How our social media audits work

Public data and sampling

Reports use network APIs and public-data providers. Depending on the network, the sample can include followers or subscribers, profile fields, post dates, formats, views, likes, comments, replies, and shares. Privacy settings and the fields returned by the provider limit what can be measured.

Recent-post samples do not represent a full account history. X reports use a popularity-ranked sample, so they do not infer posting frequency from it. LinkedIn reports focus on personal profiles and professional evidence; we do not publish LinkedIn engagement benchmark pages.

Private reach, impressions, watch time, audience demographics, sales, and conversions are not inferred. An unavailable interaction count is different from an observed zero. Incomplete reports explain their limitations and can be excluded from search indexing and public directories.

Engagement rate formulas

Public engagement combines the available likes or reactions, comments or replies, and shares returned for the eligible posts. The denominator changes what the percentage means.

By followers

Total public interactions ÷ eligible posts ÷ followers × 100. This is the average interaction count per post relative to the audience size at the time of the snapshot.

By video views

Total public interactions ÷ total views × 100, using the same eligible videos for both totals. YouTube and TikTok reports use this method when more than half the sampled posts expose positive view counts. Otherwise, they use the follower formula. Videos without usable views are excluded from the view-based calculation.

By reach

The manual calculator can divide your engagement total by reach from your own analytics. Public profile audits do not have this private denominator, so they cannot reproduce a native reach-based engagement rate.

Facebook calculations include only posts with available public response counts. Instagram samples with a strong pattern of placeholder likes are excluded from engagement calculations. Small percentages may round to 0.00%; that does not necessarily mean the sample had no interactions.

Typical results and audit scores

The median is the middle result after sorting the sampled values. For an even-sized sample, it is the average of the two middle values. We use it to show typical views or interactions alongside the strongest result and variation across the sample.

Audit scores apply platform-specific criteria and weights to the public evidence. The report shows the criteria, scores, and supporting observations. A grade is an assessment under this rubric, not a platform-issued rating or a prediction of future growth. Scoring changes can change a grade even when the underlying profile has not changed.

Rankings and benchmark samples

Rankings contain profiles with qualifying public reports and are ordered by their stored follower or subscriber count. They cover the accounts in our dataset, not every account on a network. Individual snapshots can have different collection dates.

Each profile contributes one engagement-rate observation to its audience tier. The benchmark is the median of those profile rates, not the engagement rate of a pooled set of all posts. The sample is selected by the accounts audited and the data available, so it is not a representative industry survey.

A platform benchmark page needs at least 10 qualifying profiles; each tier shows its own sample size, which can be smaller. Use small samples cautiously. Compare the same network and formula, especially where video reports can fall back from views to followers. The latest update date describes the newest qualifying report in the dataset, not a simultaneous refresh of every account.

Corrections and freshness

Read the generated date before using a report in a decision. Public counts can change after collection. A recheck can request newer data, subject to provider availability and usage limits; temporary provider failures may leave an older saved snapshot available.

If a report uses the wrong profile or contains inaccurate information, use its support link to request a correction or removal review. Include the report URL and the specific field that needs attention.