How to Build a RevOps Tech Stack That Scales With You (Data Layer First)

A RevOps stack scales when one layer owns the data and every other tool reads from it. Build that layer first. CRM, sequencer and warehouse then become swappable parts.

This guide builds the data layer against the LeadOcean API. Each step has a request you can run today.

TL;DR

• Put one data source behind your CRM, sequencer and warehouse. Size, search, enrich and export through it. Swapping a downstream tool then costs an afternoon, not a migration.
• It takes about 45 minutes. Steps 1 to 4 spend about 30 records on the free plan. The 100-row export test in Step 5 needs Solo or above.
• Count audiences for free before you spend anything, and let the API errors tell you when you hit a ceiling.

Prerequisites

• A LeadOcean account on the free plan. Sign up at app.leadocean.io/register. No card needed.
• An API key from the app, exported as $LEADOCEAN_API_KEY.
• curl and jq. Any HTTP client works the same way.
• A list of the tools in your stack today: CRM, sequencer, warehouse, enrichment. A text file is enough.

Step 1: Write the stack contract

Start with a contract: which layer owns which record, and which direction data flows. Most stacks break because two tools both think they own the contact.

layers:
  data:        # owns person and company facts, email_status, fetched_at
    source: LeadOcean API (https://api.leadocean.io)
    writes_to: [warehouse]
  warehouse:   # owns history, scoring inputs, dedupe keys
    key: person_id
    writes_to: [crm]
  crm:         # owns deals, owners, stages. Never edits data-layer fields
    reads_from: [warehouse]
  sequencer:   # owns sends and replies
    reads_from: [crm]
rules:
  - one key per person: person_id
  - send only email_status in [verified, catch_all_valid]
  - nothing writes back upstream

Keep the file in the repo next to your sync jobs. When a tool changes, you edit one line and see what else it touches.

The person_id key matters. LeadOcean's enrich endpoint takes exactly one of linkedin_url, email, phone or person_id. Use person_id as the join key and you never match on a name.

Step 2: Size your audience for free

Count before you pull. A count costs no records, so size every segment you plan to feed the stack. Do this for each territory and persona, and you know the volume your CRM and sequencer must handle before you buy seats for either.

Send count=true as a query parameter, not in the JSON body. The body is flat JSON and rejects unknown fields.

curl -s -X POST "https://api.leadocean.io/v1/people/search?count=true" \
  -H "x-api-key: $LEADOCEAN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "jobFunction": ["Operations"],
    "jobLevel": ["VP", "Director"],
    "technologyCategories": ["CRM", "Marketing automation"],
    "employeeRange": ["51-200", "201-500"],
    "country": ["US"],
    "limit": 1
  }' | jq '.meta'

Enum values are exact strings. Resolve them with GET /v1/enums/{name} or the free list_enum_values MCP tool before you guess. technologyCategories searches the roughly 7.8M companies whose sites we crawled, not the whole database.

Step 3: Pull the people for one segment

Pick one segment from Step 2 and pull a small page. You are testing the shape of the data and your mapping into the warehouse, not building the full list yet.

Search returns people as a thin projection, with no contact points. Each person returned counts one record. Set limit between 1 and 100 and page with the cursor.

curl -s -X POST "https://api.leadocean.io/v1/people/search" \
  -H "x-api-key: $LEADOCEAN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "title": ["Head of Revenue Operations", "VP Revenue Operations"],
    "emailStatus": ["verified", "catch_all_valid"],
    "emailType": ["work"],
    "limit": 25
  }' > page1.json

jq '.meta | {count, credits, nextCursor}' page1.json

To read the next page, send the same body plus "cursor": "<meta.nextCursor>". One search pages through at most 10,000 rows. Past that no cursor is issued and meta.depthCapped is true. Narrow the filters and run smaller searches.

Step 4: Enrich on demand, by person_id

Search gives you the list. Enrichment is the step you run per record, when a rep needs the email. Pass the id from search and set reveal_email to true.

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_SEARCH", "reveal_email": true}'

Revealing the email costs no extra record. A person we hold nothing for returns a 404 and costs nothing. There is no name plus domain lookup, so search by domain first, then enrich by person_id.

For phones, call GET /v2/people/phone. The v2 contact endpoints cost one record per call, including a call that finds nothing.

Step 5: Export in bulk for the warehouse

Do not page 40,000 rows through search. At 100 rows a page that is 400 calls, and the 10,000-row depth cap stops you anyway. Run an export job: create it, poll it, download the CSV. One export takes up to 50,000 rows and your account can run 500,000 rows a day. Bulk CSV export is not on Free: it is included on Solo ($97 a month) and Agency ($497 a month, 50k rows a file).

curl -s -X POST https://api.leadocean.io/v1/exports \
  -H "x-api-key: $LEADOCEAN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "revops-stack-test",
    "filters": {
      "jobFunction": ["Operations"],
      "jobLevel": ["VP", "Director"],
      "country": ["US"],
      "emailType": ["work"]
    },
    "limit": 100,
    "columns": ["profile_full_name", "current_job_title",
                "current_company_domain", "email_N_address", "email_N_status"]
  }' | jq '.data.id'

The API answers 202 with a job id. Poll GET /v1/exports/{id} until the status is done, then GET /v1/exports/{id}/download. Numbered columns such as email_N_address come back as email_1_address, email_2_address and so on.

Test with a limit of 100 first. An export reserves up to limit records when it starts and bills each row written. The app's Exports page does the same job with a filter builder and shows the record price before you start.

Step 6: Add a cost and rate guard

Read your quota on every response. The X-Quota-* and X-RateLimit-* headers ride on each call, and GET /v1/usage shows the totals.

curl -s -i https://api.leadocean.io/v1/usage \
  -H "x-api-key: $LEADOCEAN_API_KEY" | grep -i -E "^x-(quota|ratelimit)|^HTTP"

Wire two alerts into your stack: one on remaining records, one on any 429. A stack that scales reads its own limits instead of failing in the middle of a sync.

Rate limits are per plan: 5 requests a second on Free, 50 on Solo, 200 on Agency and 10,000 on Platform. Paid plans have no credits. Solo is $97 a month and Agency $497, each with a fair-use pace: past it, the key slows until the window resets. See pricing.

What you get

The Step 2 audience, counted through the free count_leads MCP tool on 2026-10-21:

count_leads
  jobFunction:          Operations
  jobLevel:             VP, Director
  technologyCategories: CRM, Marketing automation
  employeeRange:        51-200, 201-500
  country:              US
  facets:               false

total:        4020
totalIsExact: true
totalAll:     12992
credits:      0
defaults:     counted verified, catch_all_valid and catch_all addresses

That is 4,020 mailable people out of 12,992 matching the other filters. The count spent 0 records. A second count on 2026-10-21, titles "Head of Revenue Operations" or "VP Revenue Operations" worldwide, returned 856 mailable people.

A record-level response has this shape (response shape from the API reference, placeholder values):

{
  "profile_data": { "...": "profile fields" },
  "contact_data": {
    "has_email": true,
    "has_phone": false,
    "email_status": "verified",
    "email_type": "work"
  },
  "meta": { "source": "leadocean", "schema": "leadocean.person.v1" }
}

contact_emails and contact_phones appear only after you set reveal_email to true on enrich. Read email_status before you send. There is no refund or credit-back for bounced emails.

Troubleshooting

FAQ