Automation-First Approach: What to Automate First

An automation-first approach checks every repeatable task for automation before anyone does it by hand. What to automate first, and what to leave alone.

By Yannis Spanenburg · 2026-06-17 · 6 min read

Every week a business owner asks us about AI agents. Almost every week, the honest answer is that their first automation should be something far more boring, and that the agent conversation is three steps early. An automation-first approach is how you find that boring, profitable first step.

What is an automation-first approach?

An automation-first approach means every repeatable task is checked for automation before anyone is hired or trained to do it by hand. In practice that means documenting the process, simplifying it and automating what survives, so people keep the work that needs judgment.

Key takeaways

• A process is ready to automate when it happens often, follows the same steps every time, has predictable inputs and a correct result you can state in one sentence.
• Start with the first reply to a new enquiry: in a Harvard Business Review study of 1.25 million sales leads, firms that tried to reach a lead within an hour were nearly seven times as likely to qualify it as firms that tried even an hour later.
• Automation magnifies whatever it is pointed at, so document and simplify a process before you automate it, or you will only break it faster.
• Automation takes over tasks inside jobs: McKinsey found fewer than 5 percent of occupations consist of activities that can be fully automated.
• Leave alone the work customers choose you for, decisions where judgment meets real consequence, and anything that changes shape every few weeks.

What should a business automate first?

Automate the first response to a new enquiry first, then intake and qualification, recurring reports and any follow-up that depends on someone remembering. These four happen often, carry little risk and have a result you can define in one sentence, which is exactly what makes a process ready to automate.

The first reply to a new enquiry

A new enquiry that waits until tomorrow is competing with whoever answered within the hour. When Harvard Business Review analysed 1.25 million online sales leads at 42 US companies, firms that tried to reach a lead within an hour were nearly seven times as likely to qualify it as firms that tried even an hour later, and more than 60 times as likely as firms that waited a day or more.

An instant acknowledgement, routing to the right person and a booking link are usually the highest return per hour of build in the whole business, and also the least glamorous.

Intake and qualification

Every business asks new customers the same opening questions. Collecting the answers before the first conversation shortens the sales cycle and filters out people who were never going to buy. This is where language models are worth using: turning messy free text into structured fields your team can sort.

Reports nobody enjoys building

If someone rebuilds the same report every month by copying numbers between tabs, that is recoverable time with almost no risk attached. If the automation breaks, a report is late and no customer is harmed.

Follow-up that depends on memory

Quotes nobody chased, carts left behind, customers who bought once and never heard from you again. They fell through because follow-up relied on someone remembering. On one US Shopify store, automated cart recovery has brought back $201K in abandoned carts.

If you already have a cart flow and it underperforms, the usual culprits are timing and early discounts, covered in why your abandoned cart flow is not converting.

How do you know a process is ready to automate?

A process is ready when four things are true: it happens often enough to repay the build, it follows the same steps every time, its inputs are predictable, and you can say in one sentence what a correct result looks like. If any one of those is missing, the process is not ready yet.

The test costs nothing. Write the process down step by step, as if handing it to someone who starts on Monday. If you can do that in one sitting, it is a process and it can probably be automated. If you cannot, it is a habit living in one person's head, and automating it will encode whatever that person happened to do last Tuesday.

How to take an automation-first approach in 6 steps

1. List the ten tasks your team repeats most often, with how long each takes and how much judgment it needs.
2. Write each one down step by step, as if handing it to someone who starts on Monday.
3. Cut every step that only exists to solve a problem you no longer have.
4. Define a correct result in one sentence before you build anything.
5. Automate the version that survived, starting with the task that is most frequent and needs the least judgment.
6. Check the output by hand for the first two weeks, then review the process every quarter, because a process that changes shape stops being safe to automate.

The tool matters less than the order. Most first automations run happily on n8n, Zapier or Make, and our comparison of n8n, Zapier and Make covers which suits which kind of team.

Why does automating a broken process make it worse?

Because automation multiplies whatever it is pointed at. A clear process gets faster and more consistent. A messy one produces the same bad outcome more often, with fewer people watching, and it feels like progress the whole time. That is why the order is document, simplify, automate, then measure.

The idea is older than most of the tools. The benchmarking organisation APQC opens its case for why process automation won't cure inefficiency with a line from Bill Gates: automation applied to an inefficient operation “will magnify the inefficiency.”

We see the same in client work. Most automation projects that fail start as undocumented processes and become expensive once a machine starts repeating them.

Proof of Pixel's position: an automation-first approach follows one order, document, simplify, automate and measure, starting with the enquiry nobody has answered yet.

What should you never automate?

Never automate the work customers choose you for, decisions where judgment meets real consequence, such as pricing exceptions or complaints that have already escalated, or anything that changes shape every few weeks. The first removes your edge, the second removes accountability, and the third costs more to maintain than the manual work.

Automation replaces tasks far more often than whole jobs. The McKinsey Global Institute found that fewer than 5 percent of occupations consist of activities that can be fully automated, while in about 60 percent at least a third of the activities could be. In a small business that usually means people move off chasing, copying and reformatting, and onto the conversations that win work.

Proof of Pixel is a marketing and consulting agency for businesses that live on enquiries and orders, building the website, the tracking and the follow-up as one system, with offices in Dubai, New York, London, Antwerp and Kuala Lumpur. Our automation work starts with exactly this sort of list.

How do you know if an automation is working?

Pick one number before you switch it on and compare it after four weeks: first-response time for enquiries, hours spent on a report, or orders recovered from abandoned carts. If the number did not move, fix the automation or switch it off.

The five minute sort

List the five things your team does most often. Next to each, write how long it takes and how much judgment it needs. Automate the one that is frequent and low judgment, and leave the one that is rare and high judgment exactly where it is. If you want a second pair of eyes on the list, talk to us.