Author’s Note: Why Generative Engine Optimisation (GEO) Requires a New Understanding of Social Data
As generative AI models and AI-driven search engines increasingly shape how consumers discover information, understanding data provenance, or where these engines get their answers, has never been more critical.
Currently, generative engines train heavily on public data scraped from major social media platforms. However, this raises a fundamental question for brand strategy and Generative Engine Optimization (GEO): Are these engines learning from the general public, or from a hyperactive minority?
To answer this, I conducted the following evidence review of social media participation behaviors. By analyzing data from regulatory bodies and platform disclosures, this review demonstrates that public social media feeds are not representative of average consumer sentiment. Instead, public output is highly concentrated among a small fraction of power users.
For founders, marketers, and SEO professionals, the takeaway is clear: optimizing a brand for the AI era means shifting focus away from vanity social media metrics and toward producing the structured, authoritative, and original content that generative engines actively seek and trust.
Why most social media users consume more than they create, and why a small minority supplies most public content
A cross-platform research article distinguishing participation, original creation, posting, commenting, and high-volume creator concentration.
Prepared for publication | Evidence current through 22 July 2026
Institutional surveys
Executive summary
HEADLINE FINDING The evidence reveals two distinct gaps: a participation gap between people who interact and people who mostly consume, and a much sharper production gap between occasional posters and the minority that supplies most public output.
The latest broad population evidence comes from Ofcom’s 2026 Adults’ Media Use and Attitudes report. Among UK adult social-media users, 49% said they share, post or comment, while 25% mainly like other people’s posts and 24% mainly read without much interaction. The categories show a near-even active-versus-passive split, but they do not show that 49% create original work: sharing and commenting are inside the active category.
On TikTok, the threshold matters even more. Pew Research Center found that 52% of U.S. adult users had ever posted a video, 40% had publicly visible videos, and the median poster had uploaded only six public videos over the life of the account. In the linked-account sample, the most active 25% produced 98% of all observed public videos. Therefore, “ever posted,” “has public content” and “creates at scale” describe very different populations.
A Pew study of Twitter found almost the same concentration: the top quarter of U.S. adult users produced 97% of observed tweets in 2021. Only 14% of that high-volume group’s output was original standalone posts; 49% was retweets and 33% replies. This is a valuable historical benchmark, not a current estimate for X after its rebranding and operational changes.
Eurostat’s 2025 data show how easily participation can be mistaken for creation: 89.3% of EU residents aged 16-29 and 67.3% of people aged 16-74 used social networks. Eurostat’s definition includes creating a profile, posting messages or making other contributions; it does not isolate original work, posting frequency or professional creation.
Finally, huge output counts do not disclose how many unique people create. YouTube reports more than 20 million uploads per day; Reddit reports 4.4 billion posts and comments in 2025; and LinkedIn reports post growth of 41% over three years. None of those figures is a unique monthly creator rate. The defensible conclusion is not a single global percentage, but a layered one: broad participation can approach half or more of users, original and sustained creation is narrower, and public output is highly concentrated.
Methodology and definitions
This review uses evidence available through 22 July 2026. Priority was given to regulators and statistical offices, then independent institutional research with transparent sampling, then first-party platform disclosures for operational scale. Commercial creator-economy estimates were excluded when they did not publish a defensible denominator, population, time window or reproducible definition.
The primary population sources are not directly interchangeable. Ofcom surveyed 7,533 UK adults aged 16 and over between 29 September and 28 November 2025; the regulator warns that year-to-year trends are indicative because its methodology changed. Pew surveyed 2,745 U.S. adult TikTok users in August 2023 and observed 869 volunteered accounts; its linked-account sample has a margin of error of plus or minus 5.8 percentage points. Pew’s Twitter study surveyed 2,548 U.S. adult users in May 2021 and analysed 1,026 volunteered handles, with a plus or minus 6-point margin for the behavioural sample. Ofcom methodology; TikTok methodology; Twitter methodology.
A measurement ladder
Platform use: A person visits, watches, reads or maintains an account.
Participation: A person likes, comments, shares, reposts or contributes a message.
Original creation: A person makes and posts material that is substantially their own.
Public posting: The material is observable to researchers or the public, rather than private or deleted.
Active creation: A person publishes within a defined recurring period, such as weekly or monthly.
High-volume creation: A small group supplies a disproportionate share of total posts or videos.
Professional creation: A person earns income, operates a business or treats creation as an occupation.
EDITORIAL RULE Do not call someone a “creator” merely because a survey counted a like, comment or share. Use the narrowest label supported by the question wording and denominator.
What the best available evidence says, and what it does not say
| Evidence base | Population / period | Reported result | Correct interpretation |
| Ofcom 2026 | UK adult social-media users; fieldwork Sep-Nov 2025 | 49% share, post or comment; 25% mainly like; 24% mainly read. | Broad participation. Sharing and commenting are included, so 49% is not an original-creator rate. |
| Eurostat 2026 | EU residents; 2025; ages 16-29 vs 16-74 | 89.3% of younger people and 67.3% of the general population used social networks. | Participation includes a profile, messages or contributions; it does not establish regular or original creation. |
| Pew TikTok | U.S. adult TikTok users; Aug 2023 survey and linked accounts | 52% had ever posted; 40% had publicly visible videos; median poster had 6 lifetime public videos. | Ever posting, public posting and sustained posting are different thresholds. |
| Pew TikTok | 869 volunteered, valid accounts; behavior observed in 2023 | Top 25% by posting volume produced 98% of observed public videos. | An output-concentration result, not a claim that those users created 98% of all TikTok videos worldwide. |
| Ofcom: self-made video | UK social-media users; ages 25-34 vs 75+ | 42% vs 3% posted videos they had made themselves - a 14-to-1 ratio. | Original-video posting by age, across social media rather than a single platform. |
| Ofcom: livestreaming | UK adult social-media users; fieldwork Sep-Nov 2025 | 18% posted their own livestreams; 23% of video/livestream posters cited business promotion. | A specific original-broadcast behavior; the motivation percentage uses posters as its denominator. |
| Pew Twitter | U.S. adult Twitter users; public tweets, Jun-Sep 2021 | Top 25% produced 97% of tweets; only 14% of the top group’s output was original posts. | Historical pre-X benchmark. Tweets include retweets, replies and quote tweets; it is not a current X estimate. |
Note: Percentages use different countries, populations, time windows and definitions. The table is a comparison of evidence types, not a league table of platforms. Derived figures (48% never posted, 60% without public video, the 14-to-1 age ratio and percentage-point differences) are simple arithmetic from the cited published values.
1. The first gap: active participation versus passive use
Ofcom provides the clearest recent broad split. In its 2026 report, 49% of adult social-media users said they share, post or comment. A further 25% mainly like things others post and 24% mainly read with little interaction. The published categories total 98%, leaving a small residual for other or unreported responses; it would therefore be inaccurate to rewrite the finding as a precise “51% non-creator” rate. More importantly, the 49% active group mixes original posts with shares and comments. Ofcom report.
The active share was 61% in the prior wave and 59% in 2023. The latest 49% is 12 percentage points below 2024 - a 19.7% relative decline - but Ofcom explicitly cautions that the trend is indicative because the 2025 fieldwork used a new survey structure. Younger users aged 16-34 were more likely to share, post or comment (58%), as were users in AB socio-economic households (57%). Ofcom report.
The same study found meaningful attitudinal differences, but not causation. Forty-five per cent of active users said social media was good for their mental health, compared with 25% of passive users. Active users were also more likely to say social media made them feel creative (55% versus 27%). These associations may reflect self-selection - people who enjoy participation may be more likely to participate - rather than an effect caused by posting. Ofcom report.
2. Original creation narrows the field
When the behavior becomes more specific, the measured shares change. Ofcom found that 18% of adult social-media users posted their own livestreams. Among users who posted their own videos or livestreams, 53% cited fun, 49% staying in touch, 44% feeling creative and 23% promoting a business. The denominator for those motives is the group already posting video or livestream content, not all users. Ofcom report.
Age differences are particularly large for self-made video. Forty-two per cent of UK social-media users aged 25-34 posted videos they had made themselves, compared with 3% of users aged 75 or older - a 14-to-1 ratio. This result is cross-platform, so it should not be attributed specifically to TikTok, Instagram or YouTube. Ofcom report.
Ofcom also measured creative activity beyond posting. Among UK internet users, 17% edited photos, 16% made videos, 7% created websites, 7% created music and 5% created games. These figures capture making, not necessarily publishing. They are a reminder that some people create privately or off-platform and would look like “non-creators” in a study restricted to visible posts. Ofcom report.
3. TikTok: half have posted, fewer post publicly, and a quarter supplies almost everything
Pew’s TikTok study is unusually useful because it paired a representative survey with observable account data. Among 2,745 U.S. adult TikTok users surveyed in August 2023, 52% had ever posted a video, implying 48% had not. For private accounts, Pew treated a public account-level like count as evidence that a video had existed even if the video itself could not be observed. Pew TikTok report; methodology.
Only 40% of users had videos that were publicly visible at observation, so 60% did not. Even among posters, sustained volume was modest: the median poster had six lifetime public videos. The headline “52% create” is therefore true only under an ever-posted threshold; it does not mean half the platform is currently or consistently publishing. Pew TikTok report.
The production gap is much wider than the posting gap. In the 869-account linked sample, the most active quarter by posting volume produced 98% of publicly accessible videos. The remaining three-quarters produced only 2%. Pew restated the same underlying finding in a March 2026 overview, but the behavioral data remain from 2023. That time stamp matters when the result is used publicly.
4. Output concentration is not unique to TikTok
Pew observed a nearly identical pattern on Twitter before it became X. Between 12 June and 12 September 2021, the most active 25% of U.S. adult users produced 97% of tweets in the linked-account sample; the other 75% produced 3%. The top group produced 65 tweets per month on average, while the bottom group’s median was zero. Pew Twitter report.
High volume also did not mean predominantly original publishing. Original standalone tweets were 14% of the top group’s output, while retweets were 49% and replies 33%; the remainder included quote tweets. Thus, even a content-count measure can overstate original authorship. The result should be labeled as a 2021 Twitter benchmark, not a verified 2026 X statistic. Pew Twitter report.
Figure 1. Public content production was extremely concentrated in the two directly observed Pew studies.
Sources: Pew Research Center, TikTok (behavior observed in 2023) and Pew Research Center, Twitter (behavior observed in 2021). Shares refer to public output from the linked-account samples, not all global platform content.
5. High social-network participation is not the same as a creator economy
Eurostat reported that 89.3% of EU residents aged 16-29 used social networks in 2025, compared with 67.3% of the general population aged 16-74 - a 22-point age gap. Its definition includes activities such as creating a user profile, posting messages or making other contributions. It does not specify original authorship, a posting-frequency threshold or whether the activity was public. Eurostat.
The Eurostat figures are therefore best used as evidence of broad participation and an age divide, not as evidence that nine in ten younger Europeans are recurring content creators. This is the central denominator problem: a person can qualify by maintaining a profile or contributing a message, yet still never publish an original public video, article or image.
6. Platform scale can conceal the creator gap
Major platforms often disclose activity volume or growth without revealing the share of unique people responsible for it. Those figures describe the size of the content stream, not the prevalence of creators:
YouTube said in April 2025 that more than 20 million videos were uploaded on an average day - at least about 13,900 per minute if spread evenly. It also said users averaged more than 100 million comments per day in 2024. These are upload and comment events, not unique human creators; repeat uploaders, businesses and institutions are included.
Reddit reported 4.4 billion posts and comments in 2025, up 11.3% from 2024, including 2,210,675,258 in the second half of 2025. Posts and comments are combined, and the transparency report does not state what proportion of users created them.
LinkedIn said posts rose 41% over three years. Separate FY2025 highlights said comments rose more than 30% and video uploads more than 20% that year. A January 2026 disclosure said members adding “creator” to their profiles grew nearly 90% between 2021 and summer 2025. These are growth indicators - not the percentage of LinkedIn members who post, create original material or publish monthly. FY2025 highlights; 2026 disclosure.
Instagram reported 3 billion monthly active users in September 2025 and more than 4.5 billion Reels reshares per day across Meta platforms. Those values measure audience scale and redistribution, not how many unique users originate content.
Facebook reported that views and time spent watching original Reels approximately doubled in the second half of 2025 versus the same period in 2024. The disclosure defines what Meta considers original, but does not publish an active-creator share for Facebook users.
DATA GAP This review did not identify a current, independently verified and directly comparable unique-creator prevalence rate for Instagram, Facebook, YouTube, Reddit or LinkedIn. That absence is itself material: output scale should not be converted into a percentage of people creating.
7. What the gap means
Feeds are not representative samples of users
When a quarter of observed users supplies 97% to 98% of public output, the most visible voices cannot be assumed to represent the silent majority. A feed can feel socially universal while being produced by a narrow, highly active subgroup. This matters for journalism, public opinion research, brand listening and policy debates that treat online volume as a proxy for population sentiment.
Creator strategy needs a denominator, not just volume
Platforms and publishers should report at least four separate indicators: unique users who posted in the period; unique users who posted original material; output per active poster; and the share of output supplied by the top 1%, 10% or 25%. Uploads, comments and posting growth are useful operational metrics, but they cannot show whether participation is broadening or whether existing high-volume accounts are simply publishing more.
“Non-creator” is not one stable identity
Someone may be a current reader, an occasional commenter, a former poster, a private creator, a professional creating under an organisation’s account or a maker who never publishes. The TikTok evidence shows why the label must be tied to a threshold: 48% had no evidence of ever posting, 60% had no public video at observation, and a much larger share fell outside the high-volume quarter. Each figure answers a different question.
Originality needs explicit treatment
Shares, retweets, replies, remixes and comments all contribute to platform activity, but they are not equivalent to original authorship. The Twitter composition data and Meta’s current distinction between original and duplicative content show why research instruments should separate production from redistribution and conversation. Pew Twitter report; Meta’s Facebook originality guidance.
8. Limitations
No global harmonised estimate: The sources cover different countries, ages, platforms and periods. Averaging them would create false precision.
Question wording changes the answer: “Ever posted” yields a larger creator share than “posted publicly this month,” while “post/share/comment” is broader still.
Public data miss private creation: Pew could count only publicly visible videos for posting volume; private, deleted and ephemeral material is partly or wholly invisible.
Handle-sharing can introduce selection: Pew weighted the volunteered account samples, but users willing to share a handle may still differ in ways weighting cannot fully remove.
Platform disclosures are first-party: YouTube, Reddit, LinkedIn and Meta metrics are useful for operational scale but are not independent prevalence studies and usually omit unique-creator denominators.
The Twitter result is historical: The 2021 study predates the X rebrand and major changes in ownership, product rules and user behavior.
Derived comparisons are descriptive: Percentage-point gaps, complements and simple ratios are arithmetic restatements of published values, not new statistical tests.
Active versus passive associations are not causal: Ofcom’s wellbeing and creativity differences may reflect who chooses to participate rather than effects produced by participation.
Conclusion
The creator gap is real, but it cannot be reduced to one universal percentage. Under Ofcom’s broad participation measure, active and passive social-media use is close to evenly split. Under Pew’s TikTok measure, about half of U.S. adult users had ever posted, but only 40% had public videos and the median poster had uploaded six. At the production level, a top quarter supplied 98% of TikTok videos in the observed sample; on Twitter in 2021, the corresponding share was 97%.
The practical lesson is simple: consumption is widespread, participation is broader than original creation, and sustained public production is concentrated. Any credible claim about “how many people create content” must state the country, population, time period, behavior, originality rule, privacy scope and posting-frequency threshold. Without those qualifiers, a creator percentage is more likely to blur the gap than explain it.
References
Ofcom. Adults’ Media Use and Attitudes Report 2026. 2 April 2026. Fieldwork conducted 29 September-28 November 2025
Eurostat. 89% of 16-29-year-olds used socials in 2025. 9 April 2026. Source dataset: isoc_ci_ac_i
Pew Research Center. How U.S. Adults Use TikTok. 22 February 2024. Survey and observed behavior from August 2023
Pew Research Center. TikTok users: Methodology. 22 February 2024.
Pew Research Center. The Behaviors and Attitudes of U.S. Adults on Twitter. 15 November 2021. Observed tweets from 12 June-12 September 2021
Pew Research Center. Twitter users: Methodology. 15 November 2021.
YouTube. 20 ways we’re celebrating two decades of YouTube. 23 April 2025.
Reddit. Transparency Report: July to December 2025. 2026.
LinkedIn. Networks, Not AI or Search, Are the #1 Trusted Source Amid Information Overload. 26 August 2025.
LinkedIn. LinkedIn Business Highlights from Microsoft’s Q4 FY25 Earnings. 30 July 2025.
LinkedIn. A New World of Work: Global Labor Market Rotates, Not Retreats. 14 January 2026.
Meta. In India, Instagram debuts a Reels-first experience for its mobile app. 28 September 2025.
Meta. Rewarding Original Creators on Facebook. March 2026.
Publication note: Hyperlinks in the article and references point directly to the institutional or first-party source used for each claim. All sources were checked against their live pages or reports on 22 July 2026.