a clear reframing (from parenting failure to collective problem) plus a law with real penalties moved a norm across a whole market within a year.

Who shapes culture

Four kinds of organisation finance, forecast and distribute change

Grassroots chatter rarely shifts culture alone. These are the main levers, and the latest figures we could source.

Trend forecasters

Translate niche signals into products, 1–10 years ahead.

  • WGSNBought by Apax for up to £700M (2024). 6,500+ brands subscribe, so it helps coordinate demand, not only predict it.23
  • Future Today Strategy GroupAmy Webb's firm; Convergence Outlook 2026.24
  • The Future Laboratory, sparks & honey, StylusVisual, open and cross-industry trend intelligence.

Public & open foresight

Shape legislative horizons and civic readiness.

  • Policy Horizons Canada, OECD, European CommissionHorizon-scanning manuals used worldwide.
  • CSF Singapore, WRR, NestaNational foresight and policy stress-testing.
  • IFTF, Forum for the FutureNon-profit, multi-stakeholder long-horizon programmes.3132

Philanthropy & narrative change

Fund values-driven change and the people who tell its stories.

  • Gates Foundation$77.2B endowment (end 2024); $8.74B budget in 2025; 11,750+ partners.25
  • Ford & Open Society FoundationsCivil society, rights and democratic resilience.
  • Pop Culture Collaborative, Narrative Initiative, FrameWorks, Norman Lear CenterIn a 2021 network study of 76 groups, 90% said their main contribution was connecting people.22

Platforms & regulators

Set the defaults: ranking, money, access.

  • YouTube, TikTok, MetaAd reach 2.77B, 2.33B and 2.32B (Facebook), Oct 2026.12
  • Netflix, OpenAI325M+ paid memberships; ChatGPT 900M weekly users.2728
  • eSafety Commissioner (Australia)Showed that a law with large penalties can reset youth access norms across a market.15
Put it to work

Five principles for anyone trying to see or shape change

For organisations, policymakers, funders and researchers. Each one answers a common way strategies fail.

  1. Secure livelihoods before asking people to change values

    Values research and the greenlash show sustainability norms retreat when households feel insecure. Lower switching costs; don't rely on moral persuasion.

    Do this: stress-test your policy for who pays first, with Foresight Policies.

  2. Scan the edges, not the leaderboard

    Following only top accounts and headlines means late detection and a lot of synthetic noise. Mix lead users, investigative reporters and dissenting practitioners from many regions.

    Do this: start from our voices by region and the live attention radar.

  3. Budget for real-world tipping friction

    Treat ~25% as a floor for small, aligned groups. For costly habits, plan for 50% or more, or change rules and defaults so the new norm is the easy path.

    Do this: try the tipping explorer with your own context.

  4. Work through trust brokers, not megaphones

    With 70% insularity, broadcasts bounce off. Adoption travels through employers, narrative intermediaries and local creator pods.

    Do this: find trusted voices on your topic in our seven frontiers.

  5. Test every signal before you spend on it

    Separate observation from interpretation, look at all four layers, test competing explanations, and match budgets to maturity (recognise, observe, trial, plan, implement).

    Do this: use the four-step pipeline in our method.

Evidence at a glance

Mechanisms, benchmarks and their limits

Every model of change has a blind spot. This table puts them side by side.

MechanismMain sourcesWhat drives itBenchmarkLimitation
Revised modernisation theoryInglehart & Welzel, WVS/EVS2Security and personal agency111 societies mapped (2023); 0.65 correlation with industrial shareReversible under shocks; national averages hide diversity
Multi-Level PerspectiveFrank Geels3Landscape pressure + regime weakness + mature nichesTracks regime phase-out, incumbent reorientation, speed vs depthIncumbents can absorb niches without real change
Social tipping pointsCentola et al. 2018; meta-analyses19Peer coordination once a minority hits critical mass24.3–25% in the lab; 10–43% across studiesTrivial lab rewards; costly habits need 50–65%6
Semantic change in languageHamilton, Leskovec, Jurafsky 201610Frequent words resist change; words with many senses change faster200 years, 4 languagesHard to separate language drift from cultural cause
Narrative powerNarrative Initiative; Pop Culture Collaborative; Norman Lear Center2122New stories circulated across culture and media76 groups studied; 90% contribute by connecting peopleCausal impact hard to isolate; needs long funding
Trust brokeringEdelman Trust Barometer 20264Trusted intermediaries across value divides70% insularity; 62% / 57% would trust via a trusted influencerWidening income trust gap (29 points in the US)
Layered analysis & Three HorizonsInayatullah; Sharpe; HiltunenReading signals across headline, systems, worldview, mythFour-step triage; five action zonesRisk of confirmation bias without falsification tests
Read with care

How to read these numbers

  • Dates matterEach figure carries the date of its source. Some organisational figures (for example WGSN's FY2022 accounts) are the latest public ones, not this year's.
  • Estimates are labelledFake-follower shares are an automated vendor estimate. Vendor claims (such as subscriber sales uplift) have not been independently audited.
  • User identities are not peopleDataReportal counts accounts; duplicates and fake accounts mean identities exceed unique humans. Ad reach is not monthly active users.
  • Correlation is not causeCredit for a policy (such as Australia's law) is shared by many actors. We name influential framers; we do not claim they acted alone.

Spotted an error or a better source? Tell us and we will correct it, with a note of the change.

Sources

Sources behind this brief

External sources only. Numbers in the text link here.

  1. Centola, Becker, Brackbill & Baronchelli (2018). Experimental evidence for tipping points in social convention. Science. ResearchGate
  2. World Values Survey. Inglehart–Welzel Cultural Map (2023). worldvaluessurvey.org
  3. Geels. The Multi-level Perspective on Sustainability Transitions. Cambridge University Press. cambridge.org
  4. Edelman Trust Institute. 2026 Edelman Trust Barometer. edelman.com
  5. DataReportal. Global social media statistics (Oct 2026). datareportal.com
  6. Summary of Özden's 2023 critique of the 25% threshold. rufuspollock.com
  7. Reuters Institute. Digital News Report 2026: How news creators are impacting politics and media. reutersinstitute.politics.ox.ac.uk
  8. Collabstr. TikTok fake follower checker (vendor estimate). collabstr.com
  9. The Pareto effect in tipping social networks: from minority to majority (preprint). copernicus.org
  10. Hamilton, Leskovec & Jurafsky (2016). Diachronic word embeddings reveal statistical laws of semantic change. stanford.edu
  11. DataReportal. Digital 2026: Global Overview Report. datareportal.com
  12. SearchLab. Social media statistics 2026. searchlab.nl
  13. The Guardian (7 Dec 2025). How Australia became the testing ground for a social media ban for young people. theguardian.com
  14. Tech Policy Press. Early lessons from Australia's teen social media ban. techpolicy.press
  15. eSafety Commissioner. Social media age restrictions. esafety.gov.au
  16. Education Next. New Australian law restricts social media for kids. Is it working? educationnext.org
  17. NBER Working Paper w33508. Measuring work from home. nber.org
  18. Remote work statistics 2026 (compilation). statisticsfundamentals.com
  19. Carnegie Endowment (Sept 2025). Confronting backlash against Europe's green transition. carnegieendowment.org
  20. ESRI. Economic insecurity, greenlash and the demand for climate policies in Europe. esri.ie
  21. Commons Library. Narrative change: start here. commonslibrary.org
  22. Pop Culture Collaborative / USC Norman Lear Center. PCfSC connections and accomplishments. popcollab.org
  23. Apax Partners. Apax Funds complete acquisition of WGSN. apax.com
  24. Future Today Strategy Group. ftsg.com
  25. Gates Foundation. 25th anniversary fact sheet. gatesfoundation.org
  26. Wikipedia. List of most-followed TikTok accounts (Nov 2026). wikipedia.org
  27. Resourcera. Netflix subscribers 2026. resourcera.com
  28. Wikipedia. ChatGPT. wikipedia.org
  29. Influencer Marketing Africa. Creators, brokers and brands. influencermarketingafrica.com
  30. Narrative Initiative. Funding narrative change: an assessment and framework. narrativeinitiative.org
  31. World Economic Forum. Institute for the Future (IFTF). weforum.org
  32. Forum for the Future. forumforthefuture.org
  33. TIME (18 Jul 2026). What the world should learn from Australia's social media law. time.com

Seeing a shift the research has not caught yet?

Early signals come from people close to change. Share what you notice, or a solution that is already working somewhere.