The new role · Roles

Hire a marketing-ops builder or a growth engineer

Put two candidate hires in the same year-one dollars, separating measured inputs from guessed ones, and name the month-six result that sets the next hire.

casemediumExercise · 3 min · 57 of 60

On this page

One hire, two roles the task ledger created, and a founder who wants the one that sounds like growth. This practises what got automated and what did not.

The situation

The invoicing app has 5,000 customers at $50 a month, $3M of ARR, and 5% monthly churn, so a customer returns $800 of lifetime margin. ARR has been flat for two quarters. It can fund one hire at $150,000 a year loaded; a team hour is worth $100.

A marketing-ops builder would take the ten workflow tasks the team of three does by hand, 102 hours a month, down to 17 hours of exceptions, carrying the 40 hours of monthly upkeep themselves. Building takes two months. The builder would also restart the monthly upgrade send to 20,000 active free users, dropped in spring for lack of hours; when it last ran, a holdout showed it caused 25 upgrades a month.

A growth engineer would ship onboarding experiments in the product, which has no engineer free for marketing work until next year. Assume one experiment a month, one in four wins, and each win lifts activation by 10% relative, about ten more paying customers a month from the month after it ships.

The question

Decide which role to hire first, and say what you would need to see at month six to make the other role the next hire.

A strong answer · Rubric · Where answers go wrong

Write your answer first, then open

A strong answer

Put both in year-one dollars and mark which inputs are measured.

Builder. Two months to build, then 102 − 17 = 85 hours a month returned for ten months: 850 hours, $85,000. The send restarts in month 3: 25 customers a month for ten months, 250 customers × $800 = $200,000 of lifetime margin. About $285,000 against $150,000, from measured inputs: the hours from the ledger, the 25 from a holdout.

Growth engineer. Twelve experiments, three wins, shipping around months 4, 8 and 12. The first adds ten customers a month for eight months, the second for four, the third nothing within the year: 120 customers, $96,000. The lifts persist, so the engineer's second year is worth more, about 30 customers a month or $288,000 a year from the first three wins alone. But the one-in-four win rate is a guess, and at one in eight the first year is closer to $30,000.

Hire the builder first: roughly three times the year-one value, from numbers already measured, paid back inside the year. For a company that has been flat for two quarters, the first year is the horizon that matters.

At month six, make the growth engineer the next hire if the restarted send still shows at least 20 incremental upgrades a month against a holdout and upkeep has stayed near 40 hours. Then the builder's case is proven, a second builder adds little, and the next limit is activation, which only code in the product moves. If upkeep runs past 60 hours a month instead, the workflows are fragile, and the next hire supports operations.

Rubric

Puts both hires in one unit, year-one dollars or lifetime margin, with the arithmetic shown.
Separates the builder's hours returned from the lifecycle send's yield, and values the hours at $100.
Phases each benefit in: two months of building, and wins that land through the year.
Labels the ledger hours and the holdout's 25 as measured, and the one-in-four win rate as a guess, with its sensitivity.
Decides, and names a month-six measurement that sets the next hire.
In an interview, acknowledges the engineer's larger second year and says why it does not change the first hire.

Where answers go wrong

"The growth engineer, because growth is the company's problem." It never prices either hire, and leans on the one input that is a guess. The reviewer's objection: three wins landing late in the year are worth a third of what the builder returns from measured numbers.
"The builder, because AI can automate the rest of marketing too." The right hire for the wrong reason. Without the 85 hours and the 25 upgrades it cannot say what the builder is worth, and the same reasoning would cut the team.
"Neither: buy an automation tool for less." A tool still needs an owner. The 40 hours of monthly upkeep and the exceptions do not disappear with a licence, and the send still has nobody to run it.