How to audit your processes before splashing out on software
Installing software over a messy process only speeds up errors. Here is the pen-and-paper method to review your operations before paying for licences.
In short
- Automating a flawed operational process merely speeds up the frequency at which errors are generated.
- The root cause of operational delays is usually found in the waiting times between tasks, rather than in individual execution speed.
- Before signing up for any software, it is essential to simplify and standardise every step of the manual process.
- Logging unmanaged exceptions for two weeks prevents tech licences from ending up abandoned.
Why technology doesn't fix a broken process
Automating an inefficient process only results in your company making mistakes faster and on a larger scale. For years in car assembly plants and logistics distribution centres, I saw consultants arrive promising that implementing a new system would resolve delivery delays or inventory discrepancies. The outcome was almost always the same: five-figure software invoices and the exact same mess as before, only now it's digital and harder to fix.
When a manual process fails, the cause is rarely a lack of technology. In the vast majority of cases, the problem lies in a lack of clear responsibilities, poorly designed sequences or flawed input data. Software is an amplifier. If you feed in order and standardisation, you get speed and scale. If you feed in ambiguity and unnecessary steps, you get automated chaos.
Before authorising the purchase of a management platform or hiring the development of a bespoke tool, your business needs to carry out a cold-eyed operational process audit. This doesn't require external consultants with executive slide decks or expensive tools; it requires direct observation where the daily work happens, pen, paper and the willingness to ask uncomfortable questions about how things move through your operation.
Step 1: Separate active time from waiting time
The most common mistake when analysing a manual task is only measuring how long it takes an employee to execute it. If you watch an operative packing an order or an accounts assistant logging an invoice, the direct working time is usually short. However, the order takes two days to leave the warehouse and the invoice sits in the system for weeks. The difference between those two values is where the real bottlenecks are hiding.
In supply chain management we divide the total time of a process into two fundamental categories:
- Touch Time: The actual time during which a person is physically handling the product or actively entering information into a record.
- Dwell Time: The time the order, document or product sits immobile in an in-tray, on a shelf or in an email inbox waiting for the next step.
To audit this area in your business, take a sample of ten real orders or requests and record the exact start and end times for each stage. You will be surprised to discover that touch time usually accounts for a tiny fraction of the total cycle time. If a process takes 48 hours but the active work requires just 20 minutes, buying software to cut those 20 minutes down to 10 won't change your customer's experience at all. What your business needs is to remove the points where work gets frozen.
Step 2: Follow the physical flow of information
Don't audit a process from a boardroom or based on a procedures manual that nobody reads. To understand the reality of your operation, you need to carry out what is known in the Toyota production system as a Gemba walk: physically tracing the exact path an order, raw material or document follows from start to finish.
Choose an ordinary day and follow an order from the moment it enters through the sales channel until it is delivered or settled. Watch every handoff. Handoffs between people, departments or systems are the high-risk zones of the operation. At each transfer, ask the team these four direct questions:
- What information do you need to start this step?
- Who do you receive it from and in what exact format does it arrive?
- What do you do if the information arrives incomplete or with errors?
- Who do you hand the result over to and how do you know that person received it correctly?
If the answer to what happens when information arrives wrong is "I message them internally" or "I just fix it myself the best I can", you've just found a massive productivity leak that no software is going to fix on its own.
Step 3: Map invisible exceptions before you start programming
Theoretical workflows always look neat. A customer orders, the warehouse dispatches, accounts invoices. The problem is that businesses don't operate in theoretical conditions. Daily operations are full of exceptions: customers changing delivery addresses at the last minute, products without a registered code, special discounts authorised verbally or suppliers delivering incomplete quantities.
Standard software systems are built to process the regular flow. When an exception enters a rigid system that wasn't configured to handle it, the process grinds to a complete halt and employees end up creating parallel Excel spreadsheets to sort out off-system what the software won't let them do.
Practical rule of thumb: For 14 consecutive days, put up a physical whiteboard in the operations area. Ask the team to note down every single instance where they had to stop normal work to resolve an unforeseen problem. After two weeks, categorise those notes. Those invisible exceptions make up the real requirements catalogue that any operational solution—whether manual or technological—must address.
When it's actually worth investing in tools
There is a strict sequence that prevents wasting capital on operational technology projects. The sequence consists of three consecutive stages that cannot be altered:
1. Eliminate
Question every step of the process. If a sign-off signature provides no real control or if a spreadsheet report is read by no manager to make decisions, cut the task. Automating a useless task is pure waste.
2. Simplify and standardise
Reduce variations. If the dispatch process has four different ways of being carried out depending on the shift or person in charge, define a single standard way. Write the instruction clearly enough for a new person to execute it with a minimal margin for error.
3. Digitise or automate
Only when the process is linear, predictable, contains no unnecessary steps and handles clean data does it make sense to look for tech tools. At this point, automation will fulfil its true purpose: freeing up your staff's time for higher-value tasks and cutting delivery times.
| Manual process status | Recommended decision | Risk of buying software |
|---|---|---|
| Messy with multiple undocumented exceptions | Hold off on buying. Start a field audit. | High. Abandoned licences and team frustration. |
| Standardised but with high waiting time between departments | Redesign the handoff flow and responsibilities. | Medium. Software won't solve poor communication. |
| Stable, repetitive, with high manual touch time | Evaluate automation tools or software. | Low. Clear ROI and quick adoption. |
Auditing operations with the rigour of direct observation takes time and discipline, but it costs a fraction of what a failed software implementation represents. Make sure your business knows exactly how its operations walk before trying to teach them how to run.
Frequently asked questions
How long should a process audit last in a company?
For most key business processes, a practical operational audit shouldn't take more than two to three weeks. The goal isn't to write an extensive document, but to identify friction points, downtime and exceptions through the direct tracking of real cases.
What tools are needed to audit a manual process?
No specialised software is required. The best audits are carried out with direct observations at the workstation, a basic stopwatch to measure active times, a whiteboard to log daily exceptions and hand-drawn flowcharts with the staff who carry out the work.
How do you know if an operational problem is solved with software or with management?
If the problem stems from a lack of clarity on who should do what, variations in working methods between employees, or delays while information awaits approval, it's a management problem. Software only applies when the process is entirely predictable and the bottleneck is typing or processing speed.
Guillermo Campos Ciro
Industrial engineer with ten years in supply chain across automotive and Amazon. I now help businesses fix their processes before automating them.
Tell me about your process