TMS

TMS-CSA-1126

Excel & Power BI for customer service analytics

By Thursday afternoon each manager can take a messy service extract, tell the truth about it, and brief one decision: where service is slipping, why, who it hits, and what changes before next Monday.

Facilitator
Ahmad A. Khan
When
16–19 November 2026
Monday to Thursday, 09:00–16:30
Room
16 managers, in pairs
25 hours in the room · Instructor-led, in the room

Map

Full course structure

Every day, every session, the aim of each block. 16–19 November.

Handout

Participant pack

Outline, manual, editable slides, and the Excel workbook Power BI loads on Wednesday.

The four questions

Every lab, every page, and the Thursday brief has to serve one of these. A chart that does not is decoration.

  1. 01

    Are we keeping the promise on speed?

    SLA, first response, resolution. Not a blended average that lets a bad hour hide inside a calm week.

  2. 02

    Where is effort being wasted?

    Reopens, transfers, repeat customers, status-chasing. Volume is not effort. A closed ticket that comes back was not finished.

  3. 03

    Who is leaving unhappy?

    CSAT, NPS, and complaints — with the base beside the score. A person, a queue, or a policy. Do not guess which.

  4. 04

    What changes before Monday?

    One action, one owner, one number to watch for two weeks. If the chart does not change a decision, it does not go in the pack.

Who it is for

Senior customer service managers who own a queue, a site, or a channel. They already get a weekly pack they do not fully trust, and they need to brief a director without waiting on an analyst.

Who it is not for

Analysts who model in Power BI every day. New agents. Anyone whose only job is to publish a report someone else designed. This room makes decisions. It does not produce a report factory.

Before they arrive

  • Install Excel desktop (Microsoft 365) and Power BI Desktop. Open each one once, on the laptop you will bring.
  • Confirm you can save a file on that laptop. A locked-down machine is the usual failure, and it is not a Day 1 problem.
  • Read the definitions on the Brief. No other pre-reading.
  • Bring one question your director actually asked in the last month. Do not bring live customer data. Practice data stays in the room. Work data stays at work.

Four days, one file

The practice operation is TMS Care. Ticket log, complaint log, and CSAT / NPS invites for Q3 2026, read as of 1 October 2026, 18:00 Asia/Qatar. The same rows run from Monday’s trust check to Thursday’s brief.

Day 116 Nov

The question before the chart

A chart without a decision is decoration. A dirty row makes the chart gossip.

Open the run sheet

Day 217 Nov

Where the work actually goes

Volume is not effort. The expensive work is the work that should not have existed.

Open the run sheet

Day 318 Nov

One model, three files

Excel answered the question once. Power BI is how the question stays answered when the file refreshes.

Open the run sheet

Day 419 Nov

The Monday briefing

Finished means a skeptical peer cannot pull the number apart, and a director knows the ask.

Open the run sheet

Open the practice file in the lab

What they can do by Thursday

  1. 01Write a service question a sheet can answer: metric, grain, window, and the decision it serves.
  2. 02Trust-check a ticket log and flag bad rows without overwriting the extract.
  3. 03Build an Excel summary of breach rate, reopen rate, and open work that is already late.
  4. 04Read a complaint file: aging, upheld, and concentration — and refuse a remedy the counts do not support.
  5. 05Join tickets, complaints, and surveys on the right key without double-counting.
  6. 06Write a short set of Power BI measures you can say out loud, including NPS that does not treat silence as a score.
  7. 07Design one page a director can read in two minutes.
  8. 08Brief it: observation, evidence, what it does not prove, and the ask.
  9. 09Set a weekly rhythm: who refreshes the extract, which three numbers stay, which three retire.

Language of the room

These words mean one thing for four days. Do not renegotiate them mid-lab.

Grain
What one row is. In the ticket file, one row is one ticket. A survey row is one invite, not one customer.
Window
The dates the number covers. This file is Q3 2026, read as of 1 October 2026 at 18:00.
SLA breach
Resolve time past the promise, or still open past the promise at the as-of clock. P1 4h, P2 24h, P3 48h, P4 72h.
Reopen
The ticket came back. A re-closed ticket can still meet the clock. The flag is the signal, not the SLA.
CSAT
Mean of 1–5 on replies only. Always say how many replies, and how many invites went out.
NPS
Share of 9–10 minus share of 0–6, among replies. Silence is not a zero. Report it with n.
Effort
Mean of 1–5 on replies. Higher means the customer worked harder. Do not flip the scale mid-course.
Response rate
Replies divided by invites, same window. Email and phone are not comparable until this sits beside the score.
Repeat customer
Two or more tickets in the file on the same customer id.
Complaint
Its own case. Some link to a ticket. Some do not. Dropping the blanks drops the walk-ins.
Upheld
Yes or Partial means the complaint had merit. No means rejected. Open is neither.
As-of
The clock you freeze for open tickets. Here it is pinned. If you use now, tomorrow’s file disagrees with today’s.

What a complete run includes

Outcomes. Written as something a manager can demonstrate, not as topics a facilitator can cover.

Run sheets. Each day has times, an aim, the moves, and the sentence that opens the room.

One file. Labs stack. Tuesday uses Monday’s flags. Wednesday loads the same rows.

Labs before slides. The file teaches. The facilitator does not tour menus.

Assessment in advance. A check each day. Thursday’s rubric is visible on Monday morning.

Records. Attendance, a same-day feedback form, and a one-page note for the sponsor. No quotes by name.

A completion line. In the room for the labs, and a brief that scores 70 or better. Partial days are covered by the file, submitted that evening.