GTM engineer: what they actually build

The short answer

What is a GTM engineer

A GTM engineer is a technical person on the go-to-market side of a company. GTM means go-to-market: everything involved in getting a product in front of buyers and turning them into customers. The GTM engineer builds the automation behind that. Data that enriches leads, workflows that notice when an account changes, outbound that runs from those workflows, and the CRM plumbing underneath it all.

The meaning that matters in practice: a GTM engineer writes code or something close to it, but the output is pipeline, not product. Apollo describes the role as a technical operator who designs, builds and maintains sales and marketing systems. Greg Meyer at Data Operations puts it as combining people, process, tools and data so outcomes get more consistent. Both are right. The second one is the one I'd hire on.

I should say who is writing this. I'm not a GTM engineer. I'm a developer who spent years in enterprise software and now builds Merge & Tell alone, and the go-to-market part of that is the part I was always worst at. I wrote this from the postings, the vendors' own pages and the pay data linked below, all read on 7 October 2026.

Where it came from

What is GTM engineering, and where the title came from

GTM engineering is the practice; GTM engineer is the person doing it. Clay's own guide defines it as building automated revenue systems with AI, data enrichment and workflow automation, and says Clay coined the role in 2023. The same page says about 100 GTM engineer listings go live every month. Clay sells the main tool these people use, so it has every reason to make the role sound big. The postings below suggest it is real anyway.

Before the title, the work existed under other names: marketing ops engineer, sales systems, growth engineer, "the person in RevOps who knows Python." What changed is that enrichment and AI research steps got cheap enough to run on every account, and somebody had to own the machine that does it.

The work

What they actually build

Strip the vocabulary and almost every GTM engineering project is one of these.

• Enrichment pipelines. A domain or an email goes in. Out comes company size, industry, tech stack, the right contacts and their details, usually by trying several data providers in order until one answers.
• Signal workflows. Something changes at an account: a funding round, a new executive, a job post, a pricing page visit. The workflow notices, checks the account is worth it, and tells the right rep with the context attached.
• Outbound infrastructure. Sequences that get fed by those workflows instead of by a list someone built on a Tuesday. Plus the unglamorous parts: sending domains, warmup, bounce handling.
• CRM systems. Objects, fields, routing rules and automations, and the scripts that keep the data from rotting.
• AI steps. A model that reads a company's site or a call transcript and writes a field, a summary or a first line.
• Reporting. Which of all the above actually made money. The step everyone puts last and then can't do because nothing was logged.

The tools for each of these are mapped layer by layer in the GTM tools guide, and five of the workflows are written out in GTM workflows.

Real postings

The job, from three real postings

Definitions are cheap. Job posts say what a company will pay someone to do. These three were open on Ashby when I read them, and they show how wide the title stretches.

Clay's version is closest to a systems engineer for Salesforce. Mercor's is a first hire building the whole outbound machine. Attio's is post-sale, building inside customers' CRMs. Same title, three different jobs. Read the bullet points in any GTM engineer posting before you read the title.

A day of work

A week of the work

This is not a real person's calendar. It is the work those postings describe, laid out across a week, so you can see the rhythm.

Monday starts with the runs that broke over the weekend: an enrichment provider changed a field name, a webhook got a 429, a sequence sent to a list that should have been suppressed. Most of a GTM engineer's mornings are maintenance, and the ones who pretend otherwise have stopped checking their logs.

Midweek is building. Sales wants to know when a target account hires a head of data. That means finding a job-post source, matching companies to CRM accounts, deciding what counts as a match, writing the alert, and agreeing with the reps on what they do when it fires. The code is maybe a day. The agreeing takes longer.

Thursday is a meeting with RevOps about a field that two workflows write to differently. Friday is reporting: which of last month's signals turned into meetings, and which ones should be switched off.

Neighbouring jobs

GTM engineer vs RevOps vs sales engineering

The three get confused because they sit near each other and share tools. They do different things.

Apollo's comparison sums up the ops split in one line: "One governs process. The other ships revenue systems." In smaller companies the GTM engineer often sits inside RevOps. The sales engineer is the odd one out. Their customer is the buyer in a live deal, not the internal team, and their systems work ends where the demo environment does.

On the marketing side, the neighbours are developer marketing and devrel. Both reach developers directly. Neither builds the sales machine.

Titles and job descriptions

GTM engineer roles and a job description

The titles I saw on postings and pay surveys: GTM Engineer, Senior and Staff GTM Engineer, Founding GTM Engineer (first hire, builds everything), Forward Deployed GTM Engineer (builds for customers, after the sale), and GTM Engineer Manager. The founding and forward-deployed versions are the most different from each other, and the pay data below lists them separately for that reason.

If you are writing a GTM engineer job description, here is a starting point built from what the real postings have in common. Change the tools to the ones you actually run.

GTM Engineer

You will build and run the systems our sales and marketing teams use to find,
research and reach customers.

What you will build
- Enrichment pipelines that turn a domain into an account we can act on
- Signal workflows: when something changes at an account, the right rep knows
- Outbound infrastructure: sequences fed by data, not by hand-built lists
- CRM objects, fields and automations that keep the data true
- AI research steps, with a human check on anything that speaks to a customer

What you will need
- SQL, and enough Python or JavaScript to call an API and parse what comes back
- Admin-level comfort in our CRM (Salesforce / HubSpot / Attio)
- Experience with Clay, n8n, Zapier or Make
- A habit of logging what your automations did, including when they did nothing

How we will measure it
- Pipeline created by systems you built
- Hours of manual research and list building removed
- Rep trust: the alerts you send get acted on

The last line of the measurement list is the one hiring managers skip. A GTM engineer whose alerts reps ignore has built a very elaborate way to send Slack messages.

AI

The AI GTM engineer

"AI GTM engineer" mostly means the same job with more of the research done by models. Clay's Claygent is the common example: agents inside a Clay table that research accounts, find data that has no provider, score leads and write personalization. Apollo's piece on the AI GTM engineer describes the job moving toward designing workflows where AI writes and humans review.

The postings back that up. Mercor wants someone to build AI agents for lead research and routing. Attio asks for hands-on Claude Code or Cursor experience. Writing code with an AI assistant is now listed as a skill on GTM postings, which tells you how far the job has moved toward engineering.

Pay

GTM engineer salary

The honest answer is that the sources disagree, because they measure different things. Here is each one as it reports itself. I have not averaged them, because averaging a median of posted ranges with a recruiter's band and a self-reported total compensation number produces a figure that describes nobody.

For a sense of the neighbours: the US Bureau of Labor Statistics puts the median for sales engineers at $124,900 as of May 2025. The three postings above list $140K to $250K. Read the Levels.fyi figure with its sample size in mind. Twenty-six people is a dinner party, not a market.

ZipRecruiter and Glassdoor rank for this search too. Both refused to load for me, so I left their numbers out rather than copy them from a search snippet.

The signal they ignore

The signal they mostly ignore: what your own team shipped

Every signal in the usual GTM engineer toolkit is about somebody else. Their hiring, their funding, their visits to your site, their research on G2. All useful. All probabilistic, all bought from someone, and all available to your competitors at the same price.

The one signal that is certainly true, arrives first, and belongs only to you is what your product team just shipped. A merged pull request that adds SSO matters to every deal that stalled on SSO. A fixed bug matters to the account that churned over it. And in most companies, sales finds out from the changelog weeks later, if the changelog exists, which mine didn't for years.

That gap is the reason I built Merge & Tell. It reads each merged PR, decides whether it is something a customer would notice, and writes posts and a changelog entry for the ones that are. For a GTM engineer, the part that matters is that the same answer is available as data. One call returns what shipped in a window, ranked, with copy already written:

curl -H "Authorization: Bearer $MERGETEL_TOKEN" \
  "https://merge.tel/api/v1/highlights?days=2&kinds=feature,improvement&min_magnitude=minor"

Each item looks like this. The values are made up. The field names are the real ones.

{
  "pr_id": "5b0e7c1a-3f2d-4c8e-9a61-0d4b2e7f9c13",
  "repo": "acme/app",
  "number": 812,
  "title": "Add SAML single sign-on for Okta and Entra ID",
  "github_url": "https://github.com/acme/app/pull/812",
  "merged_at": "2026-10-06T15:42:09Z",
  "kind": "feature",
  "visibility": "user-facing",
  "magnitude": "major",
  "confidence": "high",
  "reason": "New sign-in option customers configure themselves.",
  "pitch": "SSO is here. Connect Okta or Entra ID from Settings…",
  "already_promoted": true
}

The same query is an MCP tool, get_highlights, documented on the MCP server page, and the public changelog is also an Atom and JSON feed for tools that only watch feeds. To be clear about the edges: Merge & Tell does not read your CRM and does not send sales email. The join between "we shipped SSO" and "these six deals asked for SSO" happens in your Clay table, n8n flow, Zap or HubSpot workflow. The closed-lost reactivation play builds it end to end, and the product signals guide makes the longer argument.

Getting in

How people get into it

Most GTM engineers I have seen describe themselves came from sales development, RevOps or growth marketing and learned to code on the job, or from engineering and got curious about why sales was copying CSVs by hand. Both routes work. The postings ask for SQL, an API you can call without help, and a CRM you can administer. A portfolio of two or three workflows that ran for real beats a certificate.

On courses: Clay runs cohorts through Clay University and says they are free to join, and its cohort list includes a four-week GTM engineering bootcamp called AlphaForge, which was taking interest-list signups only when I looked. Apollo's GTME program is a 12-week build done alongside your team, closer to a service than a course. Several GTM engineering courses on Maven were marked no longer available. The community most people point to is r/gtmengineering.

Keep reading

The rest of this guide

• GTM toolsThe GTM tech stack layer by layer: data, signals, sequencing, CRM, glue, and release signals.
• GTM workflowsFive workflow recipes with the trigger, the steps and the way each one breaks.
• Buying and product signalsBuying signals, intent signals and trigger events, and the first-party signal you already own.
• Closed-lost reactivationA deal was lost over a missing feature. You ship it. The rep hears the same day.
• Content engineeringContent built by systems instead of by hand, with a working example that writes from code.

Honesty

What this page leaves out

I have never held this job, and I am not going to pretend the week above is mine. I also have no survey of my own, so the only numbers here are other people's, linked. Pay data for a role this new moves fast and the samples are small. Check the date on any figure before you put it in an offer letter. And Merge & Tell covers one signal for one kind of company, one that ships software from GitHub. If your product team doesn't work that way, the section above is an argument, not a tool.