What social APIs will and won't tell you about your posts
Impressions are not available everywhere. Any dashboard showing you a confident number for all of them is filling in gaps.
Optional — the post reads fine without one.
When we built analytics into AttentionPilot, the surprise was not how hard the APIs were — it was how uneven the data is. The networks do not report the same things, and several do not report the headline number people most want.
What each network actually gives you
- ✓X — impressions, likes, replies, reposts, all available
- ✓Bluesky — likes, replies, reposts. No impression count exists at all
- ✓LinkedIn — likes and comments on a personal post; impressions need an organisation Page
- ✓Instagram — reach, likes, comments, shares, saves
- ✓TikTok — views, likes, comments, shares
- ✓YouTube — views, likes, comments
- ✓Facebook — likes, comments, shares
- ✓Pinterest — impressions, saves and clicks
The honest way to show that
A dashboard has two options when a network reports nothing for a metric: show a zero, or show that there is no data. A zero is a lie — it says 'nobody saw this' when the truth is 'this network does not tell anyone'.
AttentionPilot stores a null and the interface says so. It looks less impressive and it is the only defensible choice.
If a tool shows you a confident impressions figure for a Bluesky post, ask where it came from. Bluesky does not publish one.
Why readings should be a time series
Engagement accrues over days, so a single number tells you where a post ended up but nothing about how it got there. Storing every reading rather than overwriting the last one lets you see a post that grew slowly versus one that spiked and stopped — which are different signals about what to write next.