Framework — think before you build

Business Readiness Audit

Before you invest in automation or AI, answer these questions honestly. Most businesses discover they need process design, not technology. This audit helps you find out before you spend.

The Honest Business Health Check

Before you can decide if you need automation, you need to know where you're actually wasting time and money. Not where you think you're wasting it. Where you actually are.

Section A: Pain Points and Time & Money Reality

What's actually painful in your business right now?

Not "what could be improved" — what genuinely hurts? What keeps you up at night? What makes your team frustrated?

Common pain points:

  • Work sits waiting for days (bottlenecks)
  • Same mistakes happen repeatedly (errors, rework)
  • Information gets lost or is hard to find (knowledge gaps)
  • Customers complain about slow response times
  • Team works late because of manual tasks
  • You're turning away business because you can't scale

Write down the top 3 that actually hurt. If nothing genuinely hurts, you probably don't have a problem worth solving.

What tasks does your team do repeatedly?

List them. Be specific. "Admin" isn't specific enough. "Manually entering invoice data from PDFs into Xero" is specific.

How much time per week does each task take?

Be honest. Track it for a week if you're not sure. Most people underestimate by 50%.

What's the hourly cost?

Annual salary ÷ 1,800 working hours = rough hourly rate. Don't forget to include employer NI and overheads.

What's the annual cost of this manual work?

Weekly hours × hourly cost × 48 working weeks. This number is usually shocking.

If you could buy those hours back, what would you do with them?

This is the real question. If the answer is "nothing," then the problem isn't worth solving.

Section B: The Documentation Test

Could a new hire do this task from written instructions?

Yes or No. Not "probably" or "with some help." Yes or No.

Do those instructions actually exist?

In a document someone can read. Not "in Bob's head."

When the person who "knows how this works" is on holiday, what happens?

If the answer is "we wait for them to come back" or "we wing it," you have a documentation problem, not an automation opportunity.

The brutal truth: If you can't document a process clearly enough for a human to follow, automating it will be expensive and fragile. Fix the process first, then think about automation.

Section C: The Bottleneck Map

Where does work sit waiting?

Approvals? Handoffs between people? Reviews? Information gathering? List the places where things grind to a halt.

What causes the delays?

Is it people being busy? Systems not talking to each other? Lack of information? Poor communication?

What breaks when someone's sick or busy?

If the answer is "everything," you have a single point of failure problem. Automation won't fix that. Training and cross-skilling will.

The brutal truth: If the bottleneck is "Bob hasn't trained anyone else," automation won't fix that. You need knowledge transfer, not technology.

Section D: Data Quality and Availability

You can't automate what you can't measure. And you can't measure what you can't access. This section is where most automation projects actually fail.

Do you actually have the data you need?

Not "it exists somewhere" — do you have reliable access to it in a format a system can read?

  • For SMEs: Is your customer data in spreadsheets, emails, or a proper CRM?
  • For corporates: Which systems hold the data? Can you extract it without manual exports?

Is the data clean and consistent?

Missing values, duplicates, inconsistent formats (e.g., "UK" vs "United Kingdom" vs "GB") will break automation.

  • Can you trust the data quality without manual checking?
  • Do you have data validation rules in place?
  • How often does someone have to "fix" the data manually?

Who owns the data and can you actually use it?

  • For SMEs: Do you have admin access to your own systems? Some SaaS tools lock you out of bulk exports.
  • For corporates: Data governance, GDPR, privacy policies — can you legally move this data between systems?

The brutal truth: If your data is messy, incomplete, or locked in systems you can't access, fix that first. Automating bad data just creates bad outputs faster.

Section E: Success Metrics and Measurement

If you can't measure success, you can't prove ROI. And if you can't prove ROI, you can't justify the investment.

What does success actually look like?

Be specific. "Save time" isn't specific enough.

  • Time saved: Reduce invoice processing from 10 hours/week to 2 hours/week
  • Error reduction: Cut data entry errors from 5% to under 1%
  • Response time: Answer customer queries within 1 hour instead of 24 hours
  • Cost reduction: Save £15k/year in manual processing costs

The brutal truth: If you can't define what success looks like in numbers, you're not ready. "It'll make things better" isn't a success metric.

The "Is This Worth Solving?" Calculator

Not every problem is worth solving. Here's how to figure out if yours is.

Worked Example: Invoice Processing

The Scenario: A company manually processes 50 invoices per week from supplier emails into their accounting system.

Annual cost of doing nothing: 2 hrs/week × £25/hr × 48 weeks + correction time = £3,000/year

Automation cost: £3,500 setup + £500/year running costs

Break-even: 1.4 years. 3-year saving: £4,000

Decision: Probably worth it — clear ROI, reasonable break-even, ongoing savings.

Test 1: The Complexity Test

Low complexity
Same thing every time. Clear rules. No judgment needed.
Medium complexity
Mostly repeatable, but sometimes you need to think.
High complexity
Lots of moving parts. Exceptions are common. Judgment required.

The brutal truth: Are you solving a £2k problem with a £20k solution? Don't buy a Ferrari when you need a forklift.

Test 2: Governance, Compliance, and Risk

The bigger you are, the more these questions matter.

  • Who has budget authority to approve this?
  • Can you prove who did what, when, and why? (audit trails)
  • Does this vendor meet your security and privacy standards?
  • Where is the data stored? Does that comply with your policies?

The brutal truth: If you work in a regulated industry (finance, healthcare, legal, government), governance isn't optional. Factor in 3–6 months for approvals and compliance checks.

Test 3: Change Management and User Adoption

The best automation in the world is worthless if people won't use it. This kills more projects than technical issues.

  • Have you asked the affected people what they need?
  • Do they see this as "helping me" or "replacing me"?
  • Who provides support when something goes wrong?
  • Does a senior leader actually care about this and champion it?

The brutal truth: People resist change. If you haven't budgeted time and money for training, communication, and hand-holding, your fancy automation will gather dust.

Test 4: The Readiness Test

The final gate. Can you honestly answer "Yes" to all of these?

  • Are your processes documented? Not "mostly in people's heads" — actually written down.
  • Do you have executive buy-in? Someone with budget authority who actually cares.
  • Do you have someone to own this project? An actual person with time allocated.
  • Is your data clean and accessible? Not "it's somewhere in the system."
  • Can you measure success? In numbers, not feelings.

If you can't answer Yes to all five: You're not ready yet. That's not a failure — it's useful information. Fix what's blocking you first.

Not sure where to start? The AI Decision Framework helps you work out whether you need automation, process redesign, or something else entirely.

Use the AI Decision Framework Back to AI Works