In short
Counters are dated observations, not a continuous live reading of the network. Check the exact post, source and date before comparing values. A difference can reflect an older observation or a different definition; ratios must be interpreted using their own references.
Read provenance
On the video detail page, check the network, publication date and available collection information. Collection represents the observed state at a given time, not a live counter.
Check that you are comparing the exact same post and account. A repost, copy on another network or very similar variation can have its own counters. The source link and network are therefore essential to explaining a discrepancy that initially seems abnormal.
Compare the same measure
- Distinguish views and plays, then likes, comments and shares. Different sources can provide different units. A missing value remains unknown and should not be replaced by zero.
- Velocity and comparable ratios require multiple compatible observations. A corrected counter or changed source can make calculation unavailable.
For an engagement ratio, the numerator and views must come from the same compatible observation. Do not divide current likes by an old view count to reproduce the displayed value. Comparing periods requires more than a single recent screenshot of the original network.
Example: a counter changed after collection
Fictional example: Reacast shows 12,000 views at its latest available observation while the network now shows 15,000. If source and unit match, that may simply reflect growth since the observation. A new compatible observation is needed to measure change over time and calculate velocity.
If numbers drop or change abruptly, do not automatically infer audience loss: counter corrections or a changed source can make comparison unavailable. Note the values, units and dates so support can inspect the same scope.
Report a persistent discrepancy
Return after the next available collection. If the discrepancy seems inconsistent, keep the video URL, dates and compared measure for support. A difference in views alone proves neither a sales issue nor paid amplification.