Instagram Fake Follower Statistics 2026

An August 2026 analysis of 136 public Instagram creator and brand accounts found a median model-estimated fake-follower share of 18.7%, with 36.0% of accounts above 25%.

That is the short answer. The important technical qualification is that 18.7% is an account-level median in a defined, non-random sample—not an estimate of the percentage of all Instagram users who are fake. The reports were generated from publicly observable profile and engagement signals. They did not provide follower-level labels, so the results should be interpreted as audience-risk estimates rather than verified bot counts.

Definition used in this study: “Estimated fake-follower share” means the value labeled Estimated Fake Followers by the Followerus model. It may reflect patterns associated with bots, purchased followers, inactive accounts, or other low-quality audience behavior. It is not proof that a creator bought followers.

Key findings

The median estimated fake-follower share was 18.7%; the arithmetic mean was 19.0%.
The middle 50% of account estimates ran from 7.1% to 31.3%.
46 of 136 accounts (33.8%) were below 10%.
41 accounts (30.1%) were between 10% and 25%.
49 accounts (36.0%) were above 25%.
Macro accounts had a median estimate of 25.4%, and mega accounts had a median of 26.4%. Micro and mid-tier medians were lower at 18.0% and 13.1%.
Follower count had only a weak positive rank association with estimated fake share (Spearman’s ρ = 0.193).
Engagement rate had a moderate negative rank association with the estimate (ρ = −0.581). This is not independent validation because engagement rate is one of the model inputs.
Verified accounts were not immune: their median estimate was 17.0%, compared with 19.9% for non-verified accounts.

What percentage of Instagram followers are fake in 2026?

For the 136 accounts in this study, the best central estimate is 18.7%, because the median gives every account equal weight and is less sensitive than the mean to extreme profiles. The mean account-level estimate was similar at 19.0%. These figures answer a narrower and more defensible question than “what percentage of Instagram is fake?”: they describe the public accounts found through our seven discovery queries during the study window.

We did not calculate a follower-weighted average. Such a figure would let a small number of very large brand or publisher accounts dominate the result and would create false precision. The sample ranged from 1,200 to 404 million followers, while the median account had 219,000 followers. Treating each account as one observation is therefore the clearer benchmark for campaign screening.

Distribution of estimated fake-follower share, August 2026 sample

Estimated share	Accounts	Percentage of sample
Below 10%	46	33.8%
10% to below 20%	32	23.5%
20% to below 30%	16	11.8%
30% to below 40%	31	22.8%
40% or higher	11	8.1%

The quartiles matter as much as the headline median. One quarter of accounts were at or below 7.1%, while one quarter were at or above 31.3%. In practical terms, a buyer who audits only follower count can encounter materially different audience-risk profiles among accounts that look similar on the surface.

Estimated fake followers by account size

The size-tier results show a visible step upward among accounts above 500,000 followers. Macro and mega accounts had median estimates above 25%, compared with 13.1% for mid-tier accounts. The result should not be read as proof that larger accounts purchase followers more often. Large profiles are older on average, accumulate more abandoned followers, attract more unsolicited bots, and often have lower follower-based engagement rates. All of those mechanisms can affect a public-signal model.

Estimated fake-follower share by follower tier

Follower tier	N	Median	IQR	Above 25%	Median engagement rate
Nano: 1K–<10K	6	19.1%	7.5%–30.9%	50.0%	2.90%
Micro: 10K–<100K	40	18.0%	5.7%–24.9%	25.0%	5.07%
Mid-tier: 100K–<500K	41	13.1%	0.5%–26.5%	26.8%	1.97%
Macro: 500K–<1M	15	25.4%	5.1%–32.7%	53.3%	0.28%
Mega: 1M+	34	26.4%	14.1%–34.4%	50.0%	0.29%

The nano result is unstable because it contains only six accounts and should not be generalized. Across the full sample, the rank correlation between log follower count and estimated fake share was 0.193. That is a weak relationship: size provides some context, but it is not a substitute for an account-level audit.

Results by discovery query

Accounts were discovered through seven predefined keyword searches: fitness, beauty, travel, food, fashion, gaming, and tech. These rows describe the accounts returned by each search query, not mutually exclusive scientific categories. Twenty-two retained accounts appeared in more than one query, so the category counts below intentionally overlap and must not be added together.

Estimated fake-follower share by discovery query

Query	N	Median	IQR	Above 25%	Median followers
Fitness	20	30.1%	7.8%–34.3%	60.0%	265,500
Beauty	30	13.1%	2.0%–28.4%	26.7%	424,000
Travel	29	18.7%	8.5%–31.3%	34.5%	402,000
Food	29	18.7%	9.7%–31.3%	31.0%	29,100
Fashion	30	9.1%	0.1%–14.4%	10.0%	163,000
Gaming	9	26.5%	21.8%–29.6%	66.7%	316,000
Tech	13	18.7%	5.7%–31.3%	38.5%	47,800

Fitness and gaming returned the highest medians, while fashion returned the lowest. This is a descriptive result, not a ranking of industries. The gaming group has only nine accounts; account-size distributions differ by query; the discovery engine is not a random sampler; and duplicated accounts contribute to more than one row. A causal claim such as “gaming audiences contain more bots” would exceed what these data support.

Does Instagram verification imply a clean audience?

No. The 85 verified accounts had a median estimated fake share of 17.0%, versus 19.9% for the 51 accounts without a verification mark. The mean was 19.0% in both groups. Verification describes account identity or platform status; it does not certify every follower as authentic or active.

Estimated fake-follower share by verification status

Status	N	Mean	Median	IQR	Above 25%
Verified	85	19.0%	17.0%	7.3%–31.3%	32.9%
Not verified	51	19.0%	19.9%	6.4%–31.3%	41.2%

The two groups also differ substantially in size: the median verified account had 654,000 followers, while the median non-verified account had 54,100. A simple verified-versus-unverified comparison is therefore confounded and should not be treated as an estimate of the effect of verification.

Why engagement rate and the estimate move together

Estimated fake share and engagement rate had a Spearman rank correlation of −0.581: accounts with lower engagement generally received higher fake-follower estimates. This is useful operationally but should not be presented as external confirmation of the model. Engagement rate versus a size benchmark carries a nominal 25% weight in the Followerus scoring system, while the likes-to-comments ratio and engagement stability carry another 15% each. The association is partly built into the measurement procedure.

Low follower-based engagement can also have legitimate explanations: an old audience, a change in content topic, a large share of passive followers, viral reach beyond followers, seasonal posting, or a mismatch between Reels performance and follower-count formulas. A robust creator review should combine the estimate with recent post-level performance, audience geography, growth history, Story or Reel reach supplied by the creator, comment relevance, and campaign-specific conversion data.

Methodology

Sampling frame

We collected public profiles returned by Followerus keyword searches for fitness, beauty, travel, food, fashion, gaming, and tech. Accounts with fewer than 1,000 followers were removed, and usernames appearing in several searches were deduplicated for the overall analysis. The resulting candidate set contained 137 unique accounts.

For freshness, the published analysis retained reports with a “last checked” date from August 1 through August 10, 2026. One older cached report dated July 14 was excluded, leaving N = 136. All retained accounts were public at the time their report was available. The final sample included individual creators, companies, publishers, and other public accounts surfaced by creator-discovery searches.

Measurement model

The Followerus fake follower checker reports an Audience Quality Score, an Estimated Fake Followers percentage, an audience breakdown, follower count, engagement rate, verification status, and the report date. Its six documented signals and nominal weights are:

Engagement rate versus the account-size benchmark: 25%.
Likes-to-comments ratio: 15%.
Engagement stability across posts: 15%.
Followers-to-following ratio: 5%.
Follower sample quality: 30%.
Comment quality: 10%.

In every retained public report, follower sample quality and comment quality were marked unavailable. The report pages described their audience breakdowns as estimated from engagement patterns. We therefore extracted the published model output as-is and did not impute follower-level observations, infer missing signal values, or independently reweight the model. This limitation is central: the study benchmarks the model’s public-signal estimates, not a ground-truth census of bots.

Statistics

The unit of analysis was one unique account. We calculated the unweighted mean, median, 25th and 75th percentiles, categorical frequency counts, and Spearman rank correlations. Percentiles use linear interpolation. Follower tiers were defined as nano (1,000–9,999), micro (10,000–99,999), mid-tier (100,000–499,999), macro (500,000–999,999), and mega (1 million or more). Values in tables are rounded for readability; calculations used the unrounded extracted values.

Quality-control checks found no missing values for follower count, estimated fake share, Audience Quality Score, or engagement rate among the 136 retained records. The observed estimates ranged from 0.0% to 50.0%, and the rounded real, suspicious, and inactive audience components summed to 100% for every retained report.

Researcher disclosure

This is a first-party Followerus benchmark produced from Followerus search results and Followerus model outputs. It has not been independently audited or validated against Instagram’s internal account-level data. Followerus states in its terms that audience-quality scores and fake-follower percentages are statistical estimates derived from public signals, not factual claims about a specific account or person.

Limitations

Not a random Instagram sample. The accounts came from seven keyword result sets in one discovery tool. The findings should not be projected to all Instagram accounts.
Selection bias. Searchable, indexed, and highly visible profiles may differ from the broader creator population.
Mixed account types. The sample includes creators, brands, and publishers. It measures the public creator-discovery ecosystem rather than creators alone.
No follower-level ground truth. Follower samples and comment text were unavailable in the retained reports. The estimates rely on engagement and profile-level patterns.
Model dependence. The observed distribution reflects the Followerus scoring method. A different tool, threshold system, or data source could produce different percentages.
Cross-sectional design. Each account contributes one recent report. The study does not measure how audience quality changes over time.
Correlation is partly mechanical. Engagement rate is an input to the estimate, so its negative correlation with estimated fake share is not an out-of-sample accuracy test.
Niche rows overlap. Twenty-two retained accounts appeared in multiple discovery queries, and category sample sizes and follower distributions differ.

How brands and agencies should use these statistics

The 18.7% median is useful as a screening reference, not as an automatic rejection threshold. An account below the median may still be wrong for a campaign, while an account above it may have a legitimate explanation and strong conversion data. Use the number to decide where deeper review is warranted.

A practical evaluation sequence is to compare candidates within the same size and content category, inspect the latest posts for stable and relevant engagement, request first-party reach and audience-location screenshots, look for unexplained growth spikes, and then compare estimated reachable audience with the proposed fee. Where the commercial risk is material, repeat the audit and manually inspect a follower sample before contracting.

Do not convert “25% estimated fake followers” into “25% of the budget will be wasted.” Reach, impressions, content quality, algorithmic distribution, and conversions do not scale linearly with follower count. The estimate is one risk variable in a broader media-buying decision.

Frequently asked questions

What is the average percentage of fake Instagram followers in 2026?

In this sample, the mean account-level estimate was 19.0% and the median was 18.7%. The median is the preferred headline because the sample spans accounts from 1,200 to 404 million followers. These are sample statistics, not a platform-wide census.

How many Instagram accounts had more than 25% estimated fake followers?

Forty-nine of 136 accounts, or 36.0%, were above 25% in the retained sample.

Are verified Instagram accounts free of fake followers?

No. Verified accounts had a median estimated fake share of 17.0%, and 32.9% of verified accounts were above 25%. Verification should not replace an audience audit.

Do larger Instagram accounts have more fake followers?

Macro and mega accounts had higher medians in this sample, but the full-sample relationship between follower count and estimated fake share was weak (Spearman’s ρ = 0.193). Account age, niche, visibility, and engagement behavior may all influence the result.

Does a high estimate prove that an influencer bought followers?

No. Bots can follow accounts without permission, real users can become inactive, and public engagement patterns can have several explanations. The estimate should trigger additional review, not an accusation.

How should this research be cited?

Suggested citation: Followerus Research. “Instagram Fake Follower Statistics 2026: Analysis of 136 Public Accounts.” August 2026. When quoting the 18.7% figure, retain the qualifier that it is the median model-estimated share in a non-random sample of public creator and brand accounts.

Study period: reports last checked August 1–10, 2026. Analysis completed August 10, 2026. Followerus is not affiliated with Instagram or Meta.

instagram-fake-follower-statistics-2026-wordpress.html HTML instagram-fake-follower-statistics-20

Instagram Fake Follower Statistics 2026: 136 Accounts