My co-founder priced out three GTM engineer candidates in the same week and got numbers that differed by $200K depending on whether the recruiter was pitching a junior operator or a staff-level builder from an AI-native company. Nobody had told us this role has three completely different price tags depending on how you buy it, not just how senior the person is.
So here's the actual math, broken into the three ways founders are buying GTM engineering right now: fractional, agency, and full-time. Each one has a real number attached, and the number that matters isn't which is cheapest, it's which one matches how proven your GTM motion already is.
Fractional: $1,500–$4,000 a month
This is the tier we should have started with and didn't. A fractional GTM engineer, usually a contractor splitting time across three or four clients, will wire up your Clay enrichment, connect your CRM, and build a first pass at lead routing for $1,500 to $4,000 a month. That's roughly what one week of a full-time hire's fully loaded salary costs. The work is front-loaded: infrastructure and workflows get built once in the first four to six weeks, then tuned. If your ICP is still moving or you haven't run a channel long enough to know what converts, this is the only tier where a bad bet costs you a few thousand dollars instead of a quarter of runway.
Agency: $6,000–$12,000 a month, or $15,000–$50,000 per project
This is the tier most founders skip straight past, and it's the one I'd argue for if you're under $2M ARR and outbound is becoming a real channel but you don't yet have the volume to justify a full-time seat. A GTM engineering agency retainer runs $6,000 to $12,000 a month, or $15,000 to $50,000 for a scoped build like a full outbound infrastructure rollout. Compare that to a full in-house pod, strategist plus content plus outbound support, which runs $22,000 to $45,000 a month in fully loaded cost, and the agency tier looks less like a compromise and more like buying the same output at a third of the fixed cost, with none of it surviving as a liability if the channel doesn't work.
Full-time: $150K to $350K, and the number depends entirely on where you're hiring
The median salary across active GTM engineer job postings is around $127,500, but that number hides a wide spread. Base pay runs $100K to $180K, and total comp including equity and bonus runs $130K to $260K for most operators. At AI-native companies hiring senior or staff-level GTM engineers, total comp climbs to $250K to $350K or more. Vercel and OpenAI are both paying north of $250K total comp for the role, which is the number that shows up in Glassdoor searches and makes every other quote look either cheap or absurd depending on which one you saw first. Python and SQL fluency is the single biggest driver of where you land in that range, not years of experience.
What that comp data doesn't include: ramp time. A new hire, even a strong one, takes 60 to 90 days to be net-productive on your specific stack and data. Budget the first quarter of a full-time hire as pure infrastructure cost, not pipeline generation, the same way you would with the agency or fractional tiers, except now you're paying full salary for it.
The rule we use now
Do the math backward before you post the job. Take six months of a fully loaded full-time salary, roughly $75K to $175K depending on the band you're targeting, and compare it to six months of fractional or agency spend at your current usage. If your agency or fractional bill is already within 70% of that six-month number because you've scaled usage up, hire in-house, you're already paying most of the fixed cost without owning the system or the person's full attention. If your fractional spend is still a small fraction of that number, you're not ready to convert the fixed cost, you're still validating whether the motion is worth automating at all.
We waited until our agency retainer had crept up to $9,000 a month and we were adding scope every quarter before we converted to a full-time hire. That was the actual signal, not a revenue milestone or a headcount plan. The spend told us before the org chart did.