For the complete documentation index, see llms.txt. This page is also available as Markdown.

Label usage

See how your labels are being used across conversations over time — which labels are applied most, and how that breaks down day by day. Useful for understanding what's coming in (product lines, customer types, campaign sources) and spotting shifts across the week.

It's period-based and follows the shared rules in How we calculate analytics.


The key rule: counted by when a label was assigned

A label is counted on the day it was assigned to a conversation — even if it's removed later. Applying a label is a point-in-time event, and that's what this page counts: assignments that happened inside your selected period.

This has two important implications:

  • Removing a label later doesn't undo the count. If a label was applied on Tuesday and taken off on Friday, Tuesday still shows it. (This is the same "your past reports never change" principle from the calculations page.)

  • It counts assignments in the window, not labels currently on conversations. A label applied last week shows in last week's report, not this week's — even though the conversation may still carry it today.


How labels differ from ticket fields

Ticket Fields and Label Usage look similar — a value with daily counts and a total — but they count differently, so the same conversation can behave differently on each page:

Ticket Fields
Label Usage

Lives on

a ticket

a conversation

How many per item

one value at a time

many labels at once

Counted at

the value held at closure

the moment a label is assigned

If changed/removed later

only the closure value counts

the original assignment still counts

What's included

closed tickets only

any labelled conversation, open or closed

In short: a ticket field is a single settled answer captured when the ticket resolves; a label is an event you can stack, and each application is counted when it happens.


Reading the table

Each row is a label (with its colour dot), showing a count for each day in the period and a Total on the right. Cells are shaded as a heatmap — darker means more assignments that day — so heavy days jump out. The table is sorted by Total (most-used labels first).

Because each assignment is a point event, the daily counts add up to the Total (assignments are additive across days, unlike unique-count metrics).


Label sections

Labels are grouped into sections on the page. The screenshots show two:

  • Customer labels — labels applied to conversations (e.g. messenger_ads, APPLE New1, No Feedback / ไม่ตอบ, General information / สอบถามข้อมูลทั่วไป, membership tiers). These can be applied manually or by your automation rules.

  • Meta Ads and Comment Automation labels — source and post labels (e.g. ad_id.120238896238300416) that auto-assign to conversations originating from Meta ads or posts. This lets you see which ad or post drove each conversation, so you can tie support volume back to a campaign.


"Label deleted" rows

When a label is deleted from your workspace, its historical usage is preserved — the assignments that already happened still count — but the name is gone, so it appears as Label deleted. Each deleted label keeps its own row (they're not merged into one), so you'll often see several Label deleted rows, each representing a different label that was removed.


Filters and controls

  • Date period — the window the page is calculated for (shown as both the preset and the resolved range).

  • Label — narrow to specific label(s).


Good to know

  • Counted even if removed — a label counts for the day it was applied; removing it later doesn't change the number.

  • Labels are on conversations, and a conversation can carry several at once — so this page counts differently from the ticket-based pages.

  • Deleted labels still count — shown as Label deleted, by their original assignment date.

  • vs Ticket Fields — see the comparison table above for how the two pages differ.

Last updated