Process Simplification Before Automation in Banking

Simplify the Process Before You Automate

Ford cut accounts payable head count by 75% by redesigning work. Process simplification gives banks a smaller and cheaper process to automate.

Alexander Kiel

Business Analyst · Banking & Wealth Management

17.10.2026 · 7 MIN READ · PROCESS OPTIMISATION

ON THIS PAGE

1. Map the Process as It Runs

2. Apply Process Simplification to Each Step

3. Standardise Variants Before You Code

4. Fix the Inputs at the Source

5. Automate What Remains Stable

6. Measure the Result Before Scaling

Sample case: Lloyds Banking Group

In short

FAQ

Automation makes a process faster, not better. A robot that copies data between two screens repeats every unnecessary step, check and handoff, only quicker. Process simplification removes that waste first, so that what remains is worth automating. Gartner expected organisations to lower operational costs by 30% by combining hyperautomation technologies with redesigned operational processes.

The warning is old. Michael Hammer wrote in Harvard Business Review in 1990 that companies use technology to mechanize old ways of doing business. Banks still carry that pattern: workflows built around paper forms, approvals added after past incidents and manual checks that nobody dares to remove.

Automating such a process locks the complexity into code that is harder to change than a procedure. Which steps in your process would you design again if you started today?

Illustrative example of a payment investigation process: 14 steps as is, 8 steps after simplification and 8 steps with 5 automated and 3 manual after automation

STEP 1

Map the Process as It Runs

Start with the real process, not the documented one. Follow a few cases from request to completion and record every step, handoff, approval, system and waiting point. Note who touches the case and why. Ask staff where they work around the system, because workarounds point to steps that the design missed.

Fragmentation is often extreme. Hammer described an insurer where an application went through as many as 30 discrete steps, spanning 5 departments and involving 19 people. My article on as-is process mapping shows how to capture such a flow on one page before you change it.

✓

Follow real cases from request to completion

✓

Record every handoff, approval, system and waiting point

✓

Ask staff where they work around the system

How many people and departments touch a single case in your most common process?

Keep in mind: The documented procedure shows how the process should run. The cases show how it really runs.

STEP 2

Apply Process Simplification to Each Step

Challenge every step with three questions. Does it change the outcome for the client or the bank? Does a rule or a regulator require it? Would anyone notice if it stopped? Remove steps that fail all three, merge duplicate checks and cut approvals that only confirm what a system already validates. In my experience, the cheapest automation is often a step you simply remove instead of automating it.

Much of the effort in large organisations adds no value. BCG found that managers can spend 40 percent of their time writing reports and 30 percent in coordination meetings. They can waste as much as 60 percent of their time on non-value-adding activities. Every report or meeting you remove is one less task to automate.

✓

Test each step for client value, rule or real risk

✓

Merge duplicate checks and approvals into one control

✓

Remove reports and copies that nobody reads

Which approval in your process confirms something a system has already checked?

Keep in mind: Check with Compliance before removing a control. Some steps look redundant but answer a specific regulatory requirement.

STEP 3

Standardise Variants Before You Code

Processes grow variants. One team handles corporate clients differently, another accepts an extra document and a third keeps a manual list. Each variant needs its own rules once automated. Agree one standard flow with the process owners, keep only the variants that a client segment or regulation truly requires and retire the rest.

Redesign can deliver more than automation alone. Ford employed more than 500 people in accounts payable in North America, while Mazda's department consisted of a total of 5 people. After redesigning the process around the purchase order instead of the invoice, Ford achieved a 75% reduction in head count.

✓

List every variant of the process and its reason

✓

Keep variants that clients or regulation truly require

✓

Agree one standard flow with all process owners

How many variants of your process exist only because teams never agreed on one?

Keep in mind: Every variant you automate adds testing and maintenance. Retire variants before IT builds them.

STEP 4

Fix the Inputs at the Source

Many manual steps exist only to repair poor inputs: missing fields, free-text addresses or wrong product codes. Automation cannot repair what it cannot read. Define the data each step needs, make it mandatory where the case starts and reject incomplete requests early instead of fixing them downstream.

Payments show how slowly inputs change. Swift reported in March 2026 that approximately 65% of payment messages still contain unstructured addresses, months before their planned removal. Swift then agreed to extend the timeline because progress remained uneven. My article on the data quality framework explains how to set rules and an owner for each field.

✓

Define the data each step needs to proceed

✓

Make required fields mandatory where the case starts

✓

Reject incomplete requests early instead of repairing them

Which manual step in your process exists only to repair incomplete or unclear inputs?

Keep in mind: A robot that repairs bad data hides the problem. Fix the form, the channel or the upstream system.

STEP 5

Automate What Remains Stable

Only now choose what to automate. Good candidates are steps with high volume, clear rules, structured inputs and few exceptions. Steps that still change every quarter or depend on judgement should wait. Automate the standard flow first and route exceptions to people, with a named owner for each route.

Scaling automation is harder than starting it. In a Deloitte survey of more than 400 executives, only 3% had managed to scale RPA to 50 or more robots. Respondents named process standardisation as the top challenge, which is the work that simplification does before any robot is built.

✓

Automate steps with high volume and clear rules

✓

Leave judgement and frequently changing steps to people

✓

Give every exception route a named owner

Which step in your simplified process has stable rules and enough volume to automate?

Keep in mind: Automating an unstable process creates rework in IT. Wait until the simplified process has run unchanged for a while.

STEP 6

Measure the Result Before Scaling

Set a baseline before you change anything: cases per day, handling time, waiting time, rework rate and errors. Measure again after simplification and again after automation. This shows how much each stage delivered and stops automation from taking credit for gains that came from removing steps.

New technology faces the same test. Gartner predicted that at least 30% of generative AI projects will be abandoned after proof of concept by the end of 2025, due to poor data quality, inadequate risk controls, escalating costs or unclear business value. My article on how to automate reports in Excel shows a small way to prove value first.

✓

Record a baseline of volume, time, rework and errors

✓

Measure separately after simplification and after automation

✓

Scale only what shows a clear and stable gain

How much of the gain from your last automation came from removing steps first?

Keep in mind: Report simplification savings separately. Otherwise the business case for automation looks better than it is.

As Work Gets More Complex, 6 Rules to Simplify | Yves Morieux (Senior Advisor @ Boston Consulting Group)

SAMPLE CASE

Lloyds Banking Group

Lloyds Banking Group launched its Simplification programme in June 2011. The bank planned an end-to-end redesign of its processes, with automation as one part of that work. It aimed to deliver £1.5 billion of annual savings in 2014. It also planned to cut its suppliers from around 17,000 to under 10,000.

Simplification came before and alongside automation. By the end of 2013 the bank had transferred 1.7 million mortgage accounts to a single mortgage platform. It had fully automated the collection of maturity instructions from term deposit customers, reducing completion time by approximately 85 per cent.

When the first phase ended, the finance director reported annual run rate savings of £2 billion, ahead of the original target. The chief executive described the group's cost to income ratio of 51% as the lowest among its major UK banking peers.

Key takeaway: Lloyds redesigned processes, products and suppliers in one programme and automated within it. Savings came from removing complexity first, so automation had a smaller and more stable process to support.

"Our aim is to become the best bank for customers."

António Horta-Osório (Group Chief Executive @ Lloyds Banking Group)

In Short

Simplifying the process before you automate rests on six habits: map the process as it runs, remove steps that add no value, standardise variants, fix inputs at the source, automate only what remains stable and measure each stage separately.

The result is a shorter process that is cheaper to run, automation that is easier to build and maintain and savings that the business can trace to their real source.

Frequently Asked Questions

What is process simplification?

Process simplification is the removal of steps, handoffs, approvals and variants that do not add value for the client or the bank. It shortens the path from request to outcome without changing what the process delivers. In banks it usually covers duplicate checks, manual reports, redundant approvals and data repairs. A simpler process is cheaper to run and easier to automate later.

Why should you simplify a process before automating it?

Automation copies the process it is given. If the process contains unnecessary steps, duplicate checks or many variants, the automated version contains them too. Changing them later means changing code instead of a procedure. Simplifying first reduces the number of rules to build, test and maintain. It also shows which savings come from redesign and which from technology.

How does process simplification differ from business process reengineering?

Process simplification works within the existing process and removes what adds no value, step by step. Business process reengineering, a term made popular by Michael Hammer, starts from a blank sheet and redesigns the process around its outcome. Simplification is faster and carries less risk. Reengineering can deliver larger gains but needs strong sponsorship and more change for staff.