Three roles get confused in almost every hiring plan: SDR, RevOps and GTM engineer. They are not seniority levels of the same job. They fix different failures, and hiring the wrong one leaves the leak exactly where it was.
The one question that sorts them
Where does the lead die?
If it dies because nobody reached out, you have a capacity problem. If it dies because it reached the wrong person, or arrived without the data needed to qualify it, or the follow up depended on somebody remembering, you have an engineering problem. If it did not die at all but nobody can agree on how many there were, you have a systems of record problem.
That question sorts almost every hiring decision in this area.
What each role is accountable for
SDR. Creating conversations. The output is meetings with people who should be in the room. The role is a person doing a human thing: reaching out, handling a no, earning fifteen minutes. Hire when the process works and there is simply not enough of it.
RevOps. The truth of the business. Systems of record, stage definitions, process design across marketing, sales and success, and reporting leadership can act on. The output is a number everyone believes and a process everyone runs the same way. Hire when three teams report three different pipelines, or when a stage means something different depending on who you ask.
GTM engineer. The machine between a lead entering and a human closing. Enrichment before contact, scoring, routing, sequences, AI channels, the write backs into the record, and the traceability from revenue back to its source. The output is that the process happens without depending on anyone's memory. Hire when everyone agrees what should happen and it still does not happen consistently.
The symptoms, plainly
Hire an SDR when: the pipeline is thin, the offer converts when it gets in front of the right person, and your existing reps are at capacity on conversations rather than on admin.
Hire RevOps when: leadership asks for a number and gets three, stages are used differently by different reps, a report takes someone half a day to assemble, or a stage change quietly breaks something downstream and nobody noticed for a month.
Hire a GTM engineer when: leads sit untouched for hours or days, reps spend the first part of every call looking things up, follow up happens when someone remembers, the same manual step repeats hundreds of times a week, you cannot say which activity produced a closed deal, or a rollout to a second team or branch means rebuilding rather than configuring.
Why the order matters more than the titles
The common failure is hiring for capacity when the problem is engineering. More SDRs against a broken machine produces more activity and the same revenue, at a higher cost, and it usually makes the data worse.
The opposite failure exists too. Hiring an engineer when the real problem is that nobody knows who you sell to produces very elegant automation of a bad idea. A process that has not been written down should not be automated, because automating a confusing process only makes it confusing faster.
A workable default: define before you engineer, engineer before you scale capacity. RevOps or a founder writes down what should happen. The engineer makes it happen without human memory. Then adding people multiplies something that works.
How they work together once you have more than one
The cleanest division in practice is by ownership of layers.
RevOps owns the definitions: what a stage means, what a qualified lead is, which fields exist and who may change them. The GTM engineer owns the execution: what fires when a stage changes, what enriches before contact, where the lead goes, what gets written back. The SDR owns the conversation: the part that needs a person, arriving warm because the other two did their jobs.
When those boundaries blur, the usual symptom is an engineer quietly redefining stages to make an automation easier, and a reporting layer that stops matching the business a quarter later.
If you are still not sure which one you need
Open ten leads from last month that did not close. Read them like a story. If the story stops because nobody called, hire capacity. If the story stops because the record was wrong, the routing sent it to the wrong place, or there is no story at all because nothing was written down, hire engineering.
See the talent, split by whether they face your client or build inside your systems. Reading the profiles and contacting anyone is free.