Processes
Map the process before you automate it
It is a familiar scene: the company buys an automation tool, connects half a dozen systems and, three months later, has the same problems as before, only now they happen faster and in more places. The tool did not fail. It did exactly what the process told it to do. The problem is that nobody had looked at the process.
Automation amplifies what is already there
Automating means taking a sequence of steps and having a machine repeat it. If the sequence is good, you gain time and consistency. If the sequence has an unnecessary step, a rule each person reads differently or data that goes in wrong, automation repeats all of that flawlessly.
Picture a distributor where orders arrive by WhatsApp, get typed into a spreadsheet, retyped into the billing system and checked by hand by a third person because "there is always a mistake". Automating the retyping looks like the obvious move. But the useful question comes first: why does the order need to go through three places? In many cases the answer is "because it has always been that way", and half the steps can simply disappear.
Michael Hammer, one of the most important names in process management, described this in "The Process Audit", published in Harvard Business Review in 2007. According to him, the performance of a process depends on five things: its design, the people who perform it, an owner, the infrastructure (systems and information) and the metrics. Technology is only one of the five. Without a clear design and an owner, the infrastructure does not hold up.
Signs the process is not ready to be automated
- Two people describe the same process in different ways.
- There is a side spreadsheet "just to be safe".
- Nobody can say how long the process takes from start to finish.
- When something goes wrong, the fix is to check again, not to change what caused the error.
- The process stops when one specific person goes on vacation.
If you recognized three or more, it is worth mapping before you buy or connect any tool.
How to map a process in 5 steps
Mapping does not need to become a months-long project or an 80-page document. For a mid-sized process, one week of well-done work is enough.
1. Pick one process and define where it starts and ends
Start with a single process, preferably the one that generates the most rework or complaints. Name it after its result: "from order to invoice", "from first contact to proposal", "from hiring to day one". Defining the start and the end keeps the mapping from turning into a discussion about the whole company.
2. Talk to the people who do the work, not only to those who manage it
The manager knows the process as it should be. The people doing it know it as it is. Sit next to them, ask them to do the work and write down every step, including the shortcuts and workarounds. Those are what show where the official process does not work.
3. Draw the real path, with time and owner
For each step, record four things: who does it, what comes in, what goes out and how long it takes (working time and waiting time). It can be on paper, on a whiteboard or in a flowchart editor. The format matters less than the honesty of the drawing.
4. Cut, merge and simplify before automating
With the drawing in hand, ask three questions about each step:
- If this step disappeared, would anyone miss it? If not, cut it.
- Is this data being typed for the second time? Then it should come from where it was typed first.
- Does this check exist because the error is born in another step? Then fix the source.
It is common for a process to lose several steps in this phase alone, with no new technology. How many depends on each company's starting point.
5. Define the owner and the metric
Every process needs one person who answers for it end to end and one number that says whether it is healthy: total time, error rate, volume per week. Without an owner, the process drifts back into disorder. Without a metric, nobody notices.
The best automation is the one applied to the step that was left after you cut the ones that did not need to exist.
What to automate first
After the mapping, the choice gets easier. Prioritize steps that combine three traits: they happen often, they follow a clear rule and they require no judgment. Typical examples are moving data from one system to another, sending a standard document, reminding someone of a deadline and generating a report.
The McKinsey Global Institute, in its report "A Future That Works" (2017), estimated that about 60% of occupations have at least 30% of activities that could be automated with existing technology, and that fewer than 5% of occupations could be automated entirely. The practical reading is that almost every role has a repetitive part that can be taken off people's hands, and almost no role disappears. Automating well means giving the team time back for the work that needs judgment.
If your team spends a good part of the day looking for information and chasing status updates, it is also worth reading the article on work about work.
How long this takes
It depends on the size of the process and how many areas it crosses. What we recommend is not waiting for the perfect map of the entire company. Map one process, simplify it, automate the first step and measure. What you learn in that first cycle makes the second one much faster.
Well-organized processes and well-built automation reduce costs and improve margins, but how much varies from one operation to another. That is why, at delpra™, every project starts with an assessment and with mapping, before any tool. See how it works in how we work or explore our services.
Key takeaways
- Automation repeats the process that exists, good or bad.
- Map it with the people who do the work and draw the real path.
- Cut and simplify before connecting any tool.
- Give every process an owner and a metric.
- Start with what is frequent, follows a clear rule and needs no judgment.
Written by the delpra™ teamProcesses, automation and AI so operations scale efficiently.