How to Build the Ideal GTM Tech Stack (and Cut the Rest)

TL;DR

• The ideal GTM stack has four layers: data, system of record, sending and orchestration. Anything else has to earn its seat with a number.
• It takes about 45 minutes and spends about 106 LeadOcean records. Sizing is free. On Free, swap the bulk export for a 100-row search pull.
• Buy the data layer first. Every other tool is only as good as the rows you feed it.

Prerequisites

• A LeadOcean API key from app.leadocean.io/register. The free plan has 1,000 records, one-off, no card, 5 requests a second.
• The key in your shell: export LEADOCEAN_API_KEY=your_key_here.
• curl and jq.
• A CRM and a sending tool you already use, or none yet. This guide does not need either to run.
• Optional: an MCP client, to size markets by chat.

Step 1: Write the stack down as four layers

A stack is a list of jobs, not a list of logos. Write one line per job and name the tool that does it. If two tools do the same job, cut one.

data:          # who to contact, and how
  source: LeadOcean API
  job: search, enrich, verify email status, export
record:        # where deals live
  source: your CRM
  job: accounts, contacts, deals, one owner per account
sending:       # how messages leave
  source: your email or sequencing tool
  job: mailboxes, sequences, replies
orchestration: # how the other three talk
  source: n8n, a script, or an MCP client
  job: move rows from data to record to sending

Apply one test to every tool already in your stack. Name the weekly action it triggers and the number that moves. If you cannot, pause the renewal for a month and see who notices.

Three things do not need a seat on day one. A dialer, if you do not cold call. An intent feed, until you can name one action it triggers. A second enrichment vendor, until the first one's miss rate is measured.

LeadOcean covers the data layer only. It has an app with an Exports page, but no sequencer, dialer, CRM or Chrome extension. Pair it with the tools you already run.

Step 2: Size the market before you buy anything

Count the accounts and the buyers, free, before you commit to a stack. If the list is 400 people, you need a spreadsheet. If it is 10,000, you need the four layers.

Company count first. GET /v1/companies/count takes the company-search filters and returns an exact total.

curl -s "https://api.leadocean.io/v1/companies/count?hqCountry=US&employeeRange=51-200,201-500&technologies=HubSpot" \
  -H "x-api-key: $LEADOCEAN_API_KEY" | jq '.data | {total, totalIsExact}'

Array filters in a query string are comma-separated. Then people. Add count=true and limit=1 as query parameters. No records are spent.

curl -s -X POST "https://api.leadocean.io/v1/people/search?count=true&limit=1" \
  -H "x-api-key: $LEADOCEAN_API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "hqCountry": ["US"],
    "employeeRange": ["51-200", "201-500"],
    "technologies": ["HubSpot"],
    "jobFunction": ["Sales & Business Development"],
    "jobLevel": ["VP"],
    "emailStatus": ["verified", "catch_all_valid", "catch_all"]
  }' | jq '.meta | {total, totalIsExact, distinctCompanies}'

count=true goes in the query string. The JSON body rejects unknown fields. Enum values are exact strings, so look them up with GET /v1/enums/{name} before you guess.

Step 3: Test the data layer on five rows

Spend 5 records, not 500. Remove count, set limit to 5, and read what comes back.

curl -s -X POST "https://api.leadocean.io/v1/people/search" \
  -H "x-api-key: $LEADOCEAN_API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "limit": 5,
    "hqCountry": ["US"],
    "employeeRange": ["51-200", "201-500"],
    "technologies": ["HubSpot"],
    "jobFunction": ["Sales & Business Development"],
    "jobLevel": ["VP"],
    "emailStatus": ["verified", "catch_all_valid", "catch_all"]
  }' | jq '.data[] | {has_email: .contact_data.has_email, email_status: .contact_data.email_status}'

Search rows carry flags, not addresses. You see whether a deliverable email exists before you reveal it. Check each row against the person's real profile. If three of five are wrong, change the data source, not the sequencer.

Step 4: Enrich at the point of use

Do not enrich the whole database into your CRM. Enrich the person when a rep or an agent needs the address. POST /v1/people/enrich takes exactly one of linkedin_url, email, phone or person_id. Set reveal_email to true to get the address.

curl -s -X POST "https://api.leadocean.io/v1/people/enrich" \
  -H "x-api-key: $LEADOCEAN_API_KEY" \
  -H "content-type: application/json" \
  -d '{"person_id": "<person_id from a search row>", "reveal_email": true}'

A person we hold nothing for returns 404 and costs nothing. Since 17 October 2026 the response also carries career history: past roles, education and certifications, each when held. Past employers carry no company_id yet.

Step 5: Export to the sending layer

Pick the list you tested and export it. POST /v1/exports takes the same filters inside a filters object, plus a row limit. One export runs up to 50,000 rows, and the daily ceiling is 500,000 rows.

curl -s -X POST "https://api.leadocean.io/v1/exports" \
  -H "x-api-key: $LEADOCEAN_API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "name": "HubSpot shops, VP Sales, US, 51-500",
    "filters": {
      "hqCountry": ["US"],
      "employeeRange": ["51-200", "201-500"],
      "technologies": ["HubSpot"],
      "jobFunction": ["Sales & Business Development"],
      "jobLevel": ["VP"],
      "emailStatus": ["verified", "catch_all_valid", "catch_all"]
    },
    "limit": 100,
    "preset": "full"
  }'

The API answers 202 with an export id. Poll GET /v1/exports/{id} until status is done, then read the signed downloadUrl. This export spends at most 100 records. Bulk CSV export is not on the Free plan, so a free key reaches the same step through the Exports page in the LeadOcean app or a smaller search pull.

Check email_status on every column before the import. Send to verified and catch_all_valid first, and treat catch_all as a smaller test batch. LeadOcean gives no refund for bounces, so the status column is your only protection.

Import the CSV into your CRM, then into the sequencer. That is the whole orchestration layer for a team of one. Add n8n or a script once you repeat this weekly.

Step 6: Give an agent the same data

If an AI client belongs in the orchestration layer, connect it to the hosted MCP server at https://api.leadocean.io/mcp. It signs in with OAuth, so no key is pasted. The server has 13 tools. Three of the free ones do the sizing and lookups above: leadocean_count_leads, leadocean_count_companies and leadocean_list_enum_values.

Start every agent session with a count. A prompt that works:

Use LeadOcean. Count mailable VPs of Sales at US companies with 51-500
employees that run HubSpot. Do not search yet. Tell me the total and
how many distinct companies it spans.

For one account at a time, the tech stack qualifier skill checks a company against a technology rule. To see what a site runs without an account, use the free tech stack checker.

What you get

These numbers come from the free MCP tools on 2026-10-21. Filters: US headquarters, 51-200 and 201-500 employees, technologies: HubSpot.

Both calls reported credits: 0. The people count covers verified, catch_all_valid and catch_all emails only. The gap between 10,023 and 29,167 is the people you cannot mail yet.

Read the stack decision from the numbers. About 5,000 reachable accounts is too many for a spreadsheet and enough to justify a sequencer. A list of 400 would not be.

A search row looks like this. This is the response shape from the API reference, with placeholder values.

{
  "success": true,
  "data": [
    {
      "profile_data": { "...": "identity and headline fields" },
      "contact_data": {
        "has_email": true,
        "has_phone": false,
        "email_status": "verified",
        "email_type": "work"
      },
      "meta": { "...": "person_id and source fields" }
    }
  ],
  "meta": { "count": 1, "limit": 5, "credits": 1, "nextCursor": "<opaque string>" }
}

One search pages to 10,000 rows. If meta.depthCapped is true, narrow the filters or use an export.

Troubleshooting

FAQ