Which work to hand to AI first: rank your workflows by dollars

Hand AI the workflows that cost the most dollars first. Price each one with hours a week x the hourly rate of the person doing it x 52 weeks, rank the list, and automate the top three that can be checked against a source. Most guides rank by hours saved, and hours put the wrong work at the top of the list.

That last claim needs defending, because nearly every page ranking which processes to automate with AI scores them the same way: hours saved, sometimes dressed up as an effort-versus-impact grid. We run a live audit engine that prices workflows for real teams, and the dollar ranking disagrees with the hours ranking often enough that we'd call hours-only scoring the most common mistake in automation planning.

Why hours rank the wrong work first

An hour of a founder's week and an hour of a support rep's week cost different amounts, and the gap is wide enough to reorder the whole list. Five hours a week of a founder's time at $150 an hour comes to 5 x $150 x 52 = $39,000 a year. Eight hours of a support rep's week at $45 comes to 8 x $45 x 52 = $18,720. An hours-ranked list sends the team after the support queue first, while the founder's five hours, costing twice as much, stay untouched.

The expensive workflows also hide where hours-counting doesn't look. Founders and senior people burn hours every week on meeting prep, follow-ups and answering the same questions in Slack, because the answers live in their heads and nowhere else. Nobody logs that as a process, no SOP covers it, and an hours-first list built from whoever complains loudest misses it completely. The loudest complaints usually come from the cheapest hours.

The dollar-ranking method

The method takes an afternoon and a spreadsheet. It's the same math our audit engine runs, so if you do it by hand, your numbers and ours should agree.

List the repeat work. Write down everything that happens weekly or more often: the follow-up after every sales call, the brief before every meeting, the numbers assembled by hand each Friday, invoices chased, support replies drafted, the same customer questions answered again and again. If it has an SOP, or everyone agrees it should, it belongs on the list.

Price each workflow. Hours a week x hourly rate x 52 weeks, using the rate of the person doing the work. The rates our engine uses:

Role	Hourly rate
Founder / CEO	$150
Engineering	$100
Sales	$75
Finance	$70
Marketing	$65
Recruiting	$65
Operations	$55
Support	$45

Swap in your own payroll numbers if you'd like. The ranking usually survives the swap, because what drives it is the gap between roles, and that gap exists on every payroll.

One guardrail: cap any single workflow at 20% of the team's total weekly hours. If a 10-person team works 400 hours a week, no single workflow gets priced above 80 of them. When an estimate blows past the cap, it's either inflated or it's three workflows wearing one name. Split it and price the parts.

Rank and take the top three. Automation projects die from scope, so resist the urge to fix everything at once. The top three dollar figures are where agents go first, and the rest of the list waits until those are live and measured.

A 12-person agency, ranked both ways

Here's an illustrative 12-person agency. The numbers demonstrate the method, and a real quote would come from your own week. Six workflows from a typical week, priced at the rates above:

Workflow	Who does it	Hours a week	Math	Annual cost
Answering repeat questions	Founder ($150)	5	5 x $150 x 52	$39,000
Sales call follow-ups	Sales ($75)	6	6 x $75 x 52	$23,400
Meeting prep briefs	Founder ($150)	3	3 x $150 x 52	$23,400
Drafting support replies	Support ($45)	8	8 x $45 x 52	$18,720
Weekly scoreboard	Operations ($55)	4	4 x $55 x 52	$11,440
Invoice chasing	Finance ($70)	3	3 x $70 x 52	$10,920

Ranked by hours, drafting support replies wins at 8 hours a week, and this team builds a support bot first. Ranked by dollars, support drops to fourth. The founder's repeat questions, the sales follow-ups and the meeting briefs top the list at $85,800 a year combined, and the meeting briefs, sitting near the bottom of the hours list at 3 hours a week, jump to third because those are founder hours. Same six workflows, and the two rankings send the team after different projects with different paybacks.

If you'd rather have this list for your own company than for an imaginary agency, the free Leverage Audit builds it in about 2 minutes. Give it your website, your team size and the areas you care about, and it estimates the 3 workflows where agents would save you the most, with the hours and dollars shown.

What to hand to agents and what to keep

The dollar ranking tells you where the money is. One more filter decides what gets automated: can the output be checked against a source?

Agents do the top of that example list well. A follow-up drafted from the call transcript can be checked against the transcript. A meeting brief built from the calendar and the account's history can be checked against both. A weekly scoreboard pulls from the same reports every Friday, invoice chasing runs on dates and amounts, and a drafted support reply cites the doc or the past ticket it came from. Every one of those outputs carries its source, so a person can verify it in seconds before it goes out.

Judgment calls stay with people: pricing exceptions, hiring decisions, anything where the right answer isn't written down anywhere. An agent without a source guesses, and a guessing agent in front of customers costs more than the hours it saves. So no, AI won't automate everything, and a team that tries to hand it everything usually ends up trusting it with nothing.

There's also a dependency the tool listicles skip. Agents need a memory to draw on, and the reason repeat questions top the example list is that the answers live in one person's head, like the pricing exception for the big client or what was promised on last quarter's calls. When she leaves, all of it walks out the door with her. A company brain puts meetings, customer conversations and decisions in one searchable source of truth, which is what turns a drafting agent into one that answers with receipts.

What the top of your list is worth

Benchmarks from the audits our engine runs: a 1 to 5 person team lands around $124,000 a year across its top 3 workflows, a 16 to 50 person team around $300,000, and a 51 to 200 person team around $400,000. Those figures count staff time only, so treat them as a floor. They don't include the deal lost to a follow-up that slipped, or the decision remade from scratch because nobody wrote down why it was made the first time.

Against those figures, the cost of acting on the list is small. Our builds start at $15,000, fixed quote, delivered in the first month, and the full breakdown is in what it costs to build AI agents. An optional retainer from $2,500 a month covers new agents and upkeep after launch. We back the math with a 60-day savings review: 60 days after launch we run a cost review from the system's own records, and if it finds less than $15,000 a year in savings you agree are real, you get up to 2 months of the retainer free.

The method has run on our own books too. At Rye Valley, the holding company behind Hockey Stick, ranking the repeat work by dollars and handing the top of the list to agents removed $425,000 in annual expenses and multiplied shipping velocity 10x within the first month after install. The case study has the detail, and the decision log behind it is public at michaelhouck.co/decisions, so you can audit the reasoning the way we'd tell you to audit anyone's.

None of this requires buying anything to start. The ranking method is free, the rates are published above, and a spreadsheet built this afternoon will tell you which workflow to hand to AI first. Most teams expect the support queue to win and then watch a founder workflow take the top spot instead. Price the list before you pick the tool, because the ranking decides the payback and the tool only decides the plumbing.