←BaldanAll worksLet's talk

Case study · Indosat · Circle Kalisumapa · 2026

Fulfillment Manpower: every sales seat filled, or flagged.

A circle has a fixed number of field-sales seats, spread over regions, areas and MCs and staffed by outside agencies. This app shows who sits in each seat, how long the empty ones have been empty, and fulfillment against target. It replaces the shared master workbook, and still imports and exports that exact format.

Built end to endNext.js · FirestoreExcel in, Excel out
Try the rules ↓
fulfillment · dashboard
Fulfillment dashboard on demo data: quota, active, fulfillment against target, vacant slots, vacant days, fulfillment per region and role

Screens: the real app running on made-up demo data. No real people, outlets or figures.

01 · How it's used

Every seat accounted for.

  1. 1

    Sign in, get scoped

    The session says circle, region or agency, and every read is filtered to that.

  2. 2

    Read the dashboard

    Quota, active, fulfillment against target, vacant seats, average and longest vacancy, and the critical count, then a Region › Area › MC tree.

  3. 3

    Work the vacant board

    Empty seats, longest first. Filling one takes a name and a join date.

  4. 4

    Record a resignation

    A date and a reason: the seat turns vacant, the record is archived and the vacancy clock starts.

  5. 5

    Sync with Excel

    The circle uploads the full workbook (preview, confirm, automatic backup, replace); anyone exports their own scope.

02 · Try the rules

Resign a seat, let the days run.

One made-up MC with eight seats. Resign or fill seats and drag a vacancy's days: the colours and the fulfillment bar follow the app's own thresholds.

Quota

8

Active

5

Vacant

3

Avg vacant days

24.0 d

Critical > 30 d

1

Fulfillment

62.5%

DSE-01

DSE

Seat holder A

DSE-02

DSE

DSE-03

DSE

Seat holder B

DSF-01

DSF

Seat holder C

DSF-02

DSF

AE-01

AE

Seat holder D

PRM-01

Promotor

PRM-02

Promotor

Seat holder E

03 · Under the hood

Seats are the quota.

  • Quota = seats

    Each seat has an ID made from its role and a unique key. A new seat starts vacant from today, and only a vacant seat can be deleted.

  • Vacant days

    Days since the resign date, counted in Jakarta time. Over 14 days is yellow, over 30 is red and counted as critical.

  • Fulfillment

    Active ÷ quota against a 95% target: green from 95%, amber from 85%, red below.

  • Resignations

    Reasons are kept, counted per month and ranked; a resign date before the join date or in the future is refused.

  • A careful Excel import

    Up to 10 MB; the file applied must be the very file that was previewed; eight sheets' headers must match; dates are validated; duplicates and unknown areas block the import, warnings don't.

  • Backup before replace

    Every import writes a backup first (the newest 20 are kept), then replaces the data in one go.

04 · Who sees what

Three levels, enforced in the data layer.

Checked on the server for every read and write, not only hidden in the screens.

Circle

Everything: Excel upload, backups, users and the audit log.

Region

Its own region's seats; can add, edit or remove seats there.

Agency

Only the seats it staffs; can fill seats and record resignations.

05 · The small stuff

The parts nobody sees.

Preview with consequences

The upload preview counts added, removed and changed seats per role and warns about active people who would disappear.

Exports that still calculate

The exported workbook has live formulas and highlights duplicate IDs, vacant rows and long vacancies.

Safe writes

The JSON store queues writes and swaps files atomically; Firestore writes in batches, adding before deleting.

Sized for the free tier

One document per MC keeps a full load around 170 reads, with a live cache.

Business rules

Some seat types belong to one agency only, some need a PT; a unique ID can't be edited.

No open redirects

After login it only returns to a local path.

Accounts

New users get a one-time set-password link; a role change signs them out everywhere.

Two modes

Runs on a local JSON file for demos, or Firestore with real sign-in in production.

06 · How it fits together

Server components over a swappable store.

Browser

Next.js screens

Dashboard tree, vacant board, roster, resign archive, Excel upload and export.

Cloud Run

Server actions + data layer

Session → scope checks → validation and audit → the store.

Firestore

seats per MC, resigns, audit, backups

Firebase Auth

sign-in, session cookie, claims

Excel

The master workbook

Imported and exported in the same format the team already used.

07 · Tech stack

What each piece does.

Next.js 16 + React 19

Screens and server actions.

Firestore

Seats, resignations, audit and backups.

Firebase Auth

Sign-in, session cookies, role claims.

ExcelJS

Reading and rebuilding the master workbook.

Node crypto + zlib

Signed demo sessions, file hashes, gzipped backups.

Docker on Cloud Run

Hosting.