Case Study 10.2: Instagram's Like Count Experiment — What Hiding Numbers Revealed
A Test With High Stakes on All Sides
In April 2019, Instagram began one of the largest natural experiments in the history of social media product design. In seven countries — Canada, Australia, Brazil, Ireland, Italy, Japan, and New Zealand — the platform began hiding the public like count on posts. Visitors to a post could still see who had liked it (a few names were listed), but the total count was invisible to everyone except the post's creator.
The stated motivation was mental health. Instagram CEO Adam Mosseri explained in public statements that the company wanted to "reduce the social pressure of posting" and create "a less pressurized environment where people feel comfortable expressing themselves." The specific concern: that visible like counts drove social comparison anxiety, made low-performing posts feel stigmatizing, and contributed to the documented negative wellbeing effects — particularly in teenage girls — that Instagram's own internal research had been accumulating for years.
The experiment was significant for two reasons. First, it was genuinely large-scale: the selected countries collectively represented tens of millions of Instagram users, making it one of the most empirically informative natural experiments on social reward visibility ever conducted. Second, it was politically charged: removing a fundamental feature of how social approval was displayed on the platform touched the interests of professional creators, advertisers, and the platform's business model simultaneously.
What happened over the following two years revealed a great deal about what the like count actually does — and about how deeply the business model depends on it.
The Design of the Experiment
The hidden like count rollout was methodologically imperfect as a research instrument, for reasons characteristic of platform experiments at scale: Instagram did not preregister hypotheses, did not define primary outcomes in advance, and did not establish a formal control group in the same markets. Users in the test countries knew they were receiving a changed product (the absence of the count was visible), which created awareness effects. Users in other countries knew they still had the count, which was also a condition, not a neutral baseline.
Nevertheless, the scale and duration of the experiment — running across multiple countries for months, with millions of users in each country — provided substantial data on behavioral responses that would have been impossible to gather through laboratory studies or surveys alone.
The key design features:
- Post creators could still see their own like count in their insights dashboard — the hiding was specifically of the public count visible to visitors
- The change applied to all types of posts (photos, videos, carousel posts)
- Stories and live videos were not affected in the initial rollout
- The experiment was explicitly framed as a wellbeing intervention, not a general product test
This framing matters: by telling users and the public that the change was about mental health, Instagram created a context in which users understood the nature of the design change. This reduces the ecological validity of the experiment as a test of how hidden counts affect users who don't know the counts are hidden — but it accurately represents how such a policy change would function if implemented broadly.
The Behavioral Findings
Research examining the experiment's effects came from multiple sources: independent academic researchers who studied behavior in the affected countries before and after the rollout, Instagram's own internal research (portions of which were later reported by journalists), and extensive qualitative data from user surveys and interviews.
The findings were genuinely mixed — which is itself informative.
What improved. Several studies found evidence of reduced explicit social comparison anxiety among users who were aware of the like count change and who had previously reported high levels of comparison-related stress. In qualitative interviews, some users described feeling "freer" to post without fixating on the count, and some reported reduced anxiety about low-engagement posts. For users who had been consciously aware of using like counts as a comparison metric, the absence of the number did seem to reduce that specific behavior.
A 2020 study by Tara Malone and colleagues at the University of Melbourne examined Australian Instagram users before and after the rollout and found modest positive effects on self-reported wellbeing measures for users who had been high-frequency social comparison engagers. The effects were small but statistically significant in the intended direction.
What didn't improve. The more sobering finding was that the fundamental dynamics of social approval-seeking and social comparison largely persisted. Users who had been using like counts as comparison metrics adapted by using other available metrics: follower counts, comment counts, story view counts, and the small number of named likers visible even without the total count. Qualitative research found that many users quickly developed workarounds — asking followers directly how posts were performing, using third-party analytics tools that could retrieve engagement data, or simply judging relative performance by the names they recognized in the partial likes list.
Instagram's own internal research, portions of which were reported by Bloomberg in 2020, apparently reached similar conclusions: hiding like counts reduced some explicit social comparison behaviors but did not substantially change the underlying dynamics that drove those behaviors. The desire for quantified social feedback was robust; users found ways to satisfy it even when the most obvious number was removed.
The Goldie Hawn effect. Some researchers noted what one team informally called the "awareness effect": users who were told the change was about mental health, and who understood the mental health rationale, showed larger positive effects than users who were simply aware that the count was hidden without being told why. This suggests that some of the positive effects of the experiment were mediated by the changed framing of the platform experience — using Instagram with the active awareness that you're not supposed to judge your worth by the count — rather than purely by the absence of the number itself.
The Creator Response: A Loud and Economically Consequential Backlash
The most immediate and politically consequential response to the hidden like count experiment came from the creator community — the population of users who earn income from their platform presence.
For professional content creators, like counts are not primarily a psychological metric. They are an economic metric. Brands and advertisers use engagement rate — typically expressed as a ratio of likes and comments to followers — as the primary measure of creator effectiveness when selecting partners for sponsored content campaigns. A creator with 100,000 followers and a 5 percent engagement rate (5,000 likes per post) is substantially more valuable to a brand than a creator with the same follower count and a 0.5 percent engagement rate. Without visible like counts, brands could not easily assess engagement rate. Without that assessment, the commercial value of creator partnerships became harder to demonstrate and harder to price.
Creators responded with organized objection. Many reached out directly to Instagram. Professional creator associations lobbied against the change. Prominent influencers posted about the harm the hidden counts were doing to their businesses. The argument was framed partly in business terms (I can't show my value to brands) and partly in platform participation terms (creators generate the content that makes Instagram valuable, and this change makes it harder for creators to build and monetize their platforms).
The creator backlash was not merely a nuisance for Instagram. The creator community is the platform's primary content engine. Professional creators — a small percentage of users by number — produce the highest-quality, most widely-shared content on the platform. They also attract the advertising partnerships that fund the platform's growth. Any policy that systematically disadvantaged professional creators threatened the content ecosystem that made Instagram valuable.
Instagram's product team heard this feedback, and it shaped the eventual resolution of the experiment.
The Ad Industry's Perspective
Beyond individual creators, the broader advertising and marketing industry had concerns about hidden like counts that compounded the creator community's objections.
Social media advertising on Instagram works through a combination of paid placement (brands paying Instagram directly for ad impressions) and influencer marketing (brands paying creators to feature products in organic content). Both channels use engagement metrics to measure effectiveness.
For paid placement, hidden like counts on organic posts reduced the quality of engagement data available to advertisers who used engagement benchmarking to evaluate campaign performance. If the average Instagram post's like count was no longer visible, it was harder to calibrate what "good" engagement looked like for a given audience or content type.
For influencer marketing, the impact was more direct: the most commonly used metric for influencer selection and pricing (like count relative to follower count) was obscured. Brands and agencies that had built influencer marketing workflows around publicly visible like counts needed to adapt, and many reported friction in their ability to identify and evaluate creator partners.
The advertising industry response was measured compared to the creator backlash — advertisers have workarounds, including direct access to creator analytics through formal partnership agreements. But the signal was clear: the hidden like count change created friction in the commercial relationships that fund the platform.
What the Business Model Actually Requires
The hidden like count experiment, and the way it ended, is best understood as an experiment not just in whether hiding counts improves wellbeing, but in what the social reward visibility is actually for.
The conclusion the experiment forces is that visible like counts are structural to Instagram's business model, not merely one of many equivalent design features. They serve multiple critical functions simultaneously:
Engagement driver. Visible like counts motivate users to post content (to receive likes), to check their posts (to monitor accumulating likes), and to engage with others' posts (to build social reciprocity that generates likes for their own posts). The visible count is the primary feedback signal that powers the engagement flywheel.
Content quality signal. The like count functions as a crowdsourced quality signal that helps users identify which content is worth their attention. Highly-liked posts appear more trustworthy and worth engaging with than low-liked posts. This quality signal is part of what makes the platform's content ecosystem navigable.
Creator economy infrastructure. As described above, the visible count is the primary metric on which creator commercial valuations are based. Without it, the creator economy's pricing and selection mechanisms are significantly impaired.
Advertiser data. Engagement benchmarks, influencer vetting, and campaign measurement all depend on the visibility of engagement metrics including like counts.
Remove the like count, and you disrupt all four functions simultaneously. The business case for restoration was not simply "creators complained" — it was that the like count is woven into the economic architecture of the platform at multiple levels. Hiding it was an experiment in disrupting a structural element, and the disruption costs were real and immediately visible.
The Compromise and What It Means
By late 2021, Instagram had settled on what was framed as a user choice solution: users could, in their settings, choose to hide like counts on their own posts and to hide like counts on posts from others. The default was restored to visible counts.
The user choice framing is worth examining. It sounds like a balanced outcome: users who found like counts distressing could hide them; users who wanted them visible could keep them. But the practical effect of making visible counts the default is well-established in the behavioral economics literature: users overwhelmingly stick with defaults. Making visible counts the default, with hidden counts as an opt-in change, ensures that the large majority of users — including the large majority of teenage users — continue to see like counts.
The users who would most benefit from hidden counts — those experiencing significant social comparison anxiety — are precisely the users least likely to go into their settings and proactively change the default. The choice architecture makes the protective option available while ensuring it is rarely chosen by the population that needs it most.
What the Experiment Ultimately Revealed
The Instagram like count experiment is, in retrospect, one of the most informative episodes in the recent history of social media design. Several things it revealed:
Platforms conduct large-scale behavioral experiments. The like count change was, at scale, a natural experiment on the psychological effects of social reward visibility conducted on tens of millions of people. It was not described as an experiment in real-time, but it was — complete with conditions, behavioral data collection, and outcome analysis.
The business model and the psychological harm are linked. The same feature (visible like counts) that drives engagement, powers the creator economy, and funds advertising is also the feature associated with social comparison anxiety and negative wellbeing outcomes. This is not a coincidence. The feature is effective at both because it exploits the same neurological mechanism for both purposes: the deep human response to quantified social approval.
Removal of the metric doesn't remove the motivation. The most important finding for understanding the psychology of the approval economy is that hiding the number doesn't hide the desire. Users adapted, found proxies, and continued engaging in social comparison behaviors. This suggests that the problem is not the specific number — it is the broader environment of quantified social feedback and the behavioral patterns it trains. Any solution that only addresses the number while leaving the environment intact will have limited effects.
The people who built this understood what they were building. The fact that Instagram had conducted substantial internal research on negative wellbeing effects before the like count experiment — and that this research was documented, discussed internally, and ultimately did not prevent the restoration of visible counts — is a statement about the institutional priority structure at Meta. The research informed the experiment. The experiment's business costs informed the restoration. The outcome was predictable from the incentive structure alone.
The hidden like count experiment didn't hide the approval economy. It revealed it.
Sources: Mosseri, A. (April 30, 2019). "Instagram wants to be a place for connection, not competition." Instagram Blog. Internal Meta research on Instagram's effects: Wells, G., Horwitz, J., & Seetharaman, D. (September 14, 2021). "Facebook Knows Instagram Is Toxic for Teen Girls, Company Documents Show." The Wall Street Journal. Behavioral research on the experiment: Tara Malone et al. (2020) reported in academic working papers; Bloomberg reported on internal Instagram research in 2020. On the creator economy and brand partnerships: SignalFire (2023). Creator Economy Market Map. The compromise user-choice resolution was reported by multiple outlets in November 2021.