AI And Automation Playbook

Why “Automations Candidate Finder” matters (and what you’ll be able to do)

What if you could stop guessing which parts of your business deserve AI and automation - and instead pick the best targets in a repeatable way?

Most teams start by collecting ideas like “we should automate something” or “AI could help with emails.” That’s a dead end. You end up with random experiments, half-built tools, and workflows that nobody trusts. The real problem isn’t that you lack ideas - it’s that you lack a simple way to compare opportunities using the same yardstick.

After this chapter, you’ll use the Opportunity Radar Scorecard to find the best AI and automation candidates from your real workflow list. You’ll score each candidate, explain why it scores the way it does, and pick a short list that you can test fast. You’ll also learn what to watch for so you don’t waste time on “cool” projects that don’t work in your day-to-day operations.

Nadia, 34, operations manager at a small logistics firm, faces the same issue: her team receives shipment updates, exceptions, and customer messages every day, and the work piles up when volume spikes. She doesn’t need a giant automation program - she needs the right first automation that saves time without breaking trust. This scorecard gives her a way to choose that first target with confidence.

Practical takeaway / prompt: Write down three tasks your team does every week and ask: “Which one would cost us the most time if it failed?” That question sets you up to score candidates instead of guessing.

How It Works: The Opportunity Radar Scorecard (how to score real candidates)

The Opportunity Radar Scorecard turns messy workflow ideas into a clear shortlist. You’ll score each candidate using the same set of criteria, then you’ll choose the top targets that balance impact, feasibility, and risk.

Before you score anything, define what a “candidate” looks like in plain terms: one business task you do today (like “triage exception tickets” or “draft carrier status replies”) that you can improve with AI or automation. If you can’t describe the task in one sentence, you don’t yet have a candidate - you have a vague wish.

Use these criteria. Score each one from 1 to 5 (1 = weak, 5 = strong). Add them up. You can keep the math simple: total score from 6 to 30.

1. Time in the workflow (1-5): Score how much time the task consumes per week.

Why: automation pays off when it removes real recurring work. If Nadia’s team spends 8 hours weekly on exception triage, that candidate deserves a higher score than a task that takes 30 minutes twice a month.

2. Repeatability (1-5): Score whether the task repeats with similar steps and inputs.

Why: automation needs consistent patterns. If the same shipment status types show up every day, you can handle them with AI and rules more reliably.

3. Input quality (1-5): Score how clean and usable the data is today (emails, forms, tracking numbers, structured fields).

Why: AI performs better when it can read something clear. If your “inputs” come as messy screenshots, you’ll need extra steps first.

4. Decision clarity (1-5): Score whether you already know the “right” outcome or policy.

Why: if nobody agrees what “good” looks like, AI will produce inconsistent results. Nadia can define policies like “If tracking shows delivered and customer says ‘not received,’ request proof of delivery or escalate.”

5. Risk of bad output (1-5, but invert your thinking): Score how risky it is if the automation makes a mistake.

Why: you should still automate risky areas, but only when you can add guardrails. For a high-risk task (like billing changes), you need strict validation and human review.

6. Integration effort (1-5): Score how hard it will be to connect the candidate to your existing tools.

Why: the best idea fails if it can’t reach the right system. If your shipment status updates live in one place and you can access them via an API (Application Programming Interface), your integration score improves.

If you want a quick sanity check, ask yourself: “Could someone else perform this task the same way tomorrow?” If the answer is no, your repeatability score will stay low until you document the steps.

Ask yourself: Which candidate has the highest total score and the lowest “risk of bad output” score that still allows guardrails?

A concrete example: scoring an exception message triage

Nadia’s firm gets customer messages like “Where is my order?” and “Carrier says delayed - what now?” She wants to automate the first response draft and tag the message for routing.

...