A 1-hour framework for finding every manual task worth automating — and which three to fix first.
Pulling last week's numbers into a slide. Building UTMs by hand. Re-typing the same lead-routing logic into a new form. None of these tasks feel big on their own — which is exactly why they never get fixed. They quietly eat hours every week and never show up as a line item anyone questions.
The first step toward fixing this isn't a new tool. It's an honest list of what you're actually doing by hand, every week, that follows the same pattern each time.
For one week, keep a running list of any task you do more than once that follows the exact same steps. That list is your automation backlog — most people are surprised how long it gets.
Track time spent, not just frequency. A 5-minute daily task costs more over a quarter than a 2-hour monthly one — prioritize the audit by total hours, not how annoying the task feels.
Block one hour. List every recurring marketing operations task: reporting pulls, UTM creation, list uploads, form QA, lead routing, CRM hygiene, campaign QA checklists. For each one, note the frequency, who does it, and how long it takes. You'll likely find 15-25 tasks. Most teams have never written this list down in one place before.
Ask the person who actually does the task to describe it out loud, step by step, as if training a new hire. The gaps and exceptions they mention are exactly what an automation needs to account for.
Not every task on your list should be automated first — or at all. Score each one on three axes: frequency (how often it happens), complexity (how many decision points it has), and risk (what happens if it's wrong). High frequency, low complexity, low risk tasks — like formatting a weekly report or building UTM strings — are your best first candidates. High risk tasks involving customer data, spend decisions, or anything irreversible should keep a human in the loop, automated or not.
Start with a task where being wrong costs you five minutes to fix, not five thousand dollars. Build confidence on low-stakes automations before touching anything customer-facing.
AI is currently strongest at synthesis and pattern-matching tasks (summarizing, classifying, drafting) and weakest at tasks requiring real-time system state or irreversible action. Match the task type to the tool, not just the workflow.
Pick the three highest-scoring tasks from your matrix and build only those first. A weekly reporting pull that used to take 90 minutes and now takes 10. A UTM builder that enforces your taxonomy instead of relying on memory. A form QA check that flags broken fields before launch instead of after. Small, proven wins build the case — and the muscle — for automating more.
Document each automation as you build it: trigger, inputs, logic, owner, and what to check if it breaks. Six months from now, nobody will remember why it works unless it's written down.
You don't need a giant automation initiative to get the time back. You need an honest list, an honest scoring system, and the discipline to start with the three tasks that are safest to get wrong while you learn. The manual-work tax is real — and it's the easiest cost in marketing to cut once you can see it.
If you're auditing your team's manual workload and want to talk through what to automate first, reach out — I'd love to compare notes.
hello@kerrycokelekoglu.com kerrycokelekoglu.com