Chat & automation quality
Demo · invented dataConnect your Page
Chat & automation quality
ช่างใหญ่การช่าง · Messenger chats customers started in the last 7 days (12/11–18/11/2026) · vs 05/11–11/11/2026 · business hours 08:30–18:00 · data as of 11:02
Follow up first: 12 chats where the customer asked and nobody replied
78/100
Reply quality score
75 → 78
▲ 3 ptsvs 05/11–11/11/2026
measured 85 of 100 · Level: fair
Too few
Closed sales (interested who continued)
continued 9 of 19 · sold 4 chats · ฿9,540
27% bot only
Who answers chats · target bot only 50%
Bot only 27% · Handed over 23% · People only 27% · Nobody replied 23% · of 56 chats
What customers message about, and who answers
56 chats · topics from word rules on the customer's text (not AI)Who replied: Bot onlyHanded to a personPeople onlyNobody replied
| Customers message about | Chat | Who replied | Customer kept typing | Nobody replied | |
|---|---|---|---|---|---|
| Buying | 19 | 56% | 2 | ||
| General questions | 13 | 54% | 0 | ||
| After-sales | 8 | 40% | 3 | ||
| Ghosted | 6 | Nobody replied | 6 | ||
| Other | 6 | 20% | 1 | ||
| Order follow-up | 4 | 75% | 0 |
Buying: most, 19 of 56 chats · buyers who kept typing after the answer 56%
What to fix first to raise the score
Biggest point loss first · only areas with 10+ chats| Issues | Pass | Points lost | What to do | Example chats |
|---|---|---|---|---|
| BuyingCustomer continues after the answer | −1.1 | The first answer should ask one more thing (model · quantity · delivery province) so the customer keeps talkingSales team | Chats 1 · Chats 2 · Chats 3 | |
| BuyingAnswer every chat that asks | −0.9 | ตั้งข้อความตอบอัตโนมัติให้เรื่องนี้ หรือให้มีคนเฝ้า Messenger ช่วงที่ลูกค้าทักมากTeam lead | Chats 1 · Chats 2 · Chats 3 |
Fix Buying · Customer continues after the answer first — 56% pass, costing 1.1
Chats to open
24 chats · most serious first (nobody replied › handed over, not taken › slow), newest first within each · one-message chats last| Customer messaged | Page | Topic | Why look | |
|---|---|---|---|---|
| 17/11/2026 15:31 | สาขาบางนา | Buying | Handed over, nobody took it waiting 1 h 28 min | Open chat |
| 13/11/2026 22:38 | เครื่องมือช่าง | Buying | Handed over, nobody took it waiting 1 h 58 min | Open chat |
| 18/11/2026 09:44 | สาขาบางนา | Order follow-up | Handed over, nobody took it Unassigned | Open chat |
Nobody replied 12 · Handed over, not taken 3 · slow 2 · wanted to buy, went quiet 7 — a chat can have several reasons
How much automation and AI help
Messenger bot · after-hours comment replies · AI drafts · 7 days| Metric | This period | vs previous period |
|---|---|---|
| Bot only, of chats the bot answered | Too few | Too few to compare |
| Automatic after-hours comment replies | Not on yet | Comparable once it is on |
| AI drafts sent unedited | 51% | ▲ 1 pt |
| AI drafts used (edited or not) | 83% | ▲ 5 pts |
| AI read the comment right (the team did not correct it) | 100% | ▲ 1 pt |
The team sent 51% of 47 AI drafts unedited
Latest bot data
Is the bot wrong because its data is old · by Page · the ones to check first| Page | Status | Bot's latest reply | Active promotions | Bot mentions an expired promotion | Ads with a known product |
|---|---|---|---|---|---|
| ช่างใหญ่การช่าง เครื่องมือช่าง | Stale Check | 18/11/2026 10:47 | 1 promotions | “ลดล้างสต็อกดอกสว่าน 30%” 4 messages | 2 of 2 |
| ช่างใหญ่การช่าง สาขาบางนา | Latest | 18/11/2026 09:47 | 1 promotions | ไม่พบ | ไม่มีลูกค้าจากโฆษณาใน 30 วัน |
1 of 2 pages have bot data to check — start with ช่างใหญ่การช่าง เครื่องมือช่าง: the bot still names expired promotions in 4 messages
How the 78 score works — 4 measured areas · 85 points in all · open each one
- Answer every chat that asks3023.0 / 30 · 77% · 43 of 56
- First reply within 15 min2523.8 / 25 · 95% · 41 of 43
- Handed over and taken within 60 min1511.5 / 15 · 77% · 10 of 13
- Customer continues after the answer157.7 / 15 · 51% · 22 of 43
- Closed sales (vs a 30% target of interested people who continued)15Too few (9 of 10 chats) · not counted yet
Reply quality score 0–100 = weighted average of the measured areas · closed sales vs a 30% target (target reached = full marks) · areas with fewer than 10 chats are not counted yet, so read it with measured n of 100 points · time counts business hours only (08:30–18:00)
only this business's Messenger chats started in the period · Thai time · chat topics come from words in the customer's messages · phone numbers and addresses in customers' messages are hidden