Can a Low TikTok Engagement Rate Prove Fake Followers?

Collection setup

An Actor is a hosted data collection program. A Task is a saved input preset.

Input Configuration (Build 0.1.36)

• mode: knownUrls
• startUrls: Replace the entire saved sample list with the reader's single intended known public video URL.
• maxItems: 1
• includeProfileMetrics: true (adds best-effort current public profile facts)

Native Fields

The scraper returns native fields including inputUrl, postUrl, postId, profileId, profileHandle, profileUrl, publishedAt, videoDurationSeconds, views, likes, comments, shares, saves, reposts, profileFollowersCount, postMetricStatus, profileMetricStatus, status, statusReason, and scrapedAt. Missing metrics are returned as null, not zero.

Output and Control

The scraper emits a run summary called OUTPUT containing providedUrls, contentUrls, inputUrls, resolvedUrls, metricRows, noPublicMetricRows, blockedRows, failedRows, and statusCounts. Reconcile the intended known video by comparing saved native rows with this run summary. The postMetricStatus field provides per-metric availability and precision flags for each record. A current one-video control matched public post and profile IDs, with several counts marked as likely_rounded and saves or reposts marked exact. This confirms public counter and identity support on the source platform. It does not prove private analytics, video retention, fake followers, or derived rates.

How to Review Creator Engagement

1. State the decision without an accusation, such as whether a creator's public results justify more evidence before a partnership. Separate a low observed ratio from its possible causes. Do not invent universal pass or fail engagement thresholds.
2. Collect a declared known-video sample for the intended creator using knownUrls. Set maxItems to one for the pilot and includeProfileMetrics to true. The Actor, which is a hosted data collection program, adds current best-effort follower counts per video. A current follower count is not necessarily the count at publication or within the video's audience. Record video age and observation time.
3. Inspect source identities, status, precision, and optional metrics. Missing likes, comments, shares, or followers cannot become zero. Rounded denominator or numerator counts make derived ratios approximate. Keep different account identities separate, and do not join accounts by display name.
4. Define any ratio exactly. A simple public interaction ratio uses the sum of likes, comments, and shares divided by the current profile followers count, multiplied by 100. This uses count events, not unique engaged followers. A view-based ratio uses another denominator and answers another question. Reposts and saves should be included only under an explicit comparable formula. Do not change the formula across creators.
5. Build rival explanations. The selected videos may have little exposure. Posts may be old, new, or not typical. Followers may be inactive. The source may hide counters. The audience may be nonfollowers. Counts may be delayed or rounded. These are hypotheses, not diagnoses. One public ratio cannot assign shares of followers to these causes.
6. Check whether the apparent result persists in a fair declared sample and comparable observation window. Use the median and range of per-video values if suitable. Do not sum repeated follower denominators as if they were distinct accounts. Public follower changes do not establish purchased followers.
7. Request appropriate owned aggregate evidence or manually inspect public context when the decision needs more support. This guide does not collect follower lists, identify bots, or validate private access. Keep source claims, missing evidence, and unresolved risks separate. Do not contact creators through this workflow.
8. Write a conclusion such as the checked sample has low observed actions under this formula, the cause remains unresolved, and request more evidence. Avoid numerical fake-follower estimates and labels such as proven fraud. A due diligence note is an evidence record, not certification.

Worked Example Using Fictional Creators

FictionalCreatorA has 100,000 current followers. One checked video has 500 views, 250 likes, 30 comments, and 20 shares. The formula numerator is 300. The follower-based interaction ratio is 300 divided by 100,000 times 100, which equals 0.3 percent. The view-based event ratio is 300 divided by 500 times 100, which equals 60 percent. Neither is the proportion of unique people engaged. Counters can overlap and repeat. A low follower ratio is compatible with low exposure and does not identify fake followers.

FictionalCreatorB also has 100,000 followers and the same 300 interaction events, but 10,000 views. Its follower ratio is also 0.3 percent. Its view ratio is 3 percent. Both have the same follower ratio, but very different public exposure. Thus, 0.3 percent does not identify one follower-quality cause.

FictionalCreatorC has a follower count of null. The follower ratio is unavailable, not zero. If a rounded value like 100K is used, the exact arithmetic is an approximate ratio. Do not infer that A's high view ratio proves authentic followers or that B's low ratio proves bots.

Fictional Creator Comparison Table

A fictional three-video sample for FictionalCreatorA shows distinct per-video ratios:

This table describes the checked sample, not a whole-account fraud score.