GTM engineer postings grew around 205% year over year, and a single role posted in one community Slack channel last week drew over 100 applications in seven days.
Here is the problem that creates for you as the buyer: almost every one of those applications lists the same tools. Clay, Apollo, Instantly, Smartlead, n8n. The tools section filters nothing, because the tools are a month of learning. The judgement is years of breaking things.
So hiring one well comes down to separating systems-builders from tool-operators, and there are reliable ways to do it.
The three questions that separate candidates
1. "Tell me about a campaign that broke. What did you check first?"
Anyone can follow a tutorial to launch. Very few people can explain why a domain went quiet in week three and what the diagnostic order was. Tutorials cover the build; nothing covers the hour after it breaks. A real answer names specifics: bounce rate, authentication failing silently, a list that decayed, an out-of-office spike. A weak answer rewrites the copy.
2. "What do your numbers actually mean?"
If a candidate says "242 opportunities," ask how many became held meetings. The gap between what a tool labels and what actually happened is where the job lives. Someone who can say "202 replies, 107 positive, 75 held, and the drop was follow-up speed" has run this. Someone who repeats the dashboard number has watched it.
3. "What would you refuse to build?"
Knowing when not to build something is rarer than knowing how to build anything. A good candidate can name a situation where cold email was the wrong channel, a market too small for the volume someone wanted, or a tool they would not deploy. If everything is always buildable, nothing has ever been their responsibility.
What to pay
| Arrangement | Range |
|---|---|
| Freelance hourly | $125 to $250 |
| Monthly retainer | $2,000 to $10,000 |
| US full-time salary | $120,000 to $160,000, median near $159,000 |
The uncomfortable truth in these ranges: offers far below them still get applicants, because the title is new and plenty of people want it. What those offers attract is tool-operators. A systems-builder has options, and a $500-a-month offer to run three brands reads to them as a warning, not an opportunity.
If you want to sanity-check the other side of this market, the candidate's view of rates is in how to price yourself as a GTM engineer.
The screen-share test
Before any offer, do one working session. Pick a real account from your ICP and ask them to show you, live, how they would research it and write the first line.
This test is close to uncheatable. A tool-operator can talk through a workflow convincingly; watching someone actually drive Clay, read a company's site, and decide what matters takes ten minutes and settles it. Candidates who have done the work suggest this themselves.
Red flags, from reading a lot of these applications
A tools list with no numbers. The stack is table stakes. No outcomes means no ownership.
Career totals without context. "Millions of emails sent" says nothing. One campaign with sends, replies, positives and meetings says everything.
Guaranteed results. Reply rates across real campaigns range roughly 0.2% to 3% opportunities per send depending on list and offer. Anyone guaranteeing a number before seeing your market is guessing with your money.
No questions about your TAM. A real one asks how big your addressable market is before talking about volume, because infrastructure can out-scale a market and they have seen it happen.
What good looks like in the first month
Week one is infrastructure and list building, not sending — domains warming, ICP locked, data sourced and verified. Launch lands around week four at conservative volume. Anyone promising meetings in week one is either inheriting warmed infrastructure or burning something.
And from day one you should see the reporting: sent, delivered, reply rate, positive replies, meetings held, weekly, per campaign. If you have had someone book meetings before while you "never had visibility into what was working," the reporting cadence is the thing to fix in the contract, not the copy.