How do I read Instagram engagement correctly when the report has empty values?
Read Instagram engagement from the field path of the detail action that produced the saved response, and leave unavailable values blank instead of writing zero. Check the saved response’s action first. For instagram.v2.fetch_post_info, read the counters in data.data.metrics. For instagram.v3.get_post_info, find the item with the matching shortcode in data.items[] and read the matching item’s counts. Calculate a change only when both observations provide the same metric.
For example, if a Reel’s likes rise from 120 to 135, report a net change of 15 with both observation times. Leave an unavailable play count blank.
Where are likes, comments and plays?
Check the shortcode and the account first, so that both observations describe the same post. The snapshot guide already parses instagram.v2.fetch_post_info and instagram.v3.get_post_info responses separately.
How should the report handle empty values and zero?
Keep a reported zero as zero, and mark an unavailable field as unavailable. A photo without plays can still be compared by likes and comments. Do not subtract an invented zero baseline from the later play count. A play count is not unique reach.
How do I explain a change between two observations?
List the post, both collection times, the before and after values and the net change. Calculation examples: likes changing from 120 to 135 yield +15; 120 to 118 yields −2. Keep negative changes in the report for review.
Calculate percentage growth only when the baseline is available and greater than zero: (after−before)/before×100. With a zero or unavailable baseline, report the difference and the baseline state without a percentage. An engagement rate needs its own explicit denominator and observation time.
What should accompany the report?
Keep both original responses, the CSVs and the comparison JSON with the actual observation times. The snapshot guide’s comparator accepts two valid snapshots of the same post from the same detail action, and checks the source files and counts. When an edited CSV no longer matches, investigate the difference against the original responses.
A report can say: “Between these observations, reported likes increased by 15. Comment counts were unavailable, so no comment change was calculated.” Replace the figures with your actual records. The change table establishes what changed; explaining campaign effectiveness needs additional content and campaign evidence.
Parsing and comparing saved responses runs locally. New retrieval is charged at the current pricing.

