Map a task step by step and remove the steps that are only there out of habit.
Mei Ling works the front desk of a tuition centre in Tampines, and you first met her in lesson 1.1. When she finished scoring her enquiry task there, she wanted to give all of it to an automation tool straight away, which is what most people want to do. Slow down and look at the steps first. An automation repeats a process exactly as you give it, and it has no way of asking why any step is there.
A parent fills in the enquiry form on the centre's website, and a notification email arrives. Mei Ling opens it, then types the parent's name and number into the enquiry sheet. She types the same two details into the class register as well, "just in case". Next she looks up the timetable and replies to the parent. Finally she puts the details on a sticky note for whichever teacher will phone the parent back. All of this takes her about six minutes per enquiry.
Suppose one of those steps serves no purpose. Once automated, it runs on every single enquiry, and because it now happens where nobody watches, it is harder to spot. Each step in an automation is also a point where the flow can fail. Consider writing to the class register. The flow would need a connection to the register file, the right columns mapped, and the data in a format the file will take. If a colleague changes the register's columns, the flow breaks. When a messy process is automated, it runs more quickly and has more places to go wrong. Removing a step before you build takes away one of those failure points permanently.
Finding the pointless steps is easy but a bit dull, and most people skip it for that reason. Begin by listing the task in order, numbered. Each line should be a verb-and-object step, meaning an action plus the thing it acts on, written briefly and literally: "check timetable", "send reply", "copy name to enquiry sheet", "write sticky note". When a step needs more than a few words to describe, you have probably found two steps joined together, so separate them.
Mei Ling ended up with seven:
Step 1, open notification email Step 2, copy name and number to enquiry sheet Step 3, copy name and number to class register Step 4, check timetable for the child's level Step 5, write reply to parent Step 6, write sticky note for teacher Step 7, hand sticky note to teacher
Put three questions to every line: who uses the result of the step, what would go wrong if it disappeared tomorrow, and whether the same information is already kept somewhere else.
Step 3 fails at once. Only enrolled students belong in the class register, and most parents who enquire never enrol. Mei Ling had kept up the "just in case" copy out of habit from the days before the enquiry sheet existed. Steps 6 and 7 fail as well, because the teacher could look at the enquiry sheet directly if it had a column naming who should call the parent back. With steps 3, 6 and 7 gone, four steps remain, and the first of them, opening the email, also disappears once the form sends its entries straight to the sheet.
Almost every office carries three kinds of waste, and the questions tend to turn them up.
Duplicate data entry is a name, number or amount keyed into two or three places. It usually builds up because each place has a different owner, or was started at a different time. A single record that everyone reads works better than three copies that slowly stop matching.
Unused fields are the things a form or sheet collects that nobody ever looks at. Mei Ling's enquiry form asked for the parent's occupation, and no one had read that answer in two years. Each extra field is something more to map, check and protect, and when the field holds personal data, keeping it safe is also your responsibility.
Approval steps nobody reads are sign-offs and copy-ins that are always given or never acted on, such as a manager copied on every reply "for visibility". An approver who has never once said no is not checking anything. Keep the approvals that genuinely catch mistakes. Lesson 1.3, "Know which work should stay manual", covers where a human check matters, and module 7 shows you how to build a human check that people actually use.
Be careful with steps that seem pointless, since some exist for a reason no one wrote down, such as an audit, a legal record or a report a colleague puts together each month. If a step involves another person, ask them what they use it for before you take it out. A two-minute conversation now saves an awkward one later.
Mei Ling's routine now has four parts: she copies the details into the enquiry sheet, checks the timetable, sends the reply and ticks a column for the teacher. When she timed it, it took about three minutes rather than six. She handles fifteen enquiries a week (the example figure from lesson 1.1), so 15 x 3 = 45 minutes come back to her every week, and she has not used any software to get them.
A shorter manual routine is often the entire fix, and this happens a lot. If what remains takes only a few minutes each week, choosing not to automate at all is a good result. If the task still deserves automating, a tidier list gives you fewer connections to set up, fewer fields to map and fewer points where the flow can fail.
For the activity below, go back to the two tasks you circled at the end of lesson 1.1 and choose the one you are more likely to build. Write out every step as a numbered list, giving each a verb and an object, and break apart any step that needs more than a few words. Put the three questions to each step. Pay particular attention to duplicate data entry, unused fields and approval steps nobody reads, and talk to anyone else involved before dropping a step that affects them. Cross out each step you could remove with no software at all. Then time or judge the routine that remains and decide whether it is still worth automating. Deciding not to automate is an acceptable answer. Hold on to the crossed-out list, because your diagram in module 2 starts from it.
Write out the steps of one of your top two tasks. Then cross out every step you could drop without using any software.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).