Chapter 3

Skip to content

Chapter 3

The Age of Automation (1990s–2000s)

ERP promised a single source of truth. We got standardisation and speed — plus a new kind of blindness: assuming the system was right because it ran without error.

1. The ERP Proposition

Integrated suites connected sales, purchasing, inventory, finance, and HR on one database. Batch jobs replaced handoffs; journals posted themselves; month-end closed faster than ever. Dashboards showed KPIs at a glance.

Theme tie: Automation removed friction — and with it, many moments where humans once asked, “Does this make sense?”

2. Configuration Is Control

ERPs are policy engines. They do exactly what they’re configured to do — and nothing more. When UoM conversions, tax codes, or posting rules are wrong, the system will be precisely wrong at scale.

  • Master data: Items, customers, suppliers, GL mappings.
  • Posting rules: Accruals, cost flows (FIFO/Weighted), discounts, returns.
  • Authorisations: Roles, approval workflows, segregation of duties.

The illusion of control comes from screens that look official. Real control comes from tested configuration and reconciliations.

3. Reconciliation Erosion

Because subledgers auto-post, teams assume AR/AP always match control accounts. When exceptions occur — a failed batch, a custom integration, a partial migration — mismatches hide for months behind green status ticks.

4. Customisation and Lock-In

Gaps between process and software were bridged with custom fields, user exits, and add-ons. Over time, upgrades became risky and costly. Workarounds turned into shadow policy.

5. Shadow IT and the Spreadsheet Hydra

Even with ERPs, analysts exported to spreadsheets to model specials, forecasts, and reconciliations. Versions multiplied; logic forked. The ERP was the system of record, but critical decisions lived outside it.

6. What Good Looked Like

  • Change control: Every configuration change ticketed, reviewed, and tested.
  • Closed-loop reconciliations: Subledgers to control accounts monthly; inventory to GL; GRNI/PO accruals cleared.
  • Master-data stewardship: Owner per domain; dedupe and validation routines scheduled.
  • Role design: Least-privilege, SoD checks, periodic access reviews.
  • Fit-to-standard first: Customise last, with explicit business case and upgrade path.

Practice takeaway: Treat configuration like code. Version it, test it, and roll it out deliberately.

7. Tales from the Trenches

Insert your story here

Prompt: A posting rule misrouted costs for months. How did the variance surface — stock valuation, margin dip, or audit query?

Prompt: An AR/AP mismatch hid behind automated posting. What triggered discovery, and how did you unwind it?

Prompt: A custom field/add-on blocked an upgrade. What was the business case, and how did you refactor or retire it?

8. Theme Tie — The New Blind Spot

ERPs professionalised process — but they also professionalised complacency. When the machine runs smoothly, humans stop interrogating inputs. Chapter 4 will argue that AI revives this risk at higher speed and scale.

← Back: Chapter 2 Next: Chapter 4 — The Rise of AI (2010s–2020s) →

Appendix ideas to add later: a month-end reconciliation checklist, a configuration change log template, and SoD test examples.