Finding the right tasks to automate

0:00 Finding the right tasks to automate

Think about your last working week. How many hours went on copying enquiries into a spreadsheet, sending the same reply for the twentieth time, or formatting the Monday report? For most small businesses, creators and agencies it's a lot, and it's invisible because it's spread across the week in five-minute chunks. In this lesson you'll learn how to find the work worth automating, score it honestly, and choose three candidates you can actually build, starting this week.

0:33 Analogy: the leaky bucket

Here's an analogy. Your working week is like a leaky bucket. The big holes are obvious, and you already patch them. The real loss comes from dozens of tiny holes: two minutes here copying a lead, five minutes there reformatting a report. Each seems too small to matter. Together they drain hours. Automation is about finding and plugging those small, repeated leaks first.

1:00 Automate the boring, not the important

The best automation projects in small businesses are rarely glamorous. They remove small, repeated chores: copying data between apps, routine messages, formatting reports, sorting inboxes. Once those are handled, AI can take on more interesting work like drafting, summarising and classifying, with the right level of human oversight. Automate the boring, not the important. Keep people on the work that needs judgement, empathy and relationships.

1:28 The four-question filter

Score every candidate task from one to five on four questions. Frequency: how often does it happen? Daily scores high, yearly scores low. Repetitiveness: are the steps roughly the same each time? Clarity: could you write the rules or give examples on one page? And risk, scored in reverse: if the automation gets it wrong, how bad is it? Low harm scores high. Then estimate the time spent each week. High scores plus meaningful time equals a strong candidate.

2:02 Good and poor candidates

Typical strong candidates look like this. In marketing: repurposing long-form content into draft posts, summarising comments and reviews, and weekly performance summaries. In sales: capturing form and ad leads into the CRM and notifying the right person, drafting follow-ups from call notes, and reminders for stale deals. For creators and agencies: sorting inbound DMs and emails, first-draft briefs from intake forms, invoice reminders, and logging sponsored deliverables and disclosure checks. Poor candidates for now: rare one-offs, anything needing negotiation or empathy, high-consequence decisions that can't be caught before they land, and processes nobody has defined.

2:43 The one-week audit

Now run a one-week audit. Log every repeated task: what it was, how long it took, which tools, and who did it. At the end of the week, group similar tasks, score each group with the four questions, and estimate weekly time. Pick your top three with the best mix of time saved and low risk. For each, write the current steps as a simple numbered list. That list becomes your automation blueprint. And if the audit shows three people doing the same task three different ways, standardise first, because automation amplifies whatever you feed it.

3:25 Simple example: a photographer's enquiries

A simple example first. A freelance photographer notices she copies every booking enquiry from her website form into a spreadsheet, then sends the same pricing PDF by email. It happens about fifteen times a week, takes four minutes each, never varies, and a mistake just means resending an email. That scores top marks on all four questions, about an hour a week, and it's a perfect first automation.

3:55 Worked example: Dubai influencer agency

Here's a real-world shaped example. A four-person influencer agency in Dubai audits a week. Copying brand enquiries from email and Instagram into the CRM takes about four hours and scores top marks: automate it. Drafting creator briefs from intake forms and weekly client reports each take several hours and score well: automate the draft, with human review. Checking sponsored posts for disclosure is frequent and clear but risky: an AI-assisted checklist with human sign-off. Negotiating creator fees stays human. They start with enquiry capture, then move on to briefs and reports.

4:35 Business example (illustrative)

Illustrative numbers for the Dubai agency. Enquiry capture took about four hours a week; after automation it took about twenty minutes of checking. Briefs and reports together took about eight hours; with AI drafts and review, about three. That's roughly nine hours a week back across a four-person team, more than a full working day, which they spent on creator relationships and new-business calls.

5:03 Hands-on in the lesson

In the hands-on section you'll set up a shared task-audit sheet with the exact columns to log, formulas that total minutes per task group and blend your scores with monthly hours into a priority score, and a prompt that turns your top three task groups into step-by-step blueprints, marking each step as automate, AI step, human check or keep human, and asking questions instead of inventing steps. Review those blueprints with the people who actually do the work.

5:37 Common mistakes

Common mistakes when choosing what to automate. Starting with the most exciting idea rather than the most frequent task. Only asking the business owner, not the people who do the work. Automating a process before agreeing how it should be done. Forgetting that every automation needs maintenance time. And not naming an owner, so when it breaks, nobody notices for weeks.

6:04 How you'll know you chose well

How will you know you chose well? After a month, the automation runs most days without anyone thinking about it. The person who used to do the task confirms they've got time back. Errors are rare and caught quickly. And you have a measured before-and-after number you could show a partner or client with confidence.

6:28 Watch me do it: audit → blueprints

Watch me do it. I open the audit sheet. The log tab has date, person, task, minutes, tools, trigger, output and notes. After a week, it has about a hundred and forty rows. On the summary tab, I list each task group once and use sum if to total minutes per group from the log. Then hours per week divides by sixty. The team scores frequency, repetitiveness, clarity and low risk from one to five. The priority formula averages the four scores and multiplies by the log of monthly hours, so big tasks aren't drowned out by tiny perfect ones. The verdict formula says keep human when low risk scores two or less. I sort by priority: enquiry capture first, briefs second, reports third. Next, I paste those three groups into the blueprint prompt. It returns numbered steps labelled automate, AI step, human check or keep human, and asks me two questions I answer with the team.

7:36 Recap

To recap: prioritise tasks that are frequent, repetitive, clear and low risk, with meaningful time behind them. Run a one-week audit, score before choosing tools, keep negotiation and high-stakes judgement human, standardise messy processes first, and give every automation an owner. Watch out for starting with the most exciting idea instead of the most valuable, and for underestimating maintenance. Your next step: log your repeated tasks for one week and write blueprints for your top three. Next, the building blocks of no-code automation.

8:12 Try this now (10 minutes today, 30 on day five)

Try this now. Create the task-audit sheet with the columns from the lesson and share it with your team today. Ask everyone to log repeated tasks for the next five working days, even tiny ones. On day five, spend thirty minutes grouping and scoring them, and pick your top three. You'll almost certainly find at least one task that's been quietly eating hours.