The First Thing We Look at When Automating a Business Workflow

Most people assume that automating a business process starts with choosing a tool. Pick the right software, connect it up, and off you go.

That is the wrong place to start. The most common failure mode in AI implementation is building something technically correct that solves the wrong problem. It works perfectly and nobody uses it, because it was not solving the right thing to begin with.

Every engagement at AI Consultancy starts the same way. Before looking at any tool, before writing a single line of anything, a structured activity called a friction audit is conducted. It takes about twenty minutes and it completely changes what gets built.

What a friction audit actually is

A friction audit is a structured conversation about time. A business owner walks through a typical week in detail. Not in summary, but specifically: what happened on Monday morning? What took longer than it should have? What got done three times this week that has been done three hundred times in the past year?

The goal is not to find impressive things to automate. The goal is to find the tasks that are taking time that should not be taking time, and to understand exactly why they take the time they do.

This distinction matters enormously. A task that takes two hours a week because it is genuinely complex is different from a task that takes two hours a week because the information is scattered across three systems and someone has to manually copy it from one to another. The first might not be automatable at all. The second can often be resolved in an afternoon.

The three questions at the core of every audit

Within the friction audit, three questions consistently surface the most valuable opportunities:

1. What gets done more than once that feels like the same thing every time? Repetition is the clearest signal that automation is possible. Writing the same email twenty times a month with minor variations is a candidate. Manually updating the same spreadsheet weekly is a candidate.
2. What falls through the cracks? The follow-up that did not happen. The invoice that went out late. The client whose appointment confirmation was forgotten. These failures are almost never the fault of the person. They are the fault of a system that relies on human memory. AI does not forget.
3. What gets done after hours that should be happening during business hours? Answering the same client questions at 9pm that were answered at 9am is a sign that the process is broken. The goal is not to work more hours. It is to build systems that work when people are not.

What the audit consistently surfaces

Across businesses of different types and sizes, the same categories emerge from every friction audit. They vary in specifics but not in type:

• Quote or proposal generation, usually manual and slower than it needs to be
• Follow-up communications, happening inconsistently or not at all because other work gets in the way
• Appointment or job scheduling, consuming time that a tool could handle automatically
• Repetitive client-facing documents, intake forms, reports, and updates that are essentially templates with minor variations
• Information transfer, moving data from one system to another by hand because the systems do not communicate

Each of these categories has well-established solutions available today. The question is always which one fits the specific business, and what order of implementation makes the most sense. Trying to fix everything simultaneously is the surest way to fix nothing. For questions about cost, complexity, or data safety before starting, the most common questions are answered here.

Why this comes before any tool conversation

When businesses skip the friction audit and go straight to tools, they automate the wrong things. They see a compelling demo and try to implement it without first asking whether it solves a real, costly problem. They spend time and money building something that works technically but does not move the needle, because it was not addressing the right problem from the start.

The friction audit creates clarity. By the time any tool conversation begins, there is already a precise understanding of what problem is being solved, how much time it currently costs, and what a good outcome looks like. The tool becomes a means to an end rather than the end itself.

You can run this on your own business

A friction audit does not require an external consultant to conduct. Sit down with a notebook and walk through the past week in as much detail as possible. For every task, ask: did this take longer than it should have? Is this a version of something done before? Could a machine have handled this if it had the right inputs?

The challenge is usually not finding things to automate. It is prioritising which ones to tackle first and ensuring the solution fits the actual business rather than a generic template.

If you want help running a friction audit, or if you have done this yourself and want to discuss what you found, that conversation is always free.