work / automation / multi-client financial reporting

Month-end reporting for 40+ clients, done before anyone asks.

An accounting firm rebuilt every client's monthly numbers by hand. I built one workflow that pulls each client's financials and hours and fills their dashboard automatically.

role
Sole automation engineer
client
Canadian accounting firm
status
Delivered, in monthly use

Languages

JavaScript

Tools & platforms

n8nQuickBooks Online APIHarvest APIAirtableOAuth 2.0Bitwarden

AI & LLM

to build it
ClaudeChatGPTPerplexity

Skills

API integrationOAuth token managementHistorical backfillError reportingParameterized workflows
40+

client configurations from one control point

~15 min

down from hours (director-timed)

11

months of history backfilled

01// how I approached it

The thinking behind it

  1. 01 · What I noticed

    Every client was the same job with different settings. The team was rebuilding the same report 40+ times.

  2. 02 · What I decided

    Build one workflow with per-client settings, so a fix or a new metric reaches every client at once.

  3. 03 · What I delivered

    Month-end for 40+ clients in about 15 minutes instead of hours.

02// old vs new

Old system vs new system

Old systempeople as the glue
  1. Export each client report
  2. Copy numbers into sheets
  3. Pull hours separately
  4. Update each dashboard
  5. Repeat for 40+ clients
New systemthe system does the work
  1. Month closes
  2. One workflow loops every client
  3. Financials and hours pulled
  4. Dashboards updated, failures flagged
03// how it works

Watch it run

Click through the four steps. In step 3, click any box to see what it does.

synthetic data · no client screens
Reconstruction with synthetic data · no client screens
04// the hard part

One workflow, 40+ different clients.

Every client has its own connection, its own settings and its own quirks. I built a single control point that resolves each client at run time, keeps their sign-in tokens fresh, and marks any metric that fails instead of silently skipping it. Six edge-case bugs found in real runs were fixed along the way.

05// key decisions

Why it's built this way

decision 1

One parameterized workflow, not 40 copies

A fix or a new metric reaches every client at once.

Trade-off: client settings must be kept clean.
decision 2

Tokens refreshed and written back automatically

Connections keep working month after month without manual sign-ins.

Trade-off: careful handling of credentials.
decision 3

Failures are visible

A failed metric is marked as failed, so nobody trusts a wrong number.

Trade-off: more status logic.