Hiring for ‘High-Agency’: Identifying Remote Talent That Doesn’t Need Micro-Management

Two virtual assistants get the same task: schedule a call with a new client. The Calendly link is broken. Neither of them knows that yet.

The first one clicks the link, sees the error page, and messages you: “Hi! The scheduling link isn’t working.” Task returned to sender. You now have your original task back, plus a diagnosis to run.

The second one opens the link in an incognito window to rule out a cache issue, finds the event type was deleted, digs up an old confirmation email to see how the original was configured, rebuilds it, sends the invite, and mentions it in the end-of-day recap. You find out the problem existed and got solved in the same sentence.

Same job description. Same hourly rate. Wildly different math for you. If your remote hire routes every ambiguity back to your desk, you didn’t buy back time—you bought a subscription to clarifying questions. Fifteen of those a day at five minutes each is more than an hour of your day spent being someone else’s search engine.

TL;DR

High-agency hires act on the goal. Low-agency hires act on the instruction, and when the instruction runs out, so do they. Interviews can’t tell you which one you’re looking at because interviews remove the exact condition—ambiguity—where agency shows up.

A short paid trial task with a deliberate hole in it will tell you more in two hours than three rounds of interviews.

Key Takeaways

• High-agency means ownership of the outcome, not enthusiasm in the interview. The person who rebuilds the broken Calendly link is high-agency. The person who reports “link doesn’t work” is not.

• Interviews select for rehearsal. Every candidate on earth describes themselves as a self-starter. Test under ambiguity instead.

• The single best live signal is question quality. Outcome questions (“who reads this report?”) predict agency. Step questions (“what font?”) predict a future full of step questions.

• Design your trial task with one missing piece on purpose. How they handle the gap is the whole assessment.

• Once you hire a high-agency person, micromanagement will train the agency out of them in about 90 days. The hiring problem becomes a management problem fast.

What high agency actually means

Strip the buzzword down and high-agency is this: when the instruction runs out before the outcome is reached, the person keeps going.

It is not hustle. It is not personality. Some of the highest-agency people I’ve seen are quiet and methodical, they just refuse to hand a problem back in worse shape than they received it.

Low-agency people are often hardworking. They respond fast. They ask lots of questions. It looks like diligence. What it actually is: every decision in their queue getting rerouted through yours.

That’s the version that hurts most, because you can’t point to anything wrong. The work gets done. You’re just somehow still doing all of it.

Why your interview can’t detect it

Interviews are theater, and everyone rehearses. Ask a candidate if they’re a self-starter and you will get a yes, delivered with a prepared story. Behavioral questions (“tell me about a time you took initiative") don’t fix this. Stories get polished.

The candidate who interviews best is frequently the candidate who prepared best, which measures conscientiousness about interviews, not agency about work.

The deeper problem is structural. Agency shows up under ambiguity, and a good interview removes ambiguity by design. You explain the role clearly. You ask clear questions. You’ve built a controlled environment and then tried to observe a behavior that only appears in uncontrolled ones.

So stop trying to interview your way to the answer. You need to watch them work.

The signals that actually predict it

Question quality comes first. Before a high-agency person starts a task, they ask about the outcome: who’s this for, what does done look like, what’s the deadline that actually matters.

A low-agency person asks about steps: which template, what font, should the file name have underscores. Outcome questions mean they’re planning to make decisions. Step questions mean they’re planning to make you make decisions.

Second: ask what they built that nobody asked for. An SOP they wrote for a previous boss. A tracker they set up because the old process annoyed them. A shortcut they documented for the next hire. People with agency leave artifacts behind. If the answer is a blank stare, believe the blank stare.

Third: the application itself is a test you didn’t have to design. Did they follow your instructions exactly? Good. Did they follow the instructions and add something specific—a Loom walkthrough, a relevant work sample you didn’t request? Better. Did they ignore your instructions entirely to show “creativity”? That’s not agency. That’s a preview of them ignoring your SOPs later.

Design a trial task with a hole in it

Pay for one to two hours of real work, actual work from your business, not a hypothetical, and leave one piece deliberately missing.

Say you ask for a competitor pricing summary and don’t specify the format. Or you hand over a spreadsheet where Column B mixes three different date formats and say nothing about it. The gap is the assessment.

Three outcomes. They don’t notice the gap: pass. They notice and immediately stop to ask before touching anything else: maybe, at least they saw it. They notice, make a reasonable call, and flag it in the handoff (“Column B had mixed date formats, I standardized to MM/DD/YYYY, here are the four rows I wasn’t sure about”): hire that person before someone else does.

You’re not grading whether their call matched the one you’d have made. You’re grading whether they made one, and whether they told you about it without being asked. And yes, pay for the trial. Fifty to a hundred dollars to avoid a bad six-month hire is the cheapest insurance in your business.

Red flags that look like strengths

“I’ll do whatever you need.” Sounds flexible. But it might mean they’re planning to defer every decision to you indefinitely. Flexibility about tasks is fine; flexibility about judgment is the thing you’re trying to screen out.

“I’m a perfectionist.” Often translates to: I stall when instructions are incomplete, because I’m afraid of being wrong. Perfectionism and agency are close to opposites: one needs certainty before acting, the other acts to create certainty.

Instant replies to every message during the trial period. Responsiveness feels great. But someone who answers within ninety seconds every single time may be someone who can’t sit with a problem long enough to solve it. Watch for the candidate who goes quiet for forty minutes and comes back with the thing finished.

Don’t train the agency out of them

Most low-agency remote workers were high-agency once, at some job, until a manager taught them that decisions get punished and questions get praised.

If you rewrite every deliverable they submit, they will stop deciding and start guessing what you want. If a reasonable wrong call gets treated like a crisis, they’ll never make another call. Correct the decision framework, not the individual decision: “next time, default to the client’s timezone” beats redoing the calendar invite yourself in silence.

Define outcomes and constraints. Skip the steps. Give them a budget of decisions they’re allowed to get wrong. The 90-day window after hiring determines whether the agency you screened for survives contact with your management style.

---

Back to those two VAs and the broken link. The gap between them was never skill; both could rebuild a Calendly event. The gap was that one believed the outcome was theirs and the other believed the instruction was.

That gap doesn’t show up on a resume, and it hides beautifully in an interview. It only appears when something goes slightly wrong and nobody’s told them what to do about it. So build that moment into your hiring process on purpose, because your business will build it in by accident every single day after they start.

Most founders hire for skills, manage for compliance, and then wonder why they’re still the bottleneck.

Hire for the gap.

Book a strategy session today

FAQs

Can you train someone to become high-agency?

Partially. You can raise it in someone who has it suppressed, usually by making wrong calls safe and right calls visible. You generally can’t install it in someone who has never shown it anywhere. Screen for the raw material; develop it after.

Does hiring high-agency people mean I can skip writing SOPs?

No. High-agency people are the ones who actually use SOPs well, because they treat them as defaults to work from rather than scripts to hide behind. The SOP handles the 80% that repeats. Their judgment handles the rest.

How much should I pay for a trial task?

Their normal hourly rate for the time it takes, typically $30–$50 total. Unpaid trials filter out exactly the confident, in-demand candidates you want.

Isn’t “high-agency” just a fancy way of saying “hire senior people”?

No. Agency and experience are different axes. Plenty of ten-year veterans route every decision upward, and plenty of two-year Offsite Professionals will rebuild your broken Calendly link without being asked. Experience tells you what someone has seen. Agency tells you what they do about it.

What happens when a high-agency hire makes a genuinely bad call?

You fix the framework, not the person. One bad call handled calmly costs you an afternoon. One bad call handled badly costs you every future call they would have made.