What to automate first: scoring tasks by time and risk
To decide what to automate first, list the tasks your team repeats, then score each one on two things: how much time it takes and how much damage a mistake could do. Start with tasks that are high on time and low on risk. Leave high-risk tasks until you have built confidence, and think twice about low-time tasks that are not worth the setup.
Time says what is worth doing. Risk says what is safe to do first. Score each task from 1 to 3 on both, and begin with the high-time, low-risk ones. This is a rough sorting tool, not a calculation, and your own judgment should overrule it.
Step 1: What tasks are on the list?
Write down the repeated tasks in your business, not the big projects. Ask each person what they did last week that they have done many times before. Typical answers include copying details between systems, sending the same confirmation, chasing a missing document, entering orders, updating a spreadsheet, and sorting incoming email.
Aim for 10 to 20 items. For each, note who does it, which tools are involved and roughly how often it happens. If a task is not clearly defined yet, write it up first using documenting the process before you automate it.
Step 2: How do you score time?
Time is about the total hours the task consumes, not how long one instance takes. A two-minute job done forty times a week is a bigger target than a one-hour job done once a month. Use rough estimates from the people doing the work, and score:
- 1 (low): a small amount of time overall, or only occasionally.
- 2 (medium): noticeable time each week, but not a main part of anyone's job.
- 3 (high): a regular drain on someone's week.
Include the hidden cost of interruptions. A task that is quick but constantly breaks someone's concentration may deserve a higher score.
Step 3: How do you score risk?
Risk is what happens if the automation gets it wrong, or does the right thing at the wrong moment. Consider money, customer relationships, legal or privacy exposure, and how easy it is to undo. Score:
- 1 (low): a mistake is internal, visible quickly and easy to reverse. For example, tagging a record or filing a document.
- 2 (medium): a mistake is noticeable to a customer or costs some money, but can be fixed. For example, a confirmation email with a wrong detail.
- 3 (high): a mistake is hard to reverse or affects money, contracts, sensitive data or an important relationship. For example, sending a payment or a legal notice.
Also ask how the task handles unusual cases. If it has many exceptions, the risk is higher; see the exception problem.
Step 4: How do you put the scores together?
Draw a simple grid with time on one side and risk on the other, and place each task in it. Then read it in this order:
- High time, low risk. Start here. These give the most relief for the least worry. Examples are copying new enquiries into a tracker or sending internal notifications.
- High time, medium risk. Good second wave, usually with a person approving the output before it goes out. See designing a review queue.
- Low time, low risk. Fine if they are quick to build or come free with something else, but not a priority.
- High risk, any time. Wait, or automate only the preparation and keep a person on the final decision.
- Low time, high risk. Usually leave alone.
Where tasks tie, add two tie-breakers: how clean the data is, and whether the task has a clear owner. A task with tidy inputs and a named owner will go in more smoothly than one with scattered data. Our automation readiness checklist covers both.
What does this look like in practice?
Imagine a small services firm with a list of six tasks. This is an illustration of the method, not a real case or a prediction.
- Copying form enquiries into a tracker: high time, low risk. Do first.
- Sending appointment confirmations: medium time, medium risk. Do second, with a template a person has approved.
- Sending payment reminders: medium time, medium risk, as tone matters. See customer follow-up automation.
- Issuing refunds: low time, high risk. Leave with a person.
- Replying to complex customer complaints: medium time, high risk. Keep with a person, perhaps with automated summaries to help.
- Renaming and filing incoming documents: low time, low risk. Do when convenient.
What mistakes should you avoid?
- Choosing the most exciting task, not the best one. A visible project, such as a chatbot, is not always the best first step. See AI chatbot for business for where that tends to help and fail.
- Scoring on guesswork. Ask the people who do the work. They often know the true time cost.
- Starting too big. Pick one task, and one stage of it if needed, and run it beside the manual process first. See the parallel run.
- Forgetting maintenance. Every automation needs an owner and alerts. See why automations break.
- Not deciding how you will know it worked. Agree on a simple measure before you start; measuring automation results shows how.
Frequently asked questions
How many tasks should I score?
Ten to twenty is enough. More than that and the exercise becomes the project.
What if a task is high time and high risk?
Consider automating only the preparation, such as gathering information or drafting, and keeping a person on the final step.
Should I automate the task I dislike most?
Dislike is a fair signal of a task people avoid or get wrong, but still run it through the scores. A disliked task can also be risky.
Can I do this myself, or do I need outside help?
The scoring is easy to do yourself. Help is more useful for the build and for spotting hidden risks.
Want a second pair of eyes on your list? Book a free audit and we will help you rank the options.
Want this built for your business?
We will look at how your work flows today and tell you which steps are worth automating, and which are not. Free, and you keep the notes.
![]()
You will talk to Fayez or Kareem, the two founders.
