Which business tasks should you automate first? A scoring method
A practical way to choose your first automation: score each task on value and effort, check the risks, and follow a four-week plan to put it live.
Automate first the task that happens often, follows clear rules, uses information you already have in digital form and causes little harm if it occasionally goes wrong. For most businesses that means reminders, follow-ups and moving enquiry details from one system to another. Score each candidate on value and effort, start with the one that scores highest on value and lowest on effort, and measure the time saved before moving on to the next.
- Choose your first automation by value and effort, not by what a software vendor is promoting.
- Good first candidates are frequent, rule-based and low-risk: payment reminders, appointment reminders and getting enquiries into your CRM.
- Measure before you build. Count how often a task happens and how long it takes, so you can prove the saving afterwards.
- Keep a person in the loop for anything involving money, contracts or sensitive customer matters.
Why the first automation matters
Most teams spend a large share of their week on coordination rather than the work they were hired to do. Asana's 2021 Anatomy of Work Index, a survey of more than 13,000 knowledge workers in seven countries, found that around 60% of working time went on "work about work": chasing updates, searching for information, switching between applications and repeating routine tasks.
The technology to reduce this already exists. McKinsey estimated in 2023 that current AI and other technologies could, in principle, automate activities that take up 60% to 70% of employees' time. That is technical potential, not a forecast of what any single business will achieve, and the gap between the two is usually down to choosing the wrong starting point.
A first automation that works builds confidence and frees time for the next one. A first automation that stalls, because the process was unclear or the data was scattered, often ends the effort altogether. That is why the choice deserves a few hours of thought.
Step 1: list and measure your repetitive tasks
Spend one week noting every task that you or your team repeat. For each one, record:
- How often it happens (per day, week or month)
- How long it takes each time
- Which systems it touches (email, calendar, accounting software, spreadsheets, CRM)
- What goes wrong when it is late or done badly
Multiply frequency by time to get hours per month. This is your baseline, and it is the figure every automation will be judged against.
Step 2: score each task on value and effort
Give each task two scores from 1 to 5.
Value reflects how much the automation would be worth:
| Score | Value means |
|---|---|
| 5 | Saves many hours a month, or directly affects revenue, cash flow or customer experience |
| 3 | Saves a few hours a month, with some benefit to customers or staff |
| 1 | Rare, quick, or makes little difference when done late |
Effort reflects how hard the automation would be to build and run reliably:
| Score | Effort means |
|---|---|
| 1 | Clear rules, one or two systems, data already digital and tidy |
| 3 | Some exceptions to handle, three or more systems, some data cleaning needed |
| 5 | Many exceptions, judgement required, data on paper or spread across people's heads |
Step 3: plot the results and pick from the top-left corner
The chart below shows the scores from a worked example for a small service business with a dozen staff. Tasks in the shaded corner give the most value for the least effort.
Chart
Where common tasks fall on value and effort
- 1Payment reminders
- 2Appointment reminders
- 3Enquiries into your CRM
- 4Answering common questions
- 5Weekly sales report
- 6Bookkeeping matching
- 7Quoting complex jobs
- 8Supplier negotiations
Show the data as a table
| Task | Effort to automate (1 to 5) | Value (1 to 5) |
|---|---|---|
| Payment reminders | 1.6 | 4.5 |
| Appointment reminders | 1.3 | 3.9 |
| Enquiries into your CRM | 2.4 | 4.0 |
| Answering common questions | 2.9 | 4.6 |
| Weekly sales report | 2.1 | 2.4 |
| Bookkeeping matching | 3.7 | 3.3 |
| Quoting complex jobs | 4.4 | 3.9 |
| Supplier negotiations | 4.5 | 1.5 |
| Task | Hours a month today | Value | Effort | Verdict |
|---|---|---|---|---|
| Payment reminders | 6 | 4.5 | 1.6 | Automate first |
| Answering common questions | 20 | 4.6 | 2.9 | Automate next |
| Enquiries into the CRM | 8 | 4.0 | 2.4 | Automate next |
| Appointment reminders | 10 | 3.9 | 1.3 | Automate next |
| Weekly sales report | 3 | 2.4 | 2.1 | Later |
| Bookkeeping matching | 12 | 3.3 | 3.7 | Plan carefully |
| Quoting complex jobs | 15 | 3.9 | 4.4 | Assist, do not automate |
| Supplier negotiations | 4 | 1.5 | 4.5 | Keep manual |
Payment reminders win in this example even though they take only six hours a month. They follow clear rules, sit in one system and bring cash in faster, which is worth more than the hours alone suggest. Quoting complex jobs takes more time, but it depends on judgement, so the better approach is to let software prepare a draft that a person checks.
Step 4: run the risk check
Before you build anything, ask three questions of the task you have chosen:
- What is the worst outcome if it goes wrong? A reminder sent a day late is minor. A wrong price sent to a customer, or a payment made to the wrong account, is not.
- Does it involve money, contracts or personal data? If so, keep a person approving the final step, at least until the automation has a long record of working correctly.
- Can you see what it did? Every automation should leave a record: what it sent, to whom and when.
Good first automations for most businesses
- Payment reminders for overdue invoices, using the reminder features built into most accounting software.
- Appointment reminders and confirmations by text or email, which reduce no-shows.
- Enquiry capture: details from website forms, emails and missed calls placed automatically into your CRM or a shared spreadsheet, with an alert to the right person.
- Instant first replies to new enquiries. Speed matters: a 2025 study of more than 1,300 US law firms found that the median response to an online enquiry took 13 minutes and 26% of firms never replied at all.
- Answering common questions through a website chatbot or an AI receptionist, with a hand-over to a person for anything complicated.
- Review requests sent automatically a day after a job is completed.
What not to automate yet
- Processes nobody can describe. If two members of staff do the task differently, agree the method first.
- Rare tasks. Something done four times a year rarely repays the effort of building and maintaining an automation.
- High-stakes judgement. Hiring decisions, complex pricing and sensitive customer conversations need a person, although software can prepare the groundwork.
- Tasks built on messy data. If customer records are duplicated or incomplete, clean them first, or the automation will spread the errors.
A four-week plan for your first automation
| Week | What to do |
|---|---|
| 1 | Write the process down step by step, including exceptions. Record the baseline: how often, how long, how many errors. |
| 2 | Build the automation and test it with real but harmless examples, such as messages sent to your own address. |
| 3 | Run it alongside the manual process, with a person checking every output before it goes out. |
| 4 | Switch it on fully, keep a weekly check of the log and compare results with the baseline. |
After four weeks you will have a measured saving and a working method, which makes the second automation quicker and easier to justify.
What we would do
We would begin with one week of measurement, score the candidates with the method above and start with the highest-value, lowest-effort task, which for most service businesses is payment or appointment reminders. The second step is usually getting every enquiry, whether it arrives by form, email or phone, into one place with an instant first reply. Calls are often the largest source of lost enquiries, and our guide to the cost of missed calls shows how to put a figure on them.
Next Step
Find out what to automate in your business
We map the process, build the automation and measure the time it saves.
Frequently Asked Questions
What is the easiest thing to automate in a business?
Reminders are usually the easiest: appointment reminders, payment reminders and follow-up messages. They follow clear rules, run often, use data you already hold and are built into many booking and accounting systems, so they can often be switched on in an afternoon.
How much time can automation save a small business?
It depends entirely on the task. The reliable way to know is to measure first: count how often the task happens and how long it takes each time. A task done 60 times a month at six minutes each is six hours a month, and that is the figure an automation should be judged against.
Should I automate a process that does not work well?
No. Automating a confused process produces confusion faster. Write the process down, remove unnecessary steps and agree how exceptions are handled, then automate the improved version.
Do I need AI for automation, or are simple rules enough?
Many valuable automations need no AI at all, only rules such as 'when an invoice is seven days overdue, send this message'. AI becomes useful when the input is unstructured, for example reading emails, answering questions in natural language or extracting details from documents.
Next Action
Want help implementing this in your business?
Tamkoz builds AI phone receptionists, bespoke chatbots, and business automations. All work is quoted in writing before starting.