# What is GTM engineering? > GTM engineering is building the system that decides who to contact, why and when, runs the outreach, and writes every result back to your CRM. Here is what that means in practice. - Canonical page: https://warmup.ai/blog/what-is-gtm-engineering - By Mike Hurley, founder of WarmUp. Published 2026-10-27. - Category: GTM engineering (https://warmup.ai/blog/category/gtm-engineering) - Part of: GTM engineering (https://warmup.ai/gtm-engineering) - Tags: gtm engineering, signal-based outbound, clay, crm ## Key takeaways - GTM engineering builds the system behind outbound: who to contact, why, when, and where every result is recorded. - Every contact is on the list for a reason that is true this week, such as a hire, a funding round or a job change. - The system lives in tools your team can read, like Clay and your CRM, and every touch is written back to the CRM you already report from. - Built well, it is documented and handed to your team, so it does not walk out the door with the people who built it. GTM engineering is the work of building the system that decides who your company should contact, why, and when, then runs that outreach across email and LinkedIn and writes every result back to your CRM. It replaces bought lists and guesswork with signals, and it is built in tools your own team can read, run and change. That is the short answer. The rest of this post is what it looks like when we do it. ## What does a GTM engineer actually build? A GTM engineer builds four connected parts, on the stack the company already has. 1. **Signals.** A feed of reasons to reach out: a company hiring for a role you sell to, a funding round, a buyer changing jobs, a visit to your pricing page, or an event only your business can see. The feed decides who gets worked this week. 2. **Lists and enrichment.** Every contact the signals surface is enriched through a waterfall of data providers, so one gap does not drop them, then verified twice before anyone is allowed into a sequence. Contacts are scored against your own closed-won deals, not a generic fit score. 3. **Plays.** A play is the sequence that fits the signal: the email, the LinkedIn touch, the gap between them, and the call to action. Copy is drafted from your own sales calls, so it reads like your team. 4. **Infrastructure and CRM sync.** Sending domains and inboxes on one side, kept separate from your main domain. HubSpot or Salesforce on the other, with every touch, reply and meeting written back as it happens. None of the four is new on its own. What makes it engineering is that they are built as one system, with rules a person can read, instead of four tools that each know a piece of the story. ## How is it different from paying someone to send emails for you? A vendor that sends emails for you is selling activity. You get a volume of sends and a report from their dashboard, and when the contract ends, the lists, the copy and the know-how leave with them. GTM engineering sells the machine. The work is deciding who to contact and why, building that decision into tables and rules, running it, and measuring it in your CRM. Because the system is built on your stack and documented as it goes, it can be handed to your team. A vendor paid by the send has no reason to do that. The difference shows up in the questions you can answer. With a vendor, "why did we email this person?" usually has no answer. With an engineered system, the answer is on the record: the signal that put them on the list, the play it triggered, and what happened next. ## What is a signal, and why does it matter? A signal is a fact that is true this week and makes a company more likely to need you now. A hire, a funding round, a job change, a visit to your site. The signal decides the play and the timing, so nobody gets a cold email simply because they exist in a database. Signals come from three places. Public data, such as job posts, funding news and job changes. Custom scrapes, which we build with Claude Code for the signals nobody sells. And your own data: product events, CRM history, and the closed-won and closed-lost deals that show what a good customer actually looks like. Signals matter because they change what outbound is. Without one, an email is an interruption. With one, it is a reason to talk that the buyer can recognise. ## What does it run on? It runs on tools your team can open, not on a black box. A typical build for us looks like this: - **Your CRM**, HubSpot or Salesforce, as the source of truth and the scoreboard. - **Clay** for enrichment, scoring and routing, with one table per job. - **Instantly** for email and **HeyReach** for LinkedIn, so each channel has its own sending infrastructure. - **Claude Code** for custom scrapes, and **Claude** for drafting and classifying replies. - **Slack** for the moments a person needs to act, like a positive reply. The tools are examples of each step, not the identity of the work. If you already run on something else, the system is built there instead. What matters is that every step is somewhere your team can see it. ## Where do people stay in the loop? Research, enrichment, drafting and reply classification are automated. The judgment calls are not. A person reads a sample of every batch before it launches. A person approves every play. Every positive reply gets a drafted response within minutes, and a person owns the send. The call to action is fixed by you and never rephrased by a model. Nothing is drafted unless the signal behind it is verified. That split is deliberate. Automation is good at doing the same careful thing many times. People are good at deciding whether it was the right thing to do. A system that removes the people entirely tends to send confident emails to the wrong companies. ## How long does it take to get a system live? For us, four weeks from intake, if your CRM data is reachable in the first week. The first two weeks build the model from your own deals and set up the signals. Week three is enrichment and verification. Week four is the first plays, and launch. Sending domains need two to three weeks of warm-up before they carry real volume, so they are set up on day one. After launch, the numbers are read with you every two weeks and the plays that are not working are changed. The system is meant to get better on a schedule, not to be finished. ## Who should own the system? You should. This is the part most outbound setups get wrong, and it is why we built our work around it. The common failure is a system nobody owns. A contractor builds a Clay table that lives in their head. Three tools overlap. The person who understood it leaves, and the next person starts again from scratch. Every rebuild costs months and the data from the last one. A GTM engineering system is built to be handed over. The tables, the plays, the documentation and the numbers move to someone on your team, with training. We explain exactly how in [How we hand a system over](https://warmup.ai/blog/how-we-hand-a-system-over). ## How do you know if it is working? You read it in the CRM you already trust. Every touch, reply, meeting and outcome is written there, so outbound shows up in the same reports as everything else. There is no separate dashboard to believe. Every two weeks, the numbers are reviewed: which signals produce replies, which plays produce meetings, and which do not. The plays that are not working are changed. Over time the list gets cleaner, because every company that drops out leaves a reason behind, and each reason is a filter for the next pull. If you want to see what this looks like for your market, the [GTM engineering page](https://warmup.ai/gtm-engineering) walks through how we run it, week by week. ## Questions, answered ### Is GTM engineering the same as RevOps? No, but they share a foundation. RevOps makes sure marketing, sales and finance run on the same data, with one definition of a lead and one rule for who owns it. GTM engineering builds the system that finds the right accounts and works them. Both live in your CRM, which is why we often do the two together. ### Do you need Clay to do GTM engineering? You need somewhere to enrich, score and route contacts, wired to your CRM. We build that in Clay because it lets one table pull from many data providers and hand the result to the CRM. The tool is a choice; the system around it is the point. ### Can GTM engineering guarantee a number of meetings? No. Anyone who guarantees meetings is choosing the number before they know your market. What a good system guarantees is that every send has a reason and every result is in your CRM, so you can see what works and change what does not.