Automation multiplies whatever process you already have - including the broken ones. Re-engineer first, then digitise.
Every quarter we meet a company that has bought excellent software to fix a process problem, and now has an excellent, expensive process problem. The approvals that used to be slow on paper are now slow in a workflow tool. The data that used to be inconsistent across registers is now inconsistent across modules.
The lesson is old but relentlessly relearned: automation multiplies the process it is given. Feed it a broken process and it will produce broken outcomes at scale, with better logging.
Map what actually happens
Process re-engineering starts with an uncomfortable exercise: documenting the process as it truly runs, not as the manual claims. Who really approves this? Where does the file actually wait? Which steps exist only because of an incident five years ago that nobody remembers? In most mid-sized businesses, a quarter of the steps in any core process serve no current purpose.
Design for the exception
The happy path is easy; processes break at the exceptions. What happens when the approver is on leave, the vendor is new, the invoice does not match the PO? A good SOP answers these before the software vendor is even shortlisted - because every unanswered exception becomes either a workaround that bypasses control, or a ticket that stalls the business.
Write SOPs as decision documents: trigger, owner, timeline, escalation, evidence. One page per process, reviewed annually, signed by the person who runs it - not the consultant who drafted it.
Then, and only then, digitise
Once the re-engineered process is running manually and cleanly for a quarter, automation becomes an accelerant instead of a gamble. Requirements write themselves from the SOP. Configuration debates end quickly because the business already knows its own rules. And the software finally does what software does best: enforce a good process at a scale humans cannot.
Scale is not a software purchase. It is a process decision that software later makes permanent. Make the decision first.
Written by
Paramjeet Singh
Writing field notes on finance, tax, process and infrastructure - from the work, not about it.
