23 June 20268 min read
Companies that have learned to hire engineers remotely often make a mess of their first marketing or support hire, because they carry over a playbook that does not transfer. Engineering has an unusually legible assessment: there is a work sample, it can be run in an hour, and competence is visible to a competent observer. Marketing and support have neither property, and pretending otherwise produces a confident hire and a disappointing quarter.
Both are worth doing across a border. They just need different questions.
The marketer problem: results are shared
A candidate says they grew a channel substantially. That may be entirely true and still tell you almost nothing, because the outcome belonged to a team, a product, a budget and a market moment, and the CV cannot separate their contribution from the rest.
So stop assessing outcomes and assess decisions. Ask what they chose to do and what they chose not to do, what the budget was, what they tried that failed, and how they knew it failed. A marketer who can describe a channel they shut down and why is showing you judgement. A marketer whose entire history is growth with no closed experiments has either been lucky or is presenting a curated version.
A marketing work sample that is fair
The unfair version is asking for a strategy for your company, which is unpaid consulting and attracts either the desperate or the generic. The fair version is narrow, bounded and about reasoning.
- Give them your public site and one competitor, and ask for three specific criticisms with a reason for each, in writing, capped at a page
- Ask them to write one short piece of real copy for a product they do not know, then discuss the choices on a call
- Show them a plausible funnel description and ask where they would look first for the problem, and what data they would need
In all three, the discussion afterwards carries the weight. You are hiring someone to make judgement calls with incomplete information, which is what the conversation tests and what a portfolio does not.
Be specific about which marketer you need
Marketing is not one job. Performance marketing, content and organic search, lifecycle and email, product marketing, and brand are distinct crafts with different daily work, and someone excellent at one may be genuinely weak at another. Hiring a general marketer for an undefined remit is the most common way this hire fails, and it fails slowly, which is worse.
Decide what the first ninety days consist of before you write the brief, exactly as you would for an engineering role, and accept that a small team usually needs one specific craft done well rather than every craft done thinly.
The support problem: the CV says nothing
Support CVs are nearly interchangeable, and the metrics people quote, tickets handled, response times, satisfaction scores, are shaped by the tooling and the product far more than by the person. What distinguishes a good support agent is judgement under mild hostility, and there is no way to read that off a document.
Fortunately it is one of the easiest things to test directly, because the medium of the job is text and you can simulate the job precisely.
A support work sample that predicts the job
Send three real tickets, anonymised, and ask for the replies they would send. Choose deliberately: one straightforwardly angry customer who is also right, one ambiguous request where the customer has not said what they actually want, and one that should be escalated rather than answered.
The third is the important one. Knowing what not to answer is the single most valuable instinct in a support hire and the hardest to train. Someone who confidently invents a resolution to the escalation ticket has told you exactly how they will behave on your worst day.
In the replies themselves, read for register rather than grammar: calm, specific, no defensiveness, no false promises, and a clear next step with a timeframe. Then ask them to rewrite one of the replies for a customer who has already complained twice, and see whether they can change tone without changing the facts.
Ramp is longer than you expect, for both
Support competence is mostly product knowledge, and product knowledge takes weeks. A strong agent in month three looks nothing like the same person in week one, and judging them in week two produces a false negative. Plan a shadowing period, give them access to the ticket history, and let them draft replies for review before they send unsupervised.
Marketing ramps slower still, because the feedback loop is longer than the probation period. This is why the first ninety days should be judged on process rather than results: did they set up measurement, did they run experiments that could have failed, did they write down what they learned. Results in month three are largely inherited; behaviour is theirs.
What the overlap actually buys you here
For engineering, shared hours mostly reduce the cost of ambiguity. For these two roles the case is more direct. Support has to be live when your customers are awake, and coverage that starts an hour before your own office opens is a genuine operational gain rather than a convenience. Marketing needs to be in the room, or at least on the call, when product and sales decide things, because a marketer who learns about a launch after it was scheduled will always be producing catch-up work.
That is the practical reason these roles are usually better nearshore than far: not because the work is harder, but because both fail quietly when the person is outside the conversation, and neither failure shows up in a weekly report until it has been true for a month.