Quick answer: Automate the task that happens most often, follows the same steps every time, and costs you the most when it goes wrong. Surveyed workers say email, data collection and data entry eat the most repetitive time, so that’s the place to look first. Start with a single workflow, measure what it saves, then expand. The five steps below show you how to find that first task in about a week.
Busywork rarely announces itself as one big problem. It arrives a few minutes at a time: copying a form entry into a spreadsheet, sending the same follow-up email for the fortieth time with only the name changed, rebuilding the weekly report. No single task seems worth fixing, so none of them ever get fixed.
The total is larger than it looks. In a 2017 Smartsheet survey of information workers, more than 40% of workers said they spend at least a quarter of their work week on manual, repetitive tasks, and email, data collection and data entry took up the most time. Nearly 60% believed they could save six or more hours a week if the repetitive parts of their jobs were automated. Six hours is almost a full workday, every week, spent on work a system could do. smartsheet
Busywork eats a real share of the week
Share of workers surveyed
More than 40% of workers surveyed spend at least a quarter of their work week on manual, repetitive tasks, led by email, data collection and data entry. Nearly 60% estimate they could save six or more hours a week if those tasks were automated. Source: Smartsheet, Automation in the Workplace survey
The pattern hasn’t faded. In Asana’s 2023 Anatomy of Work Index, a survey of 9,615 knowledge workers across six countries, people said 58% of their day went to “work about work”: coordinating tasks rather than doing the skilled, strategic work they were hired for. And in a January 2025 Ipsos survey of 1,375 US employees for Eagle Hill Consulting, 68% said they regularly spend time on low-value, inefficient tasks.
The hard part isn’t the technology. It’s choosing what to automate first. Choose well, and the first workflow pays for the next. Choose badly, and you spend a month automating something that saves four minutes a day.
Step 1: List your busywork for one week
Don’t start with tools. Start with a notepad. For five working days, write down every task you repeat. Include the small ones, because small and frequent is where the hours hide.
Look for three signals:
You copy information from one place and paste it into another.
You write a message you’ve written before.
You check something on a schedule just to confirm nothing changed.
By Friday, you’ll have a list that’s longer than you expected. That list is your inventory.
Step 2: Score every task on frequency, time and risk
Give each task three scores from 1 to 5:
Frequency: how often it happens.
Time: how long it takes each time.
Risk: how much a mistake costs, whether that’s a missed lead, a wrong invoice or an annoyed client.
Multiply the three. A task that happens daily, takes twenty minutes and loses money when it’s done wrong scores high. A task that takes an hour but only happens once a quarter scores low, however painful it feels.
For example: appointment reminders happen daily (5), take five minutes each (2), and a missed one costs a booking (4), for a score of 40. A quarterly report takes an afternoon (5) but happens four times a year (1), and a mistake is usually caught before it matters (2), for a score of 10. The reminders win.
Then add one tie-breaker: how stable is the process? A task that runs the same way every week beats one that keeps changing, even if the changing one takes longer. Every change to the process means rebuilding the automation.
Risk is the score most owners skip, and it often matters most. Automation saves time, but its quieter advantage is consistency. Workers in the Smartsheet survey listed eliminating human error almost as often as reducing wasted time when describing what they wanted from automation. smartsheet
Prefer to let a machine do the arithmetic? Lara runs this method with you in under two minutes. Give her your three most repetitive tasks, and she'll rank them and email you the full breakdown. It is, fittingly, an automation.
| Task | How often | How long | Cost when it slips |
|---|
Step 3: Pick a task with clear rules
Sort your top scorers into two groups.
Rules-based tasks follow the same path every time. When a form comes in, create a contact, send a confirmation and notify the team. These are ideal first automations.
Judgment-based tasks change with context: pricing a custom job, handling a complaint, deciding which client to prioritize. These stay human for now. AI can draft or suggest, but a person should decide.
Your first automation should come from the rules-based group. If you can’t write the steps on an index card, it isn’t ready to automate.
Before you automate anything, ask whether the task should exist at all. A report nobody reads doesn’t need automating. It needs deleting. And if the process is broken, fix it first, or the automation will just repeat the mistakes faster.
Step 4: Clean the data, then build one workflow
Before you build anything, look at the data the task depends on. An automation doesn’t improve your records. It obeys them, so whatever is wrong with your data will now go wrong automatically.
Take the department you’re automating and answer four questions:
Where does the data live? Write down every place it’s kept: the CRM, a spreadsheet, someone’s inbox, a notebook by the phone. If the answer is “a bit of everywhere,” that’s your first problem to solve.
Which place is the source of truth? When a client’s phone number changes, one system should hold the correct version and every other tool should take it from there. Without that, five apps will keep five different versions of the same client.
Is it consistent? The same client entered as “ACME Inc.”, “Acme” and “acme inc” counts as three clients to a machine. Merge the duplicates, agree on how names, dates and statuses are written, and decide which fields are required.
Who owns it? Every field that matters needs a person responsible for keeping it accurate.
Then sketch the flow on paper before you open any tool: form → CRM → owner → invoice. You’ll see where the data enters, where it gets copied, and where it goes missing. Those gaps are where an automation fails.
If the clean-up turns out to be bigger than the automation, that’s not a detour. Getting your data into one clean, structured place is often the most valuable first project, because every workflow after it depends on that foundation.
Once the data is clean, build one workflow with one trigger and one outcome. Resist the urge to automate the whole business at once.
Then choose where to build it. Start with the tools you already pay for: many CRMs, booking systems and accounting apps have built-in automations that handle simple triggers without adding anything new. When the work has to cross between apps, use a no-code automation platform such as Zapier, Make or n8n to connect them. When it involves your own systems or complex logic, it’s time for a custom build or a specialist.
If you’re unsure where to start, our pick for most service businesses is lead response, because the research on speed is hard to ignore. In “The Short Life of Online Sales Leads,” published in Harvard Business Review in 2011, researchers audited 2,241 US companies by submitting a test lead and timing each reply. A companion analysis of about 1.25 million leads found that firms contacting a prospect within an hour were nearly 7 times more likely to have a qualifying conversation with a decision maker than firms that waited one hour longer, and more than 60 times more likely than firms that waited a day. Yet only 37% of the audited companies responded within an hour, and 24% took more than a day. Another 23% never responded at all.
More than a decade later, the gap is still there. When Workato submitted demo requests to 114 B2B companies for a study published in 2026, only one sent a personalized email within five minutes, none called within five minutes, and the average email reply took almost 12 hours.
A simple workflow closes that gap. When a lead submits your form, it gets an instant personal reply, the record goes straight into your CRM, and you get a notification with everything you need to call back. The lead never waits, and you never have to remember.
Before it goes live, try to break it. Submit the form twice. Leave a field blank. Type an email address in capitals. Watch what happens. It’s better to find the gaps yourself than to have a customer find them.
Step 5: Measure it, then expand
Before you switch the workflow on, write down two baseline numbers: how long the task takes per week, and how often it goes wrong. After a month, compare.
If the workflow paid off, go back to your scored list and automate the next task. If it didn’t, find out why before you build anything else. The usual causes are a task with more exceptions than you expected, or a workflow with no clear owner.
Give every workflow a named owner and a date to review it. Automations rarely fail loudly. They drift as your tools and process change.
Each workflow you add makes the next one easier, because the data is already flowing between your tools.
Confirmed vs. Assumed: what automation can really do
40%+
Repetitive work takes a large share of the week. Over 40% of surveyed workers spend at least a quarter of their week on it.
7×
Speed wins leads. Contacting a lead within an hour made firms nearly 7 times more likely to qualify it than waiting one more hour.
Next year
“AI will automate most of your job next year.” McKinsey’s research says something more careful. Its November 2025 report estimates current technology could, in theory, automate activities that take up about 57% of US work hours, but calls that technical potential, not a forecast of job losses, and says adoption may take decades. The chart below shows the gap. The opportunity is real, but it arrives one workflow at a time.
Replacing people
“Automation means replacing people.” The tasks worth automating first are the ones nobody wants to do. The aim is to give owners and their teams back time for work that needs a human.
Possible isn't the same as soon
Share of employees' time that could technically be automated
When about half of today's work could actually be automated
Top: technical potential of current generative AI and other technologies to automate work activities, as a share of employees' time. It is not a forecast of adoption. Bottom: McKinsey's adoption scenarios for when about half of today's work activities could be automated; bar length shows years from today. June 2023. Source: McKinsey, The Economic Potential of Generative AI
Frequently asked questions
What should a small business automate first?
Start with the task that scores highest on frequency, time and cost of mistakes, and that follows the same steps every time. For most service businesses, our pick is lead response: an instant reply, an automatic CRM entry and a notification to follow up.
How do I know if a task can be automated?
If you can write its steps on an index card and they don't change from one time to the next, it's a strong candidate. If each case needs a judgment call, keep a person in charge.
How much time can automation save?
Nearly 60% of workers in Smartsheet's survey estimated they could save six or more hours a week if the repetitive parts of their jobs were automated. Your own savings depend on your task list, which is why Step 5 asks you to measure a baseline first.
Do I need to be technical to automate my business?
No. You need to know your process well enough to describe it. The building can be done with no-code tools or handed to a specialist.
Should I automate everything at once?
No. One workflow, measured and working, beats ten half-built ones. Expand only after the first one proves itself.
