Platform Stats
The performance numbers behind compensation and campaigns — and exactly which of them arrive automatically.
Platform stats are the performance data behind compensation, campaigns, missions, and reporting. Each stat row is a snapshot for one host on one platform, stamped with a date.
Read this before you buy: what is automatic and what is not #
The honest answer is that most of it is imported. Only four numbers arrive on their own, none of them are the ones your compensation runs on, and only for TikTok.
| Number | TikTok | Bigo | How it gets in |
|---|---|---|---|
| Diamonds earned | ✗ | ✗ | CSV import, or by hand |
| Live minutes | ✗ | ✗ | CSV import, or by hand |
| Valid live days | ✗ | ✗ | CSV import, or by hand |
| Live sessions | ✗ | ✗ | CSV import, or by hand |
| Battles | ✗ | ✗ | CSV import, or by hand |
| Live views | ✗ | ✗ | CSV import, or by hand |
| Peak viewers | ✗ | ✗ | CSV import, or by hand |
| New followers | ✗ | ✗ | CSV import, or by hand |
| Total followers | ✓ automatic | ✗ | Automatic, or CSV |
| Accounts followed | ✓ automatic | ✗ | Automatic, or CSV |
| Total likes | ✓ automatic | ✗ | Automatic, or CSV |
| Video count | ✓ automatic | ✗ | Automatic, or CSV |
So: diamonds arrive by CSV or are typed in by hand, on both platforms. TikTok's API grants us profile and audience counts; it does not expose LIVE earnings, air time, or battle data to us, and Bigo has no API integration at all (see Bigo Live).
Two further things to plan around:
- The automatic pull runs every night, and when someone presses Sync now on a host's platform record. See TikTok Connections.
- You can type a stat row in by hand. On a host's stat history, Add stats by hand creates a row for a period (not a future one) and Edit changes one. Agency admins can always do this. For other staff, the permission role needs Platform stats: Enter and edit by hand and also Talent: Edit, because the host's page is an edit page. A period has one row: a later CSV upload for the same period overwrites the fields it carries and takes over the row's source, and a blank cell in an upload leaves the value alone. Saving a stat does not by itself award a mission or send a campaign, and it does not change a statement that already exists. A statement generated afterwards, for a month that has none yet and for a host not deleted before that month, adds up the new numbers. Missions read a host's latest period and campaigns read the latest two, each the next time it runs. Hand-entered stats are not recorded in the change history.
If your current process is a spreadsheet per week, Live Agent Pro replaces what you do with that spreadsheet — compensation tiers, alerts, campaigns, statements, points — not the act of exporting it from the platform.
Where stats come from #
- CSV import — the main route. Two paths: a per-host Upload CSV (Live Stats) button on the host's stat history, and the Documents area's reusable column mapping for spreadsheets whose headings don't match ours. See Documents & CSV Import.
- Automatic TikTok pull — the four audience counts above, on demand. See TikTok Connections.
- By hand — Add stats by hand on the host's stat history, for a number you have and no spreadsheet for.
Every row records which of these wrote it last, so a typed row and an imported row are distinguishable. A TikTok pull does not write a row (see below).
A note on dates #
An imported row carries the period date from your spreadsheet, which is what makes week-over-week comparison work. TikTok's pulled counts are lifetime running totals rather than a period's activity, so they are not a stats row: a host's follower count is "as of the last sync", not "gained this week". The host's TikTok platform record shows the last pulled value of each count. Campaign conditions on the four counts, the follower count in message templates, and followers missions use whichever is newer: the last pull, or the latest stats row if it has the count (by its period date; on the same day, the stats row). The last pull still counts after a host's TikTok connection is removed. Each day's pull is also kept, one entry per day, and included in your full data export.
Comparing reports within a month #
A month-to-date spreadsheet imported through a Documents column mapping, with a
range in its date column (for example 2026-11-01 ~ 2026-11-08), is also kept as
a report: the numbers each host had for that range. The month's stats row
holds only the latest report, so the reports are what lets one be compared with
the last. Importing the same range again updates its report; a later range adds
one. Two campaign conditions read them, from the reports of the
month a host's latest stats are for:
- Unchanged since the last report is yes when the host's hours and diamonds equal those of the most recent earlier report this month that had either. An earlier report with no hours and no diamonds for them is passed over. It is empty when there is no such earlier report this month.
- In the previous report is yes when the host is in the report just before this one, on this platform this month, whatever their numbers. It is no when this is the first report of its range start this month.
Both compare only reports that start on the same day, so reports for weekly
ranges (2026-11-08 ~ 2026-11-14) never count as one another's previous report.
Both are empty for a host who isn't in the latest report for the host's
sub-agency: each sub-agency is compared with its own reports (and
platform accounts in no sub-agency with theirs). Each report belongs to the
sub-agency its platform accounts were in when it was imported, whoever imports
it. After a platform account moves to another sub-agency, or its sub-agency is
deleted, it is compared from its new group's next report on. An empty value
is neither yes nor no: "equals 1" and "equals 0" leave the host out, and "not
equals 1" and "not equals 0" take them in.
Only Documents imports with a range make reports. A per-host upload, a TikTok pull or stats typed in by hand don't, so both conditions are empty for a host with no report in the month of their latest stats. Reports start with the first such import after this feature was released: that import can't show anyone as unchanged, and shows everyone as not in the previous report. Reports are included in your full data export.
Momentum: comparing the last two periods #
Several features compare a host's two most recent periods to show whether they're improving: diamonds, live duration, and valid live days each have a change value. A change value is empty until at least two periods exist for that host.
You can use these as conditions in campaigns and compensation tiers — for example, to target only hosts who are improving, or only those who have dropped off.