Back

Methodology

Last updated: November 7, 2026.

1. Data sources

Influencer Camp collects statistical data on creators' social accounts through two distinct channels, which differ by access method and legal basis.

  • Authorised access via OAuth: when a creator connects their account through the third-party platform's OAuth flow (Instagram, TikTok, YouTube, Facebook, Twitch, Kick), we gain programmatic access to the metrics the user explicitly authorised at connection time: following, followers, published content, views, and the further metrics covered by the authorisation scopes granted. For TikTok, the connection goes through the TikTok API for Business; existing TikTok connections made through Login Kit keep being refreshed until the creator renews them, but they provide neither reach nor demographics. Facebook Pages and Kick channels are measured only through this channel. In this channel, processing is based on the user's direct consent.
  • Public data without an OAuth connection: for monitored profiles that have not connected their account via OAuth, public data for YouTube channels comes from the official YouTube Data API and public data for Twitch channels from the official Twitch API. For Instagram and TikTok profiles, we rely on third-party providers specialised in collecting publicly accessible information from profile pages. Monitoring of a specific profile is activated by the customer (agency or brand); for that processing, Influencer Camp and the customer act as joint controllers under Art. 26 GDPR, on the basis of the legitimate interest of the customer activating the monitoring. We process this information in compliance with applicable data protection law and guarantee data subjects the exercise of their rights under the GDPR. Controllership and legal bases for this processing are described in the Privacy Policy (§4-§5).

2. Observed, derived and estimated data

We classify metrics into three kinds: Observed, Derived and Estimated. Where the Service shows a methodology indicator, it states the kind, the source and, for Derived values, the formula.

  • Observed: a value returned by the platform or by the collection of publicly accessible information as is, at the time of collection (e.g. follower count, post count, total channel views, platform-reported reach). It reflects what the source platform reported.
  • Derived: a value we compute from observed data with a stated formula (e.g. engagement rate, the time-weighted average viewers of a Twitch live stream, total followers across platforms). Given its inputs, the result is exact.
  • Estimated: a value produced by a model or an inference, shown with a confidence level and never as an exact figure (e.g. the prevalent language and probable country of the Twitch and Kick audience, described below).

Metrics that go through our methodology registry and have no documented method are not shown.

Reach

Where platform-reported reach is not available, the reach value shown in the Service is a proxy:

  • Instagram, TikTok and Facebook: the account's follower count (Observed).
  • YouTube: subscriber and channel-view counts from the official YouTube Data API (Observed).
  • Twitch: the time-weighted average of concurrent viewers sampled while live (Derived).
  • Kick: concurrent viewers sampled at collection time (Observed).

These values describe a potential audience for planning: they are not measured reach or a number of unique people. Platform-reported reach (Instagram, Facebook, and TikTok through the TikTok API for Business) appears only with insight access.

Engagement rate

The engagement rate is a Derived value. We compute it over the content published in the settled window, that is the 30 days ending 7 days before the calculation, with one formula per platform. Insight access means access to the metrics the creator authorised through OAuth for your team (section 1).

  • Instagram and Facebook: with insight access, the sum of interactions divided by the sum of the content reach; content without reach, or without interaction data, is left out of both sums. If no content in the window has reach, or the summed reach is zero, the average of likes plus comments per content item, divided by followers. Without insight access it is not available, because per-content data comes only through the creator's authorisation.
  • TikTok: with insight access, the sum of likes, comments, shares and saves divided by the sum of the reach reported by the TikTok API for Business, over the posts (videos and photo posts); posts without reach, or without interaction data, are left out of both sums. If no post in the window has reach, or the summed reach is zero, the average of likes plus comments per post, divided by followers. Without insight access it is not available.
  • YouTube: with insight access, the sum of likes, comments and shares divided by the sum of engaged views; videos without engaged views, or without interaction data, are left out of both sums. Without insight access, or if no video in the window has engaged views or their sum is zero, the average of likes plus comments per video, divided by subscribers. This second formula uses the channel's public data and is available without insight access.
  • Twitch and Kick: we do not compute an engagement rate.

The engagement of a single content item is the total reported by the platform (Observed), when the platform provides one; otherwise we compute it as likes + comments + shares + saves (Derived).

Audience data

Audience demographics are available only for accounts connected through OAuth and come from the platform (Observed), except the Twitch and Kick estimates:

  • Instagram: gender, age, countries and cities of the followers; not available under 100 followers.
  • YouTube: gender and age of the channel's signed-in viewers and the countries its views come from, as reported by YouTube Analytics over the last 90 days. Countries are a share of views, not of viewers. YouTube hides low-volume groups, so small channels may have no data.
  • TikTok: follower demographics through the TikTok API for Business, in TikTok's own age bands, and follower activity by weekday and hour (average of the last 4 weeks). Activity requires Advanced Access or a verified Business account. Neither is available under 100 followers.
  • Facebook: countries and cities of the Page's followers; not available under about 100 followers. Age and gender are not collected.
  • Twitch and Kick: the platforms provide no demographics. We show an estimate (Estimated) of the audience's prevalent language, inferred from the declared language of the live streams over the last 90 days: "predominantly" that language when it covers at least 80% of the live hours with a declared language, otherwise "mixed"; at least 10 live hours with a declared language are needed. A probable country is given only when the language is spoken almost only in one country, with high confidence when the language covers at least 95% of those hours and medium from 80%; confidence stays medium when live streams with a declared language are less than half of the live hours. These estimates are based on the language of the live streams, not on who watches them; they are labels or bands with a confidence level, never percentages, and are not platform data.

For Instagram and Facebook, the demographics of an account with fewer than 1,000 followers are flagged as low reliability.

3. Collection frequency

Metrics are collected periodically. The frequency depends on the platform and the access method:

  • profiles connected through the platform's OAuth flow: every hour (for YouTube, see the next item);
  • Instagram and TikTok profiles monitored without an OAuth connection: typically every eight hours;
  • Twitch: channel data every hour and, during live streams, the viewer count about every 25 seconds;
  • Kick, only for channels connected through OAuth: channel data (followers and active subscribers) every hour and, during live streams, the viewer count about every 25 seconds; Kick notifies us when each live stream starts and ends;
  • Facebook, only for Pages connected through OAuth: Page data every hour; each collection refreshes the posts and reels published in the last 90 days, while older content keeps its last collected value;
  • YouTube, for both channels connected through OAuth and monitored channels: channel data (subscribers, total views, number of videos) every hour. Data for individual videos is refreshed at a frequency that depends on the video's age: every hour up to day 10 after publication, every 6 hours from day 11 to day 20, every 12 hours from day 21 to day 40, and once a day, at 6:00 Italian time, from day 41 on. For live streams and Premieres, age is counted from the start of the broadcast; ongoing and scheduled live streams are refreshed every hour, whatever their age.

For YouTube channels, new videos are detected every hour among the most recent uploads and, once a week, the channel's full video list is checked again to add any missing videos and recognise removed ones. The daily refresh of videos from day 41 on also recomputes the channel's number of long videos and Shorts and their average views.

For YouTube channels connected through OAuth, shares and engaged views of individual videos come from YouTube Analytics, the statistics service YouTube provides to the channel owner. YouTube Analytics finalises data with a delay of 48 to 72 hours, so these values may not include the last two or three days. They are refreshed every hour up to day 40 (for live streams, up to 90 days after the broadcast ends); after that, the last collected value is kept. The other video values (views, likes, comments) come from YouTube's public counters, at the frequency stated above.

For TikTok accounts connected through the TikTok API for Business, the reach of each video is reported by TikTok. TikTok may stop reporting it for a video inactive for more than 7 days: in that case the last collected value is kept. From the date of the TikTok API for Business connection, shares include in-app sends, so the series restarts at that date; views reported by TikTok for Business include paid views. Audience demographics are read once a day for Instagram, YouTube and Facebook and once a week for TikTok.

The exact collection time may vary depending on source platform availability, the daily usage limits of its APIs and operational load. If the daily limit is reached, video refreshes are not skipped: they are postponed and caught up once the daily limit resets. A value shown always refers to the latest available collection: for videos from day 41 on it is normally no more than about 24 hours old, and it may be older when a refresh is postponed. The time series shown in the Service are based on snapshots drawn from the last available collection: hourly for accounts; for individual content items, hourly for the first 35 days after publication and daily afterwards.

4. Identification and consistency

Each monitored account is uniquely identified by the numeric identifier assigned by the source platform, not by its username (which may change). Time series are built from snapshots of the last available collection, hourly for accounts and, for individual content items, hourly for the first 35 days after publication and daily afterwards; obviously inconsistent data points are discarded before insertion.

5. Data quality controls

The Service applies baseline checks to identify and discard anomalous or implausible data and preserve the integrity of the time series:

  • Empty responses: if a platform returns a profile with no data, the data point is discarded and not inserted into the time series.
  • All-zero values: a payload with every counter at zero is considered implausible and discarded to avoid overwriting a valid series.
  • Unreachable profiles: profiles that are consistently unavailable are flagged as errored and removed from the collection cycle.

These checks aim to preserve the integrity of the time series, not to provide exhaustive fraud detection.

6. Industry standards alignment

Our measurement methodology is informed by standards and guidelines published by industry bodies, including the Media Rating Council (MRC), particularly regarding data-source transparency, the distinction between observed, derived and estimated data, and data quality controls.

This alignment reflects our commitment to measurement transparency and quality. It does not imply any form of accreditation, certification, or formal compliance by the MRC or any other body.

We review this methodology at least once a year and whenever a source, formula or classification changes. Changes are listed in the log in section 10.

7. Limitations

The metrics shown by the Service are based on data provided by third parties and are observed, derived or estimated as described in section 2. Their accuracy depends on the quality and availability of the data returned by the source platforms, which we do not control. In particular:

  • platforms may change their APIs or the data they expose without notice;
  • public metrics may be rounded or delayed by the source platform;
  • the platforms' daily API usage limits may delay a refresh until the limit resets;
  • data collected from publicly accessible sources reflects the page content at the time of collection and is not guaranteed by the platform.

For further details on warranty disclaimers, please refer to our Terms of Service.

8. What we do not measure

For transparency, we explicitly state what the Service does not, or cannot, measure:

  • Private or inaccessible accounts: profiles that the source platform does not expose publicly and that have not been connected via OAuth cannot be measured.
  • Unsupported platforms: coverage is limited to the platforms the Service supports; unsupported platforms are not measured.
  • Metrics that require an authorised connection: metrics available only through authorised access via OAuth (such as platform-reported reach and other insight metrics) are not available for profiles monitored solely through publicly accessible information.
  • Audiences not deduplicated across platforms: totals computed across multiple platforms are a gross, non-deduplicated audience. A single person who follows a creator on several platforms is counted on each, so such totals do not represent the number of unique individuals reached.
  • Invalid-traffic filtering: filtering of invalid traffic (bots, fake engagement) depends on the upstream source (the platform's API or third-party collection providers) and is not independently verified by Influencer Camp.
  • YouTube Premieres not counted as live streams: while it is scheduled, a Premiere is not stored or counted. From the moment it airs it counts as a long video, not as a live stream or a Short, whatever its duration. A Premiere not detected while it was scheduled is recognised when the broadcast ends and counts as a live stream until then; a Premiere first made public after it aired stays a live stream.
  • Twitch and Kick demographics: the platforms do not provide them; for these accounts we show only the estimates described in section 2.

9. Contact

Questions about our methodology can be sent to privacy@influencer.camp.

10. Change log

  • November 7, 2026: three data kinds (observed, derived, estimated), reach, engagement rate formulas, audience data, TikTok API for Business, Facebook and Kick collection frequency, Twitch viewer sampling about every 25 seconds, YouTube Premiere rule, annual review commitment.
  • October 21, 2026: YouTube video refresh frequency by video age.
  • September 14, 2026: added the "What we do not measure" section.