Instagram Engagement Rate & Fake Followers: Influencer Audit

For agencies, brands and creator teams comparing Instagram accounts before a campaign. About $0.04 per account audited: a shortlist of 20 accounts costs about $0.80, plus Apify's small start charge. No login. The default checks one account, based on the last 6 posts available from its public embed.

Get a readable engagement verdict, a grade against a follower-size benchmark, public audience-quality signals, available video views and a sponsored-post planning estimate. Download the account rows, read the free run summary, or open REPORT.html to review the shortlist and its underlying posts.

“Fake followers” means a limited audience-quality estimate from public engagement signals. This Actor does not list followers, identify fake accounts, estimate a fake-follower percentage or prove that anyone bought engagement. A low rate can reflect content, timing, audience interests, inactive followers or platform distribution. Treat flags as questions to investigate before paying a creator.

What you get

Output	Meaning
verdict	Plain-language finding first, explicitly based on the last 6 posts by default
username, profileUrl, checkedAt	Account identity, public source link and UTC collection time
followers, posts, verified	Public follower total, total posts and verification indicator
engagementRate	Mean of likes plus comments per measurable post, divided by followers, as a percentage
medianEngagementRate	Median post interactions divided by followers; used for the grade
medianLikes, medianComments	Typical public likes and comments in the measured sample
meanLikes, meanComments	Averages, useful for seeing the effect of unusually successful posts
sizeTier, engagementNormForSize, engagementGrade	Creator-size reference, reference percentage and A-E comparison grade
postsRequested, postsSampled, postsAnalyzed	Requested size, returned sample and usable engagement records
sampleDescription	Exact sample-size explanation, including excluded counts
medianReelViews, viewsPerFollower	Median available video views and that median divided by followers
videosSampled, videosWithViews	Coverage counts so missing views cannot look like zero views
postingFrequencyPerWeek, sampleWindowDays, lastPostDate	Observed recent cadence, observation span and newest returned post
audienceQualityEstimateScore	Transparent 0-100 screening score, not audience authenticity or a probability
audienceQualityRedFlags	Each measured signal, its explanation and score deduction
commentToLikeRatioPct, engagementSpikeRatio	Comment-to-like percentage and largest post relative to the median
sponsoredPostSharePct	Share with a public paid-partnership label or recognized caption disclosure
topHashtags	Hashtags ranked by the number of sampled posts containing them
sponsorPriceLowUsd, sponsorPriceMidUsd, sponsorPriceHighUsd	Gross planning estimate for one sponsored post, in US dollars
sponsorPriceNiche, sponsorPriceMethod, sponsorPriceSource	User-selected niche, reproducible calculation and reference
recentPosts	Source links, dates, captions, likes, comments, available views and disclosure evidence
limitations, status, error	Coverage caveats and reasons an account could not be audited

Following counts, category, bio contact details, generic or repeated comment text and follower-growth history are unavailable from the reliable embed source. Their corresponding fields are null, including following, category, businessEmail, website, followerFollowingRatio, followerGrowthSpike, genericCommentSharePct and repeatedCommentSharePct. They are never filled with guesses or zeros.

When to use it

Compare a fitness creator shortlist, review beauty campaign candidates, check the typical engagement behind a large follower count, or revisit a partner's public performance before another campaign. The price estimate gives a consistent starting point for budget discussions when you do not yet have a rate card.

Brand accounts can also be measured, but the grade's reference population is creators. A retailer, institution or product brand may have a different engagement pattern. Use direct competitors and comparable content formats for important decisions.

Start in three steps

1. Add Instagram handles to usernames, or profile links to urls. The prefill is a small working example.
2. Keep postsToAnalyze at 6. Select a niche only if you want to specify the assumptions behind the sponsored-post estimate. Set your maximum run charge.
3. Run the Actor, then open the verdict dataset, the activity or pricing view, and REPORT.html. Export JSON, CSV or Excel through Apify as needed.

Input reference

Input	Default	Example and behavior
usernames	Prefilled example in the console	["blogilates", "@gymshark"]; public handles or profile links
urls	Empty	["https://www.instagram.com/blogilates/"]; merged with usernames
postsToAnalyze	6	From 3 to 6; values outside this range are adjusted with a summary note
niche	general	general, fitness, beauty, fashion, food, travel, technology or finance
saveRawData	false	Save decoded source evidence in the run's RAW record for verification

For integrations, handles, query, queries and startUrls are accepted aliases. startUrls can contain strings or objects with a url property. Comma-separated or newline-separated text works through string aliases. Input case, leading @, URL query strings and duplicate accounts are normalized. A maximum of 1,000 input entries is processed in one run; split larger lists into batches.

{
  "usernames": ["blogilates", "thebodycoach"],
  "postsToAnalyze": 6,
  "niche": "fitness"
}

Post URLs are not account inputs. Invalid links, empty inputs and inaccessible embeds produce clear free notes. The Actor cannot determine from an unavailable embed alone whether an account is private, missing, has disabled embedding or is temporarily refusing access.

A real cloud example

On 14 October 2026, a cloud audit of blogilates returned 3,147,757 followers, six measurable posts, mean engagement of 0.4504%, and median video views of 107,651.5. Views were available for all six sampled videos. A two-account exploratory audit finished in 59.1 seconds; the console prefill uses one account to keep the first run smaller.

{
  "username": "blogilates",
  "followers": 3147757,
  "engagementRate": 0.4504,
  "engagementGrade": "B",
  "postsSampled": 6,
  "postsAnalyzed": 6,
  "videosSampled": 6,
  "videosWithViews": 6,
  "medianReelViews": 107651.5
}

Calculation and interpretation

The main engagement rate is 100 * sum(likes + comments) / measured_posts / followers. Hidden or missing engagement counts exclude a post from this calculation. At least three posts with visible likes and comments, and a positive follower count, are required for a paid audit. True zero counts remain zero.

Grades use the median post engagement to reduce the influence of a viral post. The reference is HypeAuditor's September 2026 worldwide creator benchmark, read on 14 October 2026. The follower bands and median percentages are 1K-10K: 2.11%; 10K-50K: 0.62%; 50K-100K: 0.37%; 100K-500K: 0.34%; 500K-1M: 0.33%; and 1M+: 0.32%. Bands use an inclusive lower boundary and exclusive upper boundary. Accounts below 1K have no grade.

The Actor's own grade thresholds are explicit: A is at least twice the reference, B is at least the reference, C is at least half, D is at least a quarter, and E is below a quarter. These letters are a comparison aid, not a validated fraud classification. The headline mean can be high while the typical-post grade is lower.

Audience screening starts at 100. Median engagement below 25% of the size reference deducts 35 points; below 50% deducts 20 instead. A very low comment-to-like ratio below 0.05%, or a high ratio above 20%, deducts 15 when the minimum interaction thresholds are met. A post with at least 1,000 interactions and eight times the median deducts 10. Exact reasons appear in each flag. Giveaways, keyword campaigns and organic viral posts can explain these signals. Unavailable checks do not count as passed checks, and a score of 100 is not verification of a real audience.

Cadence uses the time between the oldest and newest sampled posts, not a complete month of activity. Sponsored share counts explicit paid-partnership evidence and standalone #ad, #advert, #advertisement, #sponsored or #paidpartnership tags. Unlabeled commercial content can be missed. Video views may be available for only part of the sample; inspect coverage before comparing them.

Sponsored-post planning estimates

The estimate follows an engagement-based budgeting approach described in Shopify's influencer-pricing guide. It multiplies followers by median engagement as a fraction, then by an assumed dollar value per interaction. General assumptions are $0.05-$0.08; beauty and fashion use $0.10-$0.15. The midpoint is halfway between the low and high values. Other listed niches currently use the general assumption.

Niche is supplied by you, not inferred from a biography. The calculation uses typical organic engagement, which may differ from sponsored performance. These are gross planning figures, not actual influencer earnings, a quotation or expected profit. Production minimums, usage rights, exclusivity, location, agency fees and taxes can substantially change the negotiated amount. Compare the estimate with a creator's media kit and campaign evidence.

Price and spending limits

Useful accounts audited	Standard account charge
1	$0.04
20	$0.80
50	$2.00
100	$4.00

Apify's start charge is $0.00005 at the default memory. Bronze account audits cost $0.038, Silver $0.036, and Gold or higher $0.032. Failed account lookups, duplicates, invalid inputs, insufficient measurable posts and the summary carry no account audit charge. The start fee still applies. Platform usage is included in the Actor's pricing.

The Actor checks the remaining maximum charge before fetching another account. It stops cleanly when another useful audit would not fit. A cloud test with a $0.041 maximum delivered one useful account from a three-account list and stopped. A $0.001 maximum produced no account charge and did not fetch an account. There are no external paid APIs and no special cap for free-plan users.

API and AI agents

Use the same input through Apify's API. Keep the token in your own environment:

curl -X POST \
  "https://api.apify.com/v2/acts/vant47~instagram-influencer-audit/run-sync-get-dataset-items?maxTotalChargeUsd=0.10" \
  -H "Authorization: Bearer $APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"usernames":["blogilates"],"postsToAnalyze":6}'

import os
from apify_client import ApifyClient

client = ApifyClient(os.environ["APIFY_TOKEN"])
run = client.actor("vant47/instagram-influencer-audit").call(
    run_input={"usernames": ["blogilates"], "niche": "fitness"},
    max_total_charge_usd=0.10,
)
accounts = [r for r in client.dataset(run["defaultDatasetId"]).iterate_items()
            if r.get("status") == "audited"]

With an AI agent connected to Apify MCP, ask: “Audit this Instagram influencer shortlist using the last six public posts, compare engagement by account size, and show audience-quality caveats.” Provide the handles and a spending limit. The agent can inspect the input schema, run the Actor and read its dataset or summary. Filter recordType == "account" for account rows; the final recordType == "summary" row is free.

Apify schedules, webhooks, Make, Zapier, n8n and Google Sheets integrations can consume the same output. Scheduled runs are independent snapshots. The Actor does not create a private cross-customer history or infer follower growth from one snapshot.

FAQ

Why only six posts? The reliable signed-out profile embed returns up to six recent posts. The default verdict, sample description and report say they are based on the last 6 posts. Requesting twelve does not manufacture twelve records; it is capped and noted.

Does this detect fake followers? It estimates audience-quality concerns from public engagement. It cannot inspect follower identities or prove fraud. Review source posts, creator analytics and campaign results before drawing a conclusion.

Why is a public account unavailable? Public profiles can disable embedding or temporarily refuse requests. Some are private, renamed or missing. Such accounts are free notes, with no invented engagement figures.

Why are comments or views missing? Hidden engagement is excluded; unavailable video views stay null. Comment text is not collected, so repeated or generic comment detection is unavailable. A comment count is not a comment-text sample.

Can I use the grade for brands? The arithmetic works, but the benchmark describes creators and excludes brands. Treat the grade as context, not a universal performance target.

Are the price figures actual earnings? No. They are a disclosed gross budgeting model for one sponsored post. They do not measure contracts, past revenue, net income or future sales.

Do I need a login or cookies? No. Only public signed-out embed endpoints are used. The Actor does not accept account credentials or solve CAPTCHAs. Availability remains subject to Instagram's public interfaces and terms.

Where is the report? Open the output link for REPORT.html. The free SUMMARY record and summary dataset row explain completed accounts, skipped duplicates, input adjustments and any spending-limit stop.