The most expensive mistake in automation projects is also the most common: taking a manual process, exactly as it runs today, and scripting it. The result is a faster version of the same dysfunction – automated handoffs of information nobody needs, in an order that no longer makes sense, at a speed that just gets you to the wrong outcome sooner. Process redesign has to come first. Here is the method that works in practice.
Map the process as it actually runs
Every redesign starts with an honest map of the current state. Not the process from the six-year-old SOP document – the real one. Interview the people who touch the work, follow a few real cases end to end, and record every step, handoff, system, spreadsheet, and approval. The map should capture three dimensions for each step: who does it, how long it takes, and what triggers it.
Expect surprises. In most audits, the documented process and the actual process diverge within the first five steps. Those divergences – the shadow spreadsheets, the email approvals, the re-keying between systems – are precisely where the waste hides.
Question every step before automating it
With the as-is map in hand, run each step through a strict sequence of questions:
- Eliminate: Does anything downstream actually use this output? Steps that exist only because “we’ve always done it” should be deleted, not automated.
- Simplify: Can the step be made smaller, standardized, or merged with a neighbor? A form with forty fields usually means nobody knows which twelve matter.
- Standardize: If the process runs differently in three regions, automation will cement the variation. Pick one canonical flow first.
- Automate last: Only now, automate the steps that remain – starting with the highest-volume, lowest-judgment work.
The order is the point. Elimination and simplification routinely cut the scope of an automation project in half before a single workflow is built.
Redesign the handoffs, not just the steps
In B2B processes, the delays live between steps, not inside them. An invoice sits for two days waiting for someone to notice it in a queue; a lead cools for six hours until sales checks the CRM. When you redesign, turn implicit handoffs into explicit, event-driven triggers: a record submitted in one system should arrive, enriched and routed, in the next – with an SLA attached.
This is where workflow platforms earn their keep. A well-built automation makes the handoff instant, logged, and observable, so process owners can see queue depth and cycle time instead of guessing.
Design for the exception, not just the happy path
Every process has a happy path – and the 20% of cases that do not follow it consume 80% of the operational pain. A redesign that only automates the ideal case simply concentrates all the remaining manual work into a firefighting queue. For each automated workflow, define explicitly: what happens on validation failure, on timeout, on a missing field, on a duplicate? Build a dead-letter path with a human owner, and measure exception rates as a first-class metric.
Measure, then iterate
Baseline before you automate: cycle time, touch time, error rate, cost per transaction, and the number of people who have to touch each case. After go-live, compare against the baseline monthly. Redesign is not a project with an end date; it is a loop. The teams that treat their first automation as version one – instrumented, reviewed, and improved quarterly – compound their gains. The teams that declare victory at go-live usually rebuild the same process from scratch within two years.
Map honestly, eliminate ruthlessly, automate what remains, and design the exceptions. That is the whole discipline – and it is what separates automation programs that transform operations from those that merely decorate them.
Read next: SaaS Sprawl Is Eating Your Budget: How to Audit and Rationalize Your Stack