Skip to main content

GTM engineering: what it is, how it works, and how to start

GTM engineering is the practice of building the systems that find, score, and reach the right accounts, so a revenue team grows pipeline with code, data, and automation instead of only adding headcount. It covers the account list, enrichment, scoring, buying signals, outbound and inbound automation, and the CRM plumbing that ties them together.

This guide is about the practice: what it is, what gets built, and how a company starts doing it. If you want the job itself (pay, skills, how to get hired), read what is a GTM engineer instead.

By Garrett Wolfe, founder of OneGTM, employee #9 at Unify, and author of the State of GTME report. Updated .

Key takeaways

  • GTM engineering builds the machinery a revenue team runs on. RevOps governs it, and reps use it.
  • Most of the work falls into five workflows: TAM and scoring, signals, enrichment waterfalls, inbound routing, and reply handling with CRM sync.
  • Fit and timing are different things. Keep the fit score and the signal in separate fields, and only reach out when both are there.
  • Start with one motion end to end, measure it, then add the next. A half-built stack across everything produces nothing.
  • Hire in-house for steady system work. Bring in a GTM engineering agency to get the first systems live before the hire starts.

What GTM engineering is

For most of the last decade, a B2B company that wanted more pipeline hired more SDRs. Each rep built their own lists, guessed at timing, and wrote their own emails. GTM engineering replaces the guessing with systems: a scored account list, a feed of events that say an account might need you now, verified contacts, and sequences that fire when the two line up.

The term took off in 2024 and 2025. GTM engineer job postings went from 63 in 2024 to more than 3,300 in 2025, and the live count is on the GTM engineer jobs tracker. The GTME Data Report adds who is hiring, how many roles are remote, and where the people in the role came from. The tools caught up at the same time: Clay and its competitors made waterfall enrichment something an operator could run, and AI coding agents like Claude Code made custom pipelines cheap to build.

I did this work as employee #9 at Unify, where a growth team of two ran 40+ plays a day across 100+ sequences. Now I build the same kind of systems for other companies at OneGTM.

GTM engineering vs. RevOps vs. sales ops

GTM engineeringRevOpsSales ops
OwnsThe systems that create pipeline: data, signals, scoring, outbound and inbound automationProcess, forecasting, reporting, and CRM governance across the revenue orgTerritories, quotas, comp, and rep tooling
Typical outputA scored TAM, signal feeds, live sequences, routing, a reply botClean CRM, stage definitions, dashboards, forecastTerritory plans, comp plans, rep onboarding
ToolsEnrichment APIs or Clay, a sequencer, the CRM API, Python or TypeScript, AI modelsMostly the CRM and a BI toolThe CRM, CPQ, and comp software
Measured byMeetings and pipeline per signal, cost per verified contactForecast accuracy, data quality, cycle timeQuota attainment, ramp time

At a seed or Series A company one person often covers all three. As the team grows they split, and the GTM engineer usually reports into RevOps or growth. If your gap is the governance column, a RevOps consultant is the better first call.

The GTM engineering stack

LayerWhat it doesCommon toolsWhat I run
Data sourcesCompanies, people, emails, phones, firmographicsApollo, ZoomInfo, People Data Labs, LinkedIn data APIsSeveral providers behind one API, picked per field
Enrichment and orchestrationWaterfalls, AI research, scoring, schedulingClay, Deepline, n8n, custom codeDeepline plays and Claude Code; Clay for teams already on it
SignalsEvents that say an account may need you nowJob posting feeds, visitor ID, funding data, intent providersCustom feeds matched to the scored TAM
CRMSystem of record for accounts, people, and activityHubSpot, Salesforce, AttioWhatever the client runs; Attio for OneGTM itself
SequencingCold email and LinkedIn outreachInstantly, Smartlead, lemlist, HeyReachInstantly for email, HeyReach for LinkedIn
AI modelsClassification, scoring, research, drafts, reply triageOpenAI, Claude, GeminiCheap models for classification, stronger ones for copy and judgment
Storage and hostingWhere the pipeline lives and runs on a schedulePostgres, Supabase, BigQuery, Railway, VercelSupabase plus scheduled cloud jobs

Every category has its own comparison in the GTM tool guides. The model choices, with per-token cost math, are in best LLM for GTM.

Five workflows GTM engineers build

1. TAM build and fit scoring

Pull every company that could buy, from data providers, public registries, or scraped directories. Add firmographics, then have a model read each website and answer the questions that matter for your ICP (do they sell to hospitals, do they run field reps, are they PE backed). Turn the answers into a 0 to 100 fit score and sync it to the CRM so reps see it on the record.

Check the distribution before anyone uses it. If most accounts land on the same score, the model is not separating anything. For one vertical SaaS client, with a TAM of about 45,000 companies, a rebuild moved the median score from 0.25 to 0.50 and left 2,001 accounts above 0.75, which is the shape you want: a clear top tier to work first.

2. Signal to outreach

Watch for events: a new VP in your buyer seat, a job posting that describes the problem you solve, a funding round, a pricing page visit. Match each event to an account in the scored TAM, find the right person, verify their email, and drop them into a sequence written for that signal. The signal goes in its own field, separate from the fit score, because a signal fades in weeks and fit does not.

The full list of signal types, and how to detect each one, is in the buying signals guide.

3. Enrichment waterfall

No single provider finds everyone. A waterfall asks the cheapest good source first and only pays the next one when the first comes back empty, then verifies every address before it is used, including catch-all domains. Track coverage and cost per verified email by provider, and reorder the waterfall when the numbers move. The providers are compared in email finder tools and email verification tools.

4. Inbound routing

A form fill gets enriched, scored, and assigned to the right rep, and the buyer books a meeting on the spot instead of waiting for a reply. Speed to lead is the whole game here, and most companies lose more pipeline in the gap between form and first touch than anywhere in outbound. Tools for this are in lead routing tools.

5. Reply handling and CRM sync

Every reply gets classified (interested, not now, wrong person, out of office, unsubscribe), posted to Slack with a suggested response, and written back to the CRM with the campaign and signal that produced it. That last part is what lets you report pipeline by signal, which is how you decide what to build next.

What a GTM engineering build looks like, week by week

  1. Week 1: ICP written as fields a machine can check, the first TAM pulled, and the signals worth testing picked.
  2. Weeks 1 to 3: Sending domains and mailboxes bought and warming. This runs in the background and is why outbound never starts on day one.
  3. Weeks 2 to 4: Scoring calibrated against known customers and lost deals, waterfalls wired, signal feeds matched to the TAM.
  4. Weeks 3 to 5: Sequences written per signal, every merge variable checked, first sends in small batches.
  5. Week 5 onward: Weekly review of replies and meetings by signal and segment. Kill what is not working, add the next signal, and write the handoff so your team can run it.

The full module list is on how I work.

How to start GTM engineering at your company

  1. Pick one motion: Outbound to one segment, or inbound routing. Not both at once.
  2. Write the ICP as fields: Industry, size band, region, tech, ownership. If a rule cannot be checked against data, it is a hope, not an ICP.
  3. Inventory your data: What is in the CRM, what is missing, and what it costs to fill. The TAM calculator prices the enrichment.
  4. Build one workflow end to end: From account to booked meeting, with every step logged. Ugly and complete beats elegant and half-done.
  5. Guard the send path: Most outbound failures are silent. A sequencer will happily send "Hi ," when a first name is blank, and it will not bounce or warn you. Check every variable on every enrolled lead before a campaign goes live.
  6. Measure by signal: Positive reply rate and meetings per signal and segment, reviewed weekly. Then decide what to build next.

Hire a GTM engineer, use an agency, or both

Hire in-houseGTM engineering agencyBoth
Best whenSteady system work and someone to manage itYou need systems live in weeks, or before the hire startsAn agency builds, then hands over to the hire
Cost to knowFully loaded salary plus tools; see the cost calculator and salary dataMarket range in the State of GTME report: about $2,000 to $33,000+ a monthBoth, for an overlap period
RiskA long search, and one person who knows how everything worksKnowledge leaves with the agency unless the handoff is realPaying twice if the overlap drags on

Openings and who is hiring are on the GTM engineer jobs tracker. The field of agencies is compared in best GTM engineering agencies. OneGTM's own offer is the GTM engineering agency page. OneGTM is me plus a small team: the person who scopes the build is the person who builds it, and your team owns everything afterward.

Learning GTM engineering

You can learn this for free if you build as you go. I put together a free GTM engineering course: an eight-week path with one project per week, using free tools and the guides on this site. The GTM glossary covers the vocabulary, and the open-source Claude Code skills show how the send path is guarded in practice.

GTM engineering FAQ

What is GTM engineering?
GTM engineering is the practice of building the systems that find, score, and reach the right accounts: account lists, enrichment, fit scoring, buying signals, outbound and inbound automation, and the CRM plumbing behind them. It lets a revenue team grow pipeline with code, data, and automation instead of only adding headcount.
How is GTM engineering different from RevOps?
RevOps owns process, forecasting, reporting, and CRM governance across the revenue org. GTM engineering builds the machinery that creates pipeline: data, signals, scoring, and outreach automation. At small companies one person often does both, and as the team grows the GTM engineer usually reports into RevOps or growth.
What tools do GTM engineers use?
A data layer (several enrichment providers, often through Clay or an API like Deepline), a CRM such as HubSpot, Salesforce, or Attio, a sequencer such as Instantly or HeyReach, AI models for classification and drafting, and increasingly code written with AI coding agents like Claude Code, stored in a database like Supabase and run on a schedule.
Is GTM engineering just Clay?
No. Clay made waterfall enrichment accessible and many GTM engineers learned in it, but the practice is the system design: what to score, which signals to watch, how leads route, and how results get measured. The same workflows can run in Clay, n8n, or code.
Do you need to code to do GTM engineering?
Not to start. No-code tools cover a lot. Some SQL and the ability to read and run scripts written with an AI coding agent will take you much further, because custom logic, scheduled jobs, and cost control at volume are easier in code.
How do I start GTM engineering with no budget?
Pick one motion, write your ICP as checkable fields, pull a small account list from a free source, enrich a few hundred records on free tiers, and build one workflow end to end. Measure replies and meetings by signal before adding anything else.
Should I hire a GTM engineer or use a GTM engineering agency?
Hire in-house when there is steady system work and someone to manage the role. Use an agency when you need systems live in weeks or before your first GTM engineer starts. Many teams do both: the agency builds, then hands the system to the hire.

Want the first GTM systems built before you hire for them?

Related

Start GTM engineering with OneGTM

Tell me what you're building. I reply the same day.

Call with Garrett

Takes about 20 seconds. Times show up right here, and the call is with Garrett directly. See our privacy policy.

Systems we've built for

Rather do it the old way? team@getaob.com LinkedIn