Six targets from the job description, computed from ticket data. Load your own export (CSV columns: id, channel, lang, advisor, driver, created, frt in minutes, res in minutes, reopened, csat 1-5, qa 0-100) or use the synthetic month.
Daily volume
What the lead does with this
Advisors · weekly 1:1 input
| Advisor | Tickets | Chat FRT | Email FRT | CSAT | QA | Reopen | Coaching focus |
|---|
Contact drivers (tags)
| Driver | Tickets | Share | Last 7d vs prior | CSAT | Published answer |
|---|
Approve or push back on leave with numbers. Chat arrivals per hour (from the ticket data) run through Erlang C to get the advisors needed for the 4-minute first response, then email workload and shrinkage are added. Chinese-language cover is checked hour by hour.
Hour by hour (GMT+8)
| Hour | Chats / h | Needed | On shift | 中文 have / need | Who |
|---|
Leave on file
Method
Erlang C gives the servers needed for the target share of chats to be answered within 240 seconds; servers are divided by chat concurrency to get advisors. Daily email workload is spread over a 16-hour day. Shrinkage covers breaks, coaching and QA time. Weekends use weekend volume. The 16 Sep incident day is excluded from the baseline.
A QA evaluation in two parts. Automatic checks read every advisor line: facts are held against Xapo Bank's published FAQ, compliance checks look for tipping-off, investment advice and credential requests, and process checks cover identity on email, complaint recognition and timeframes. The reviewer then scores four manual items. A critical finding fails the evaluation.
Or paste a transcript
Draft the next reply with AI, then QA it
Workers AI drafts from the published facts only; the same QA checks then run on the draft. The advisor edits and sends.
Findings
Manual scoring
Coaching note
QA log
| When | Ticket | Advisor | Reviewer | Score |
|---|
Calibration
Reviewers who score the same ticket more than 5 points apart need a calibration session before their scores count toward the 95% target.
The complaints queue on its clocks. Xapo's published complaints procedure promises an update if a complaint is not resolved within 15 days and a final response by day 35; it does not say business or calendar days, so both are selectable. The report at the bottom is the monthly pack for the MLRO.
Register
| Complaint | Opened | Lang | Root cause | Status | 15-day update | Final response |
|---|
Monthly pack for the MLRO
Two JD duties: the four-eyes check on manually created documents before they reach a customer, and the six-monthly audit of tags and shortcuts (macros). Shortcuts are checked against the same published facts as QA.
Four-eyes: balance confirmation letter
Checker result
The IBAN check is the ISO 13616 mod-97 checksum, with the Gibraltar length of 23 characters.
Shortcut audit
| Shortcut | Text | Finding |
|---|
Tag audit (6 months)
| Tag | Uses | Action |
|---|
Checks that run against Xapo Bank's public pages on each visit: the fact base against the FAQ, every help-centre article against those facts, and the Chinese help-centre experience.
Fact base, verified against the live FAQ
| Fact | Used by QA | Status | Live quote |
|---|
Help centre (customersupport.xapo.com)
| Article | Conflict | Sentence |
|---|---|---|
| Run the check to read all articles (about 20 seconds). | ||
Article coverage by contact driver
| Driver | Articles | Newest |
|---|
Matched on article URL keywords.