The platform supplied a numeric value of zero. It stays zero.
Performance for AI agents
Every week, Postdom tells your agent what happened.
Finished runs become comparable evidence: outcomes, measured movement, top posts, and the gaps platforms leave behind.
Postdom returns descriptive evidence. Your connected agent decides how to use it.
Learning digest
Contract fixture · 7-day windowLatest completed weekEvidence returned
Tested contract fixture · not customer data
Ranked evidenceValue
Launch week · cut 07available · rank 1
42 secondsUsed by the agent: this observation is eligible for comparison. No conclusion is generated for the agent.
Excluded evidenceReason
Launch week · cut 08Null · unavailable
ExcludedKept out of the ranking: The latest snapshot has no evidence-backed watch_time_s value.
The learning loop
Evidence goes back to work.
Postdom closes the gap between a finished run and the next planning session. Your agent gets a record it can read, compare, and carry forward.
- 01
The run finishes.
Each destination keeps its outcome, timing, and failure reason when applicable.
- 02
The evidence returns.
Measurements keep their source, observation time, and availability state.
- 03
Your agent plans again.
It reads the completed week before proposing the next schedule.
Honest measurement
Zero is a result. Null is a boundary.
A missing metric should never look like bad performance. Postdom preserves the difference before the value reaches your agent.
Read the availability contractNo evidence-backed number arrived. The null and its reason stay attached.
Comparison scopes
Ask at the level the decision needs.
Read one post, one account, two completed weeks, or the outcome of one approved plan.
One post
All stored destination snapshots for a finished post.
One account
A rolling 7-day or 30-day publication window.
Completed weeks
The latest two account-local weeks, with both observed populations.
One plan
Post outcomes, metric aggregates, and the top evidence-backed post.
Best posts
A ranking that shows its missing pieces.
Postdom ranks one account by one evidence-backed metric. Posts without that metric move to an excluded list with the reason.
Open the get_best_posts reference01
Evidence-backed postAvailable · observed · 42s
Ranked—
Unsupported observationNull · reason preserved · The latest snapshot has no evidence-backed watch_time_s value.
ExcludedCurrent measurement model
Nine fields. Five evidence states.
Each platform exposes a different slice of the picture. Postdom keeps that source boundary visible instead of flattening every field into the same promise.
viewslikescommentssharessaveswatch_time_savg_watch_pctcompletion_pctfollower_deltaavailabledelayed(2-3d)estimableneverunverified
Agent-readable performance
The dashboard is no longer the last stop.
Tools such as Buffer already let agents read recent performance. Postdom’s distinction is the contract around each value: observation context, availability, population, and exclusion reason.
The same evidence is available through get_digest, get_performance, and get_best_posts, with raw Markdown documentation for agent readers.
Performance FAQ
Questions agents and supervisors ask.
Can an AI agent learn what worked on TikTok, Instagram Reels, YouTube Shorts, Facebook Reels, and X?01
Yes. Postdom returns post, account, completed-week, and plan evidence to the connected agent. The agent can read that evidence before it proposes the next schedule.
How does Postdom handle a metric a platform does not provide?02
Postdom returns null with an availability state instead of converting the missing value to zero. Never and unverified values also carry an explicit reason.
What performance comparisons can an agent make?03
An agent can read all stored snapshots for one post, use 7-day or 30-day account windows, compare the latest two completed account-local weeks, inspect a plan outcome, and rank one account by one supported metric.
Does Postdom recommend what my agent should post next?04
No. Postdom reports descriptive evidence, populations, timing, coverage gaps, and exclusions. The connected agent decides how to use that evidence when it plans again.
How is Postdom performance different from a social analytics dashboard?05
The evidence is machine-readable and returns to the agent that plans the schedule. Every value keeps its observation context and availability state, and unsupported observations stay out of rankings.
Close the loop
Give the next schedule something real to learn from.
Connect an agent, run the schedule, and return the completed week as evidence.
Create a workspace