The GTM tech stack by company stage: the layers, the tools and what to build

A GTM tech stack is the set of tools a company uses to find buyers, reach them and turn their replies into pipeline. The tools change as revenue grows. The jobs stay the same at every stage: data, verification, signals, enrichment, reasoning, sending, CRM and reporting.

We build these stacks for clients every week, and this is the map we work from. It covers what the stack looks like at each company stage, what each layer does, and the tools we run in it. Every tool we use has its own entry in our GTM tools directory.

01The GTM tech stack by company stage

We first tried sizing stacks by team size, and it broke down quickly. Some companies hire reps before $1M in revenue, and some lean operations run well past it without one. Revenue held up as the axis, so the map below runs on it.

At the founder-led stage the stack exists to buy back the founder's time. There is no team to run yet, so the founder works LinkedIn by hand and Claude covers the research, the scoring and the first drafts.

Around $1M the functions start to split and real automation shows up. This is the first stage where every layer further down this page has a tool of its own, and Claude Code moves from helper to orchestrator.

At around $10M the lean tools give way to platforms for data, intent, engagement, booking and conversation intelligence. Past that, our map puts spend at $30K a month and up as an example, and the strongest teams stop adding seats. They build custom internal agents on Claude Code and keep their data in their own warehouse.

Claude is the one tool that shows up at every stage of the GTM tech stack. Only its job changes. Early on it covers the people you cannot hire yet, and later it builds the tooling you would otherwise keep renting.

02The eight layers every GTM tech stack covers

Strip the logos away and every stage covers the same jobs. This is how we group the tools we run in production.

Data and verification

No single database covers everyone, so the order you call sources in decides both quality and cost. We check Supabase first, so anyone already on file is reused rather than bought again. BlitzAPI then works domain-first, finding the people at each company on the list. Prospeo works title-first and only sees the people those two steps missed.

Every address gets verified, and BounceBan settles catch-alls without sending a message. A clean result stops the chain. Failures go to Findymail, then LeadMagic, one at a time, and every new address is re-verified. The paid steps only ever see the leftovers, which keeps the cost per verified contact down. Waterfall enrichment and B2B data decay each have a planned guide of their own.

Signals

Signals decide timing. RB2B identifies who is on the website, down to the person. Trigify logs who engages with LinkedIn posts, and Sales Navigator shows who has been viewing a profile. All three land in the same data layer, so a signal becomes a record that can be enriched and scored.

Enrichment and orchestration

Clay runs its enrichment waterfall, the signal layering and the per-segment personalisation for the outbound motion. n8n moves a lead from sourcing to the reasoning step without anyone touching a spreadsheet. It is optional early on: add it when manual handoffs start costing real time.

The AI brain

Claude scores the leads, drafts the copy, builds the campaigns and classifies the replies. Claude Code holds the custom infrastructure, GitHub stores it, and Supabase sits underneath as the system of record. Each function has its own knowledge base, and each task has a skill file Claude reads before it runs.

The other seven layers produce data. This one decides what to do with it.

Sending

Email runs on EmailBison, Smartlead or Instantly, with domain warm-up and inbox rotation as volume grows. LinkedIn runs on HeyReach across several sender profiles. Kixie puts a call into the sequence when an account is worth a live conversation. We pick the senders per client.

Every sending tool holds its own copy of the campaign. Something has to decide who moves from email to LinkedIn to a call, and stop all three the moment someone replies. In our systems Claude makes that call, so each tool only has to send. Which channel to lead with is its own decision, and we cover it in LinkedIn outreach vs cold email.

CRM

Attio holds the relationship and the pipeline in most of our builds, and HubSpot does where the client already runs it. iClosed books the call and records the outcome. A CRM only stays useful if its structure is built before the first email goes out, so we work through a CRM setup checklist in build order.

Reporting

Weekly client reports and daily performance monitors pull the numbers, render them and post to Slack. We build these with Claude instead of buying a dashboard, because each one is shaped to how the team works.

03Where Claude and Clay sit in the stack

We have built 300+ Clay tables for clients. For a long time we ran Clay as the centre of everything: enrichment, scoring, routing and copy. It could do all of it, so we let it.

The trouble was context. The centre of a GTM system has to remember what happened, and a table holds a row. When a campaign ended, whatever the system had learned ended with it.

So the hierarchy changed. Claude now orchestrates the whole system end to end. Clay still runs inside outbound, where it is the best tool we have for enrichment waterfalls and signal layering. Supabase holds the record underneath, so nothing evaporates when a campaign closes.

If you are weighing Clay against other enrichment tools, our guide to Clay alternatives maps the field.

04The tools we run for clients

On sales calls, questions about our GTM tech stack usually come down to which tools are in it. For a typical outbound build, the short answer looks like this.

• Sourcing: Prospeo, with BlitzAPI and Apollo in the contact-finding waterfall we build in code.
• Verification: BounceBan, with DeBounce alongside it.
• Signals: RB2B for website visitors, Trigify for LinkedIn engagement.
• Enrichment: Clay inside the outbound motion, with n8n for the handoffs.
• Email: EmailBison or Smartlead, chosen per client.
• LinkedIn: HeyReach.
• Brain and memory: Claude and Claude Code, with Supabase as the record.
• CRM and booking: Attio or HubSpot, with iClosed.

For the wider market, including the tools we do not run, the top 100 GTM tools encyclopedia sorts them by category. Our outbound service builds and runs this stack for clients.

05Sales tech stack vs GTM tech stack

A sales tech stack usually means the tools a sales team works in every day: the CRM, a sequencer, a dialler and a booking tool. The GTM tech stack covers the whole route to revenue. That includes the data, the signals and the reasoning that decide who the sales team talks to in the first place. The sales tools sit inside it as the sending and CRM layers.

06What to build and what to buy

The line between building and buying moved this year. We built an email tool in about an hour that drafts replies in your own voice and triages what needs an answer first. It runs on one Gmail connection and one Claude project. A year or two ago that build would have taken a small team most of a quarter.

Testing whether you can build something yourself went from a quarter of engineering time to about an hour. So the line gets re-drawn twice a year, tool by tool.

07Who runs the GTM tech stack

A stack needs one owner. In a founder-led setup we have built it as one founder posting and one junior running the pipeline behind the posts. Further up, the owner is usually a GTM engineer, the person who builds the system the revenue team sells from. We cover that discipline in our guide to GTM engineering.

The title matters less than the ownership. Whoever owns the system end to end is the one who knows which two or three APIs to wire together.

08How to build your GTM tech stack

1. 01Write down the jobs firstList the eight layers and mark which ones your motion needs today. The functions matter more than the tools.
2. 02Match the stack to your revenueUse the stage table as a ceiling. Anything bought ahead of your stage becomes a seat nobody operates.
3. 03Start with one tool per channel you useIf your buyers live in their inbox, start with one email tool. Add HeyReach when your buyers are active on LinkedIn.
4. 04Put the reasoning layer in earlyClaude can research, score and draft from day one. At the founder-led stage it covers the people you cannot hire yet.
5. 05Build the CRM before the first send, then add memorySet up the CRM structure before any email goes out. Add Supabase as the memory layer once the loop is running and there is history worth keeping.
6. 06Re-check build-versus-buy twice a yearGo tool by tool. Keep buying sending and data, and build the decision layer and reporting yourself once testing it costs about an hour.

You do not need every tool on this page. Different combinations work for different clients, depending on budget, stage and what you sell. Cover the layers first, then pick the tools.