You will produce a ranked shortlist of automation candidates from your real working week, with the numbers behind each choice.
Most people misjudge their own working week. Ask how often a task comes up and you'll get a confident answer, but once every occurrence is written down the figures usually shift. Memory holds on to the jobs that annoy you and loses track of the ones that take time without anyone noticing. If you pick an automation target from memory, you can easily pick the wrong task, so the exercise starts by tracking what you really do instead of guessing.
A typical guess sounds like "maybe five times a week, ten minutes each". In an invented illustration, three days of notes turn the five into eleven and the ten minutes into four, and a task nobody mentioned turns out to take an hour every day.
The log covers any task that moves information from one place to another. Copying an order into a spreadsheet counts, and so do forwarding an email to a colleague, typing receipt amounts into an expense claim, updating a status in a project tool and sending a weekly reminder.
Keep the log somewhere you'll really use during the day: the notes app on your phone, a paper pad beside your keyboard or a sheet pinned in your browser. When you finish a task, jot down what it was and about how long it took. You don't need accuracy to the minute, since a rough tally kept honestly beats a perfect estimate made from memory.
Run the log for three working days. That is long enough to pick up your daily tasks and most of your weekly ones. Monthly tasks are easy to miss in a short window, so look back through your calendar or sent mail and add those separately.
Once the log is done, open a spreadsheet with seven columns: task, frequency per month, minutes per run, monthly minutes, error cost, stability, notes.
Put every frequency in monthly terms so that tasks on different rhythms can be compared fairly. Count a daily task in a five-day week as roughly 22 times a month, and a weekly task as roughly four times a month.
Monthly minutes is frequency per month multiplied by minutes per run. The easiest way to get it is a formula in that column. Keep checking the arithmetic as you fill in rows, because one slip in monthly minutes will quietly reorder the whole list and you may not notice when you sort.
Stability rates how fixed the steps of a task are. Score it high when the steps are identical every time, medium when they change now and then, and low when each run is different from the last.
Error cost is what a mistake in the task costs you, and most people find it the hardest column. There are two honest ways to fill it in. One is a rough dollar figure that you label as your estimate. For instance, a missed enquiry might lose a student whose fees would be several hundred dollars a term. The other is a plain description of what happens, like "wrong client gets the quote, embarrassing, an hour to fix" or "late payment to supplier, relationship strained". All three are invented examples.
Use whichever option you could explain to your manager without hand-waving. A true sentence is more use than a number invented to look precise. With either version, also give the task an error rating of low, medium or high so you can sort the sheet on it.
Arif works at a renovation firm in Ubi. He tracked his work for three days and then estimated his own figures, which are invented for this example. Three of his rows show how the scoring works.
The first row is entering supplier invoices into the accounts sheet. This happens about 40 times a month at 5 minutes each, which is 200 minutes, or about 3.3 hours a month. He rated the error cost high because a wrong amount means paying a supplier the wrong sum. He rated stability high because every invoice has a supplier, a date, an amount and GST.
The second row is sending quotes to clients. That is about 12 a month at 25 minutes each, which comes to 300 minutes, or 5 hours a month. The error cost is high since a wrong price is hard to take back. Stability is low because every job is measured and priced differently.
The third row is booking site visits into the shared calendar: about 30 a month at 3 minutes each, so 90 minutes, or 1.5 hours a month. He gave it high stability and a medium error cost, since a double booking means a wasted trip.
Quotes take the most time of the three, but they have low stability and depend on judgement, so Arif marks them for a human check at most. Supplier invoices go to the top of his list because they combine real hours, a high error cost and fixed steps. Site visits come second, and he keeps them on his list for later.
Sort your sheet by monthly minutes, then read down the error and stability columns. The task you build during the course should meet all of these:
it takes meaningful time its error cost is something you can describe its stability is high or medium it carries the label "automate" or "automate with a human check" from lesson 1.3, "Know which work should stay manual"
If a task takes a lot of time but has low stability and depends on judgement, mark it for a human check at most, as Arif did with his quotes.
Your second choice is the one you'll come back to later. Make it different enough from the first that building it teaches you something new. If your first task is a form task, for example, an email task would make a good second.
Before you commit to the first task, look again at the steps you wrote out for it in lesson 1.2, "Fix the process before you automate it". Check that the task still makes sense after the cuts you made there.
Your sheet is finished when it has at least ten rows, every column filled in, monthly minutes calculated correctly, an error cost you could defend in one sentence and two tasks clearly marked. Once sorted, it should either match your instinct about where your week goes or show clearly why that instinct was wrong.
Once your three days of tracking are done, you have everything you need to fill in your own sheet and make the choice.
Complete the task scoring sheet for at least ten tasks and write two sentences on why your chosen candidate ranks first.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).