Before you automate a workflow, map it on one page

“Can we automate this?” is one of the most common requests we hear, from founders building operations products and from agencies whose clients are drowning in spreadsheets and WhatsApp messages. The answer is usually yes. The better question is: automate what, exactly?

Automation does not fix a process. It repeats it faster. If the process has unclear owners, missing information, or exceptions nobody wrote down, automation will repeat those problems at scale.

The one-page map

Before any tool is chosen, map the workflow on a single page. For each step write five things:

1. Trigger: what starts this step? A form submission, an email, a time of day, a status change.
2. Owner: who does it today, by role, not by name.
3. Input: what information they need, and where it comes from.
4. Action: what they actually do, in plain words.
5. Output: what changes when they are done, and who is told.

Walk through it with the people who do the work, not only the managers who describe it. The real process is almost always different from the official one.

Find the decisions and the exceptions

Mark every point where someone makes a judgement: approve or reject, which supplier, how urgent. Then ask, for each one, “what happens when this goes wrong?” A missing document. A customer who replies in a different channel. A payment that arrives in two parts.

Exceptions are where most of the work hides. A process that is 80 percent routine can still consume most of the team’s time in the remaining 20 percent.

Decide what to automate first

With the map in front of you, sort steps into three groups:

• Automate now: high volume, clear rules, few exceptions. Copying data between systems, sending standard notifications, generating documents from templates, reminders.
• Assist, do not replace: steps that need judgement but involve a lot of preparation. Software gathers the information and suggests an answer; a person confirms it.
• Leave to people: rare, sensitive, or relationship-driven steps. Automating these costs more than it saves and annoys customers.

Start with one end-to-end slice of the workflow rather than one step across everything. A complete small flow proves value and exposes integration problems early.

Measure before and after

Record a simple baseline before building anything: how long the workflow takes from trigger to outcome, how many items are handled per week, and how often something has to be redone. Without this, nobody can say whether the automation worked, and the next investment is harder to justify.

Plan for when it breaks

Every automated workflow will meet a case it does not understand. Decide in advance what happens then: the item goes to a named person’s queue, with the reason, and nothing is silently dropped. A clear failure path is what makes teams trust automation enough to stop double-checking it by hand.

Bring the page to the first meeting

Whether you build in-house, hire a partner, or use an off-the-shelf tool, the one-page map makes the conversation faster and the estimate more accurate. It is also a useful document on its own: teams often fix half their problems simply by agreeing, on paper, who owns each step.

Once the map is clear, we can automate it: an AI assistant for the questions your team answers every day, or a custom web app for the whole workflow.