A self-assessment tool for service businesses

The
Workflow
Audit.

Five questions that reveal exactly where your business is losing time. And what to automate first.

For service businesses of 2 to 30 people · 20 minutes · Switzerland & Italy

This audit covers
Q 1Where does your time actually go?
Q 2Which tasks follow a repeatable process?
Q 3What breaks when a task is done late?
Q 4Where is information getting lost?
Q 5What would 8 recovered hours buy you?
+Priority scoring matrix
+30-day first automation plan
Introduction

AI is not a technology problem.
It is a workflow problem.

Every service business, from a two-person accounting practice to a twelve-person marketing agency, loses a measurable portion of its week to tasks that are repetitive, rule-based, and entirely automatable. The problem is rarely a lack of available tools. It is a lack of clarity about which workflows to address, in what order, and why.

This audit is built on the same framework that Andrew Ng's team used to guide enterprise AI transformations at Google and Baidu. Adapted for businesses that don't have a Chief AI Officer, a dedicated engineering team, or a $10 million technology budget. The principles are identical. The scale is human.

22%
of the working week lost to repeatable tasks that could run automatically
8.8 h
average hours recovered per week after a MEIKAI workflow implementation
2 wk
median time to implement a first automation sprint

What this audit will give you

Work through the five questions in order. Each one builds on the previous. By the end, you will have:

How to use this document: Print it. Use a pen. The act of writing forces specificity. And specificity is what makes the difference between an automation that works and one that doesn't.
Questions 1 & 2 of 5

Finding the tasks worth automating.

Question 1
Where does your time actually go?

List every task that you or your team performs more than once a week, in roughly the same way. Don't filter for what feels important, write down everything, invoicing, reminders, client updates, booking confirmations, report preparation, data entry, follow-ups.

The playbook principle: The highest-value automation wins come not from the most impressive technology, but from the most repetitive work. Start by making it visible.
Task Who does it Hours / week How often (daily / weekly / monthly)
Question 2
Which of these tasks follow a consistent, repeatable process?

For each task in your list above: could you write it down as a numbered sequence of steps? If yes. It can almost certainly be automated. If no, if it depends heavily on context, judgment, or relationship. It belongs in a different category.

The rule: A task that can be documented as a process can be delegated to a machine. A task that requires judgment cannot: yet. The first category is larger than most business owners expect.
Task (from Q1) Can you write it as steps? (Y / N) Notes

Go forward only with tasks where the answer is Yes. The others are not ready for automation. And pushing them before they are costs more than it saves.

Questions 3 & 4 of 5

Measuring the cost of the problem.

Question 3
What breaks when this task is done late, or not at all?

For each task you marked "Yes" in Q2, ask: what is the downstream consequence if it's delayed by two days? A missed invoice reminder delays cash. A late appointment confirmation increases no-shows. A delayed client status update damages trust. Name the consequence specifically.

Why this matters: The consequence tells you the priority. A task whose delay costs CHF 400€ 400 in no-shows is worth automating before a task whose delay costs 20 minutes of re-work.
Task Consequence if delayed or missed Priority (High / Med / Low)
Question 4
Where is information getting lost between people, tools, or steps?

Information that lives in multiple places gets lost. Every time you copy from an email to a spreadsheet, from a spreadsheet to an invoice tool, or from a booking system to your calendar: there is a gap. That gap is where errors live and where time disappears.

The playbook principle: Workflows that require data to flow between systems are the highest-leverage automation targets. A well-connected system enters data once and distributes it automatically.

Mark every gap that applies to your business:

  • Client intake information re-entered manually
  • Time or project data copied from notes to invoices
  • Booking confirmations sent from a separate tool
  • Payment status tracked separately from accounting
  • Client communication history stored in email only
  • Appointment reminders sent from a different system than the calendar
  • Reports assembled from multiple spreadsheets
  • Supplier or contractor info tracked in a notebook
  • Project status updated manually per client
  • Compliance documents assembled file by file

Every item you checked is a workflow gap. Most can be closed in a single automation sprint.

Question 5 of 5

Calculating what recovered time is worth.

Question 5
If automation recovered 8 hours of your week, what would you do with them?

This is not a hypothetical. It is your ROI calculation. Eight hours per week is the average outcome of a MEIKAI workflow implementation for a service business of 2–10 people. Some recover six. Some recover fourteen. The question is: what are those hours worth to you specifically?

The honest version of this question: Are the hours currently lost to these tasks hours that would go to billable work, to client relationships, to business development, or simply to rest? All four answers justify the investment. Only one gets factored into most ROI calculations.

Your personal ROI estimate

Hours recovered per week (conservative estimate) _________ h
Your effective hourly rate (or cost of an employee's hour) CHF _______
Weeks worked per year 46

Annual value of recovered time CHF _______ / year

For reference: a 3-person law firm recovering 9 hours/week at CHF 180/hour = CHF 74,520€ 180/hour = € 74,520 in recovered billable capacity per year. Their implementation cost was CHF 490 (introductory)€ 490 (introductory) for the Clarity Scan plus a two-week build sprint.

If recovered hours go to billable work

Multiply recovered hours × your billing rate × 46 weeks. That is revenue currently being left on the table.

If recovered hours go to growth

Sales calls, client relationships, service development. These don't have an hourly rate: but they have a compounding effect that manual admin never will.

The question that matters most: You have now identified specific tasks, their weekly cost, and what you would do with recovered time. The next step is not to buy a tool. It is to prioritize: which workflow first, in what order, and why.
Priority Matrix

Score your workflows.
Start with the highest number.

Take the tasks you identified in Q1 and Q2 (those that follow a repeatable process). Score each one across three dimensions. The task with the highest total score is your first implementation priority: not because it is the most impressive, but because it costs the most and is the most ready.

Score each dimension 1 (low) to 3 (high):
Task
Hours / week
1 = <2h · 2 = 2–5h · 3 = 5h+
Follows a process?
1 = loosely · 2 = mostly · 3 = always
Consequence if late?
1 = minor · 2 = significant · 3 = costly
Data gets lost?
1 = rarely · 2 = sometimes · 3 = often
Total
/12

Reading your results

Score 9–12: Start here

This workflow is costing you the most and is technically ready. It is your first automation sprint. Do not skip to a lower-scoring task because it seems more interesting.

Score 5–8: Second sprint

Valuable, but either less frequent or less standardized. Come back to these after your first automation is running. They become easier once you have the infrastructure in place.

Score 1–4: Not yet

Low score means low readiness, low frequency, or low impact. Do not automate these first. Many businesses make the mistake of starting with the most visible workflow rather than the highest-cost one. The matrix prevents that mistake.

The ground rules

What automation can and cannot do.

The most common mistake in workflow automation is trying to automate the wrong things. This reference is the distillation of the DeepLearning.AI framework's core insight: AI and automation work best when they can observe a set of inputs and produce a defined output: not when they need to exercise judgment.

Automate these
  • Scheduled communications (reminders, confirmations, follow-ups)
  • Data transfer between systems (intake → CRM → invoice tool)
  • Document generation from existing templates and data
  • Status updates triggered by defined events (payment received, task completed)
  • Appointment scheduling and conflict detection
  • Report generation from structured data
  • Recurring billing and payment tracking
  • Onboarding sequences triggered by a sign-up or contract
Leave these to people
  • Client relationship conversations that require reading the room
  • Complex commercial negotiations or pricing decisions
  • Novel situations with no established precedent
  • Anything where "it depends" is genuinely the right answer
  • Creative work, strategic decisions, ethical judgment
  • Sensitive client communication after a problem or complaint
  • Tasks that require empathy as their primary ingredient
30-Day First Sprint

Your implementation plan.

W1

Choose one workflow. Document every step.

Take your highest-scoring task from the matrix. Write out every step it currently involves: who does what, in which tool, triggered by what event. If you cannot write it down completely, you are not ready to automate it. The documentation is the design specification.

W2

Identify the tools. Map the connections.

Which systems does the workflow touch? Where does data enter, where does it need to arrive? You are looking for the fewest connections that produce the desired output. Complexity is the enemy of a durable automation.

W3

Build the first automation. Run it in parallel.

Implement the automation while keeping the manual process running. Do not turn off the manual version yet. Run both for at least one full cycle, one week, one billing cycle, one onboarding. Compare outputs.

W4

Verify. Turn off the manual version. Measure.

If the automated output matches or exceeds the manual version: stop doing it manually. Track the hours recovered in week five. That number is your baseline. And your argument for the second sprint.

If you'd rather not do this alone

The Clarity Scan does this for you. Professionally. In five days.

This audit gives you the framework. The Clarity Scan gives you the answer.

We talk to your team, study your tools, and send a targeted questionnaire. Over five business days of independent analysis, we map every key workflow, identify your highest-ROI automation opportunities, score them using exactly the methodology in this document. And deliver a written report that tells you what to build first, what to leave alone, and what to watch out for.

The report is yours to keep. Whether you work with MEIKAI for implementation or take it to someone else is your decision entirely.

What the Clarity Scan delivers

  • Complete workflow map of your business operations
  • Opportunity matrix: every automation opportunity scored and ranked
  • First action for each: specific, not generic
  • What not to automate, and why
  • Tool recommendations based on your existing systems
  • Implementation timeline estimate
Investment
CHF 890 CHF 490

Fixed price. Yours to keep regardless of what you decide next.

Timeline
Discovery: interviews + questionnaire
5 days: analysis and report
1 call: report walkthrough
Request the Clarity Scan

Every engagement on our Results page began here.

Thirty minutes. Five days. A report that replaces the guesswork you just did in this audit with a professional diagnostic built on your actual workflows.

CHF 890 CHF 490 Fixed price · Report yours to keep · No lock-in
Request the Clarity Scan → meikai.ch/en/contact