FREE PREVIEW · MODULE 1 · LESSON 1 · 15 MIN
Spotting automatable tasks: frequency × time × rules
A simple scoring method that separates tasks worth automating from tasks that only feel tedious.
Not every annoying task is worth automating. The ones that pay back share three traits: they happen often, they take real time on each occasion, and they follow rules you could explain to a new colleague. A useful way to compare candidates is to score each trait from 1 to 5 and multiply the scores. Frequency × time × rules gives a number between 1 and 125. Multiplying rather than adding matters, because a task that scores 1 on any trait (done once a quarter, done in ten seconds, or decided case by case) should sink to the bottom of the list.
Use fixed anchors so that scores are comparable between tasks and between colleagues. Frequency: 1 means monthly or less, 3 means a few times a week, 5 means several times a day. Time per occurrence: 1 means under a minute, 3 means five to fifteen minutes, 5 means over thirty minutes. Rules: 1 means judgement is needed every time, 3 means rules cover most cases with some exceptions, 5 means you could write all the rules on one page. A supplier invoice that arrives twenty times a week, takes eight minutes to check and key in, and follows clear rules might score 4 × 3 × 4 = 48.
The rules score is where AI changes the picture. Traditional automation needed perfectly structured input: a form field, a fixed column, an exact keyword. AI steps can handle loosely structured input, such as a free-text email, a scanned receipt or a customer comment, and turn it into structured data. That means tasks you once scored 2 on rules, like “read the enquiry and work out which service it is about”, may now score 3 or 4. They do not score 5, because AI output still needs checking, which Module 2 covers in detail.
The common mistakes at this stage are predictable. People pick the task they personally dislike most rather than the one with the highest score. They underestimate frequency because they count only their own share of a team task. And they ignore the cost of errors: a task where a mistake means a wrong invoice paid or a client email sent to the wrong person deserves a lower priority for a first project, however well it scores. Add a simple risk flag (low, medium or high) next to each score, and start with something high-scoring and low-risk.
Example: a dental clinic in Lisbon scores its admin tasks
The practice manager at a four-chair clinic listed nine recurring tasks. Three stood out. Confirming next-day appointments scored 5 × 2 × 5 = 50: it happens daily, takes a few minutes per batch, and the rules never change. Sorting the shared inbox into bookings, billing questions and insurance queries scored 5 × 3 × 3 = 45, a good fit for an AI classification step. Replying to complaints scored 2 × 4 × 1 = 8, because every case needs judgement. She flagged inbox sorting as medium risk, since a misfiled insurance query could delay a payment, and chose appointment confirmations as the first project, with inbox sorting second once the team trusted the approach.
Illustrative example: the business, people and figures are fictional.Key takeaways
- Score frequency, time and rules from 1 to 5 and multiply them
- AI raises the rules score for messy text input, but never to a perfect 5
- Flag error risk separately and start with high-score, low-risk tasks
Exercise
List ten recurring tasks from your last two weeks of work, using your calendar, sent emails and browser history as memory aids. Score each one for frequency, time and rules using the anchors in this lesson, multiply the scores, and add a low, medium or high risk flag. Circle the three highest-scoring tasks that are not high risk.
Continue with 11 more lessons, module quizzes and 3 downloads.
Turn repetitive office tasks into reliable AI-assisted workflows, with checks that keep a person in control. One payment of $99, lifetime access, 14-day refund policy.
See the full course