A verified LINE Official Account can publish an average reply-time score on its public Business Profile — a number a customer sees before sending a message. LINE’s own help documentation says the calculation counts only genuine one-on-one replies: auto-response and AI response messages are excluded entirely, so turning on a chatbot doesn’t move the score either way.
What the “average reply time” actually counts
LINE’s help center defines it plainly: “the reply time is an estimate of how long it will take a LINE official account to reply to messages from users.” It covers one-on-one chats only, over a rolling 30-day window. A measurement starts at a user’s first message and counts only replies sent inside the next 24 hours — a later reply drops the session out of the average rather than dragging it down, and the user’s next message starts a fresh one. By LINE’s own wording, auto-response messages, AI response messages, group chats, and spam-flagged users are all excluded (LINE Official Account Manager help center). It only runs once Chat is on under Response settings — an OA that only broadcasts has nothing to measure.
Why an AI agent doesn’t move the score
A message your AI agent answers alone is, by LINE’s own definition, an AI response — filed with a canned auto-reply, not a human one. The number measures how fast a person replies inside a one-on-one thread, not how fast your shop “answers” generally. An AI agent that resolves most questions neither helps nor hurts this figure; it just removes those chats from the sample. What moves it is the opposite case: a question the AI can’t close, where a customer waits on a person — the moment covered in when a chatbot should hand the cart back to a human. As a recommended default, treat anything the AI escalates as worth routing fastest.
Turning the public score on, carefully
Showing this number is opt-in: the toggle sits in Business Profile settings, under Basic info, as “Show average reply time,” and publishes whatever the last 30 days looked like — flip it before checking the team’s real numbers, and a slow week publishes itself. Our own recommended default, not a published benchmark: watch the number privately for two weeks, then recheck after any change to staffing or escalation rules.
ตั้งค่าใน 1CS: let a human take the handoff inside LINE itself
The actions that move the score are the ones where a person decides fast, and 1CS’s AI ปรึกษาผ่าน Line (“AI Supervisor”) menu, under AI, is built for that moment — the LINE-side counterpart to writing an escalation rule from the words customers actually type. Connect a team LINE group once: add the LINE OA “1CS Thailand” as a friend, invite it into an admin group, then paste a connection code from the same page into the group. Turn on “ถาม LINE group ก่อน escalate” (“ask the group before escalating”); whenever the AI would otherwise hand off, it posts a card into the group with the question, its draft answer if any, and a 10-minute clock. Any admin can tap ส่งเลย (“send now”) to approve the draft, แก้ก่อนส่ง (“edit before sending”) to type the reply, or รับเคสเอง (“take the case”) to pull it into 1CS. No answer in 10 minutes, and the case still routes into 1CS as normal, losing only the chance to close it from LINE. This sits inside LINE Admin Consult; steps are in 1CS’s LINE AI Supervisor guide (Thai only).
The checklist for every card that lands in the group
Keep this next to the group so the 10-minute window never runs out by accident:
- Card lands at [time] — the clock is already running, not from when someone notices it.
- Draft is accurate as written → tap ส่งเลย.
- Draft is missing or wrong on one detail → tap แก้ก่อนส่ง and type the fix.
- No usable draft, or the case needs more than one line → tap รับเคสเอง.
- Nobody free in 10 minutes → say so in the group, so whoever opens the case in 1CS next knows it’s still waiting.
LINE publishes what counts, not the exact math, so none of this guarantees a number. A fast, consistent handoff does guarantee that conversations an AI agent can’t finish stop being the ones that sit longest — the half of the score a shop actually controls.
