A workflow is a good first automation when it answers yes to most of seven questions: is it frequent, rule-based, done by hand today, high-volume, costly when it goes wrong, a bridge between two apps, and contained enough to review safely? Score the workflow from 0 to 7, run every candidate through the same seven questions, and automate the highest scorer first. A score of 5 or more is a strong first automation. Anything below 3 belongs later, or never.
This checklist turns "pick your first workflow" from a gut call into a number you can compare. If you would rather we run this for you, see how we deliver AI business automation agents. Everything below is yours to use on your own.
Why does the first workflow choice matter so much?
Because adoption is everywhere and value is rare. McKinsey calls it the gen AI paradox: roughly 80% of companies have deployed generative AI, yet about the same share report no material impact on earnings. Around 90% of function-specific use cases never leave pilot, and fewer than 10% of organizations are scaling AI agents in any function.
That gap is a targeting problem, not a technology problem. The companies that capture value do not sprinkle AI across everything. They pick two or three high-value use cases and redesign the workflow around the agent instead of bolting it on. In McKinsey's data, workflow redesign is the single factor most correlated with bottom-line impact, and high performers are nearly 3x more likely to have fundamentally redesigned how the work runs.
The upside when you choose well is real. An IDC survey of more than 4,000 business leaders found an average return of $3.70 for every $1 invested in generative AI, rising to $10.30 per $1 for the top leaders. The distance between $3.70 and $10.30 is mostly about picking the right first workflow, not about the model. The checklist below exists to put you on the right side of that line.
The 7-point checklist: is this a good first automation?
Take one candidate workflow. Ask each question. Score one point for every yes, zero for every no. Add the points. The maximum is 7.
| # | The question | Score 1 if yes |
|---|---|---|
| 1 | Frequent. Does this run many times a day or week? | It repeats often, not occasionally |
| 2 | Rule-based. Do the steps stay roughly the same each time? | Little human judgment is needed |
| 3 | Manual. Is a person doing it by hand today? | No tool already handles it |
| 4 | High-volume. Does it move a meaningful number of items or records? | The throughput is real, not a trickle |
| 5 | Error-costly. Does a mistake cost money, time, or trust? | Errors today are painful to fix |
| 6 | App-bridging. Does a person copy data from one app into another? | It connects two systems by hand |
| 7 | Contained. Can the result be reviewed without touching the whole business? | A clear, bounded output |
That is the whole tool. Run every candidate through the same seven questions and you get a ranked list instead of a hunch. The experts name these criteria one at a time across their guides, but none package them into a single score you can act on. This does.
Two of the seven carry extra weight, so weigh ties toward them:
- App-bridging (point 6) is the highest-signal item. The strongest first candidate is almost always a spot where a person reads data out of app A and types or pastes it into app B. McKinsey Global Institute estimates 69% of data processing and 64% of data collection work is technically automatable, and these manual bridges are exactly that work.
- Contained (point 7) is the safety item. A workflow you can review in isolation can be piloted without risking the rest of the operation. A workflow that touches everything cannot.
How do I read the score?
The number tells you what to do next, not just how the workflow feels.
- 6 to 7: automate this first. It is frequent, mechanical, manual, and safe to review. This is the textbook first workflow, and it is usually the boring one nobody fights over.
- 4 to 5: a solid candidate. Worth doing, but check which point it failed. If it lost the point on rule-based or contained, design a human checkpoint into it before you build.
- 1 to 3: not yet. It may still be worth automating later, but as a first project it will stall. Pick something higher and come back once you have a working template.
- 0: leave it to a person. A workflow that scores zero is judgment work, and automating it removes the value rather than the grind.
The score also surfaces the trap most teams fall into. The most exciting workflow is often the one that fails point 2 or point 7, while a dull, high-frequency, app-bridging task quietly scores 7. Trust the number over the excitement.
Prefer to run it yourself? You can Hire AI Agents and put one to work on your highest-scoring workflow today.
Why does each point earn its place?
The checklist is short on purpose, but each item is doing real work.
- Frequent. A workflow that runs many times a week is where even a small per-run saving compounds into real hours. Automating something that happens twice a month is not worth the build.
- Rule-based. When the steps barely change, an agent can follow them reliably. Asana frames the best candidates as tasks that happen often and follow consistent steps, and HubSpot's clean test for a first workflow is "manual, frequent, and rule-based."
- Manual. If a tool already handles it, there is nothing to win. You are hunting for the work a person still does by hand, because that is the time you get back.
- High-volume. Volume is what turns a saved minute into a saved week. It is also what justifies the cost of building and running the automation properly.
- Error-costly. Asana's third trait for a good candidate is that the cost of errors or wasted time is high enough to justify automating. A mechanical task nobody minds getting wrong now and then is a weaker first pick than one where mistakes are expensive.
- App-bridging. The Zapier playbook's single best heuristic is that any place a person bridges two software platforms by hand is a prime first candidate. These bridges are frequent, rule-based, and pure overhead with no judgment a customer would miss.
- Contained. HubSpot warns small teams to automate one workflow first, not everything at once, and to pick something high-impact but contained. Containment is what lets you pilot, measure, and prove the return before you touch anything else.
Notice how the points reinforce each other. A workflow that bridges two apps is usually frequent and rule-based too. That clustering is the signal: the best first automations score high across several points at once, not on a single dramatic one.
What does a high-scoring workflow look like?
Some workflows reliably score 6 or 7 across most businesses. Match the example to your own audit rather than copying the list, but these tend to win first:
- Lead capture and routing. A new lead arrives in the inbox, gets copied into the CRM, and an owner is assigned. Frequent, rule-based, manual, and a textbook app-to-app bridge. Sales reps spend about 70% of their time on non-selling, mostly administrative work, and automating it can lift sales productivity by around 14.5%.
- Invoice and data entry. Order or invoice details re-keyed from email into accounting software. High frequency, high volume, high error cost, low judgment. Gartner estimates up to 30% of accounting work is automatable.
- Behavior-triggered follow-up. A renewal reminder or a nudge after a customer action. Contained and easy to review, and instant automated responses make a follow-up about 42% more likely to land.
- Support triage and repetitive answers. The same five questions, drafted by an agent with a person reviewing. Language-heavy, frequent, and safe to check.
Each of these fits the profile: frequent, rule-based, manual, and contained enough to review. The right answer for you is whichever candidate scores highest on your own seven questions, not whichever is most common in an article.
What does a bad first candidate look like?
The checklist is just as useful for ruling things out. A workflow makes a poor first automation when it fails the points that matter most:
- It needs heavy human judgment (fails rule-based). Pricing a custom deal or handling a sensitive complaint is judgment work. Keep the human and automate the prep around them.
- It runs rarely (fails frequent and high-volume). A quarterly report is real work, but automating it returns little and is hard to justify as a first project.
- Its steps change every time (fails rule-based). If you cannot describe the process clearly enough for a new hire to follow it, an agent cannot follow it either.
- It triggers high-stakes actions with no review step (fails contained). Anything that moves money or makes irreversible changes needs a human checkpoint before it qualifies.
Aim for the sweet spot the work-management playbooks describe: automation that removes the grind without removing the judgment that still needs a person. A low score on rule-based or contained is not a failure of the workflow, it is a signal to design a checkpoint in, or to pick a different first project.
How do I turn a high score into a working automation?
A high score tells you what to automate first. It does not build it. Once you have your top scorer, the path is short and the same every time:
- Map the current workflow. Write down who touches it, what triggers it, what information they read, what decision they make, and what they produce. This map defines the trigger, conditions, and actions, and it is your proof later of what "working" means.
- Decide where a human checkpoint belongs. Use the points the workflow lost. If it scored low on contained, that is where the agent escalates instead of guessing.
- Run one pilot and measure it. Run the new workflow alongside the old one for a short period and compare against the map. Pick the metric up front: hours saved, records processed without errors, faster follow-up.
- Prove the return, then reuse the loop. Once the pilot clears the bar, you have a template. Score your next candidate with the same seven points, funded by the savings from the first.
This is the part where most do-it-yourself pilots stall, and it is exactly why ~90% of function-specific use cases never leave pilot. The score is the easy half. The build, the integration, the monitoring, and the iteration are the half that keeps a workflow out of the pilot graveyard.
How to get started
You do not need a transformation program to begin. This week, list your candidate workflows, run each one through the seven questions, and automate the one that scores highest. Map it, pilot it end to end, measure it, then reuse the checklist on the next one. That is the whole method, and it is the difference between the companies that get $3.70 back per dollar and the leaders that get $10.30.
If you want the fastest path, you can skip the trial and error. We score your candidate workflows with you, pick the single highest-ROI first automation, then plan, build, integrate, and run the agent, including the monitoring and iteration that keep it out of the pilot graveyard. Book a free consultation below and we will decide your first automation together.