The 30-minute audit that finds the hours your team is losing.

The most common failure mode in workflow automation isn't the build. It's picking the wrong thing to automate in the first place.

We see it constantly: teams that spend three months and $40k automating a workflow nobody really cared about, while the thing that actually bleeds them 20 hours a week sits untouched because nobody mapped the right candidates. The problem isn't capability — it's triage.

This post is the framework we use on the first call with every new customer. You can run it yourself in about 30 minutes with a blank sheet of paper. It won't replace a proper scoping exercise, but it will tell you which three workflows are worth looking at first.

Step 1: list the top 10 time sinks (5 minutes)

Don't overthink this. Ask your team a single question: "If you could never do one recurring task again, which one would you pick?" Collect 10 answers. Write them down exactly the way your team describes them — not cleaned up, not categorized. The messy language matters.

You're looking for things like:

• "Going through invoices every Monday and matching them to POs"
• "Copying lead data from forms into HubSpot"
• "Chasing PMs for status updates for the weekly report"
• "Classifying customer support tickets"
• "Reconciling Stripe payouts with the accounting system"

If your team gives you abstract answers ("reporting," "data entry"), push for the specific version — what exactly they're doing, where, and for whom.

Step 2: score each one on four signals (15 minutes)

Here's the actual framework. Each workflow gets scored 1–5 on four dimensions. The sum tells you whether it's a good candidate.

Signal 1: Repetition (how often does this happen?)

1 = monthly or less. 3 = weekly. 5 = multiple times a day. High repetition means the automation earns back its build cost fast. Low repetition means even a perfect automation won't move the needle.

Signal 2: Rule clarity (could a new hire learn it in a day?)

1 = requires deep judgment or domain knowledge. 3 = mostly rule-based with occasional judgment. 5 = fully rule-based, no human judgment needed. This signal tells you whether the workflow is an automation problem (high score — deterministic rules) or an agent problem (middle score — some judgment), or something to leave alone for now (low score — you're not ready).

Signal 3: Volume (how much time does it eat?)

1 = under an hour a week total across the team. 3 = 5–10 hours per week. 5 = a full person's worth of time or more. High volume means big payback. Low volume means don't bother.

Signal 4: Delay cost (what does it cost when it's slow?)

1 = no downstream consequence. 3 = mildly annoying — someone waits. 5 = serious business cost — customer experience suffers, deals stall, compliance risk. This is the signal most teams miss. Even a low-volume workflow is worth automating if the delay costs you a deal or a customer every week.

Sum the four scores for each workflow. You'll have a total between 4 and 20.

Step 3: sort and look at the top three (10 minutes)

Rank your 10 workflows by total score. Look at the top three. These are your high-ROI automation candidates.

You're looking for anything scoring 15+. That's your tier one. Tier two is 11–14 — worth automating, but only after tier one is done. Anything below 11, put on a shelf. You may come back to it, but it won't be the thing that moves the needle.

The patterns in the top three

When we run this with customers, the top three almost always cluster into a small number of shapes:

• Document intake and classification — invoices, contracts, tax forms, applications. High repetition, high volume, high delay cost, moderate rule clarity (because of edge cases). This is the single most common winner.
• Cross-system data sync — CRM to billing, billing to accounting, support tickets to product backlog. High repetition, clear rules, medium volume, medium delay cost. Classic automation territory.
• Status rollups and reporting — weekly reports, QBR prep, pipeline snapshots. High delay cost (slows decisions), medium volume, medium rule clarity. Often a good agent candidate because of the synthesis component.
• Routine response and triage — support ticket routing, first-touch responses, lead qualification. High volume, clear rules for the common cases, high delay cost because customers are waiting.

If your top three look like any of these, you're in well-charted territory. The build is predictable, the payback is predictable, and you'll see value inside the first month.

The trap to avoid

The most common mistake teams make at this stage is picking the most interesting problem instead of the highest scoring one. Interesting problems are interesting because they're novel, complex, or visible to leadership. High-scoring problems are often boring — invoice matching, data entry, classification. But those are the ones that compound.

Do the boring one first. The interesting one gets easier to tackle once the boring one has freed up 15 hours a week of team capacity to work on it.

What happens after the audit

Once you have your top three, the scoping questions get concrete: what systems does the workflow touch, what's the current error rate, what's the acceptable AI error rate, what's the handoff if something goes wrong. Those are the questions that turn a score into a buildable spec.

If you'd rather not do that scoping alone, book a 30-minute call. We run this exact audit with you on the call, rank the candidates together, and tell you which two or three are worth building — with a realistic cost and timeline for each. No deck. No follow-up sales sequence. Just the ranking.