How to Overcome Firmographic Data Accuracy Challenges (With the LeadOcean API)

TL;DR

• Firmographic errors come from four places: estimated fields, missing values, loose industry labels and stale records. Each has a check you can script.
• The whole run takes 30 to 45 minutes and spends about 10 records. The sizing steps are free.
• The rule that fixes most of it: filter on brackets and labels you have verified, then sample-check before you commit a list.

Prerequisites

• A LeadOcean account. The Free plan gives 1,000 records for life, no card.
• An API key in your shell as $LEADOCEAN_API_KEY.
• curl and jq.
• Optional: the LeadOcean MCP server at https://api.leadocean.io/mcp, to run the free count tools from an agent.

Step 1: Size the segment before you trust it

Start with a free count. It shows how big the segment is and how the data splits, and it spends nothing.

curl -s -G "https://api.leadocean.io/v1/companies/count" \
  -H "x-api-key: $LEADOCEAN_API_KEY" \
  --data-urlencode "industry=Software Development" \
  --data-urlencode "hqCountry=US" \
  --data-urlencode "employeeRange=51-200" | jq '.data.total'

The response also carries breakdowns by HQ state and NAICS code. Read them. A segment packed into one state is a different market from one spread evenly. A long tail of odd NAICS codes tells you the industry label is loose (Step 3).

Step 2: Filter on brackets, not exact numbers

Total headcount is also a blunt field. A 400-person company can be mostly engineers or mostly support, and the number does not say which. Pair the size bracket with a people filter such as jobFunction when the mix matters to your offer.

Exact headcount is the field teams over-trust. Use employeeRange for sizing, and use minEmployees and maxEmployees only when you need a cut inside a bracket.

The two filters do not match the same companies. minEmployees and maxEmployees return only records that hold an exact figure. A company with a bracket but no exact count is dropped.

Run both and compare:

curl -s -G "https://api.leadocean.io/v1/companies/count" \
  -H "x-api-key: $LEADOCEAN_API_KEY" \
  --data-urlencode "industry=Software Development" \
  --data-urlencode "hqCountry=US" \
  --data-urlencode "employeeRange=51-200" | jq '.data.total'

curl -s -G "https://api.leadocean.io/v1/companies/count" \
  -H "x-api-key: $LEADOCEAN_API_KEY" \
  --data-urlencode "industry=Software Development" \
  --data-urlencode "hqCountry=US" \
  --data-urlencode "minEmployees=51" \
  --data-urlencode "maxEmployees=200" | jq '.data.total'

The gap is the number of companies your exact-number filter silently hides. The same rule applies to minRevenue and minFounded: the filter only matches records where we hold the value.

Step 3: Check the industry label against its neighbours

Industry labels overlap. Resolve the enum before you filter, then decide which neighbours belong in your segment.

curl -s "https://api.leadocean.io/v1/enums/industry?q=software" \
  -H "x-api-key: $LEADOCEAN_API_KEY" | jq '.data.values'

Filter values are exact strings. A wrong value returns an empty result, not an error (see Troubleshooting).

To go finer than a label, add keywords, which matches the company description, specialties and tags. Use excludeKeywords to drop false fits.

curl -s -X POST "https://api.leadocean.io/v1/companies/search?count=true" \
  -H "x-api-key: $LEADOCEAN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "industry": ["Software Development"],
    "hqCountry": ["US"],
    "employeeRange": ["51-200"],
    "keywords": ["payments"],
    "limit": 1
  }'

count=true is a query parameter and the call spends no records. That is how you test a narrower definition of your segment before you pull a row.

Step 4: Sample-check a few records against your own data

Never validate a field by trusting it. Pull a small sample and compare it with what you know. Ten records is enough to see a pattern.

curl -s -X POST "https://api.leadocean.io/v1/companies/search" \
  -H "x-api-key: $LEADOCEAN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "industry": ["Software Development"],
    "hqCountry": ["US"],
    "employeeRange": ["51-200"],
    "limit": 10
  }' > sample.json

For accounts already in your CRM, enrich by domain and diff the fields. A company that is not in the data returns a 404 and costs nothing.

curl -s -X POST "https://api.leadocean.io/v1/companies/enrich" \
  -H "x-api-key: $LEADOCEAN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"domain": "acme.com"}' | jq '{name: .data.company_data.company_name, fetched_at: .data.meta.fetched_at}'

Score the sample by field and write the result down.

The numbers in that table are an example of how to log your own run, not a LeadOcean result. If headcount is wrong on 3 of 10 and HQ country on 0 of 10, trust country and treat headcount as a bracket. Private companies do not publish revenue, so treat that field as a rough band and never as a hard gate.

Two habits make this step worth the ten records. First, sample the same segment again after each refresh, so you see drift and not a one-off result. Second, keep the sample file with a date in its name. When a rep says a list was wrong, you can check what the data said on the day you pulled it.

Step 5: Re-enrich on a schedule

Records age. The dataset is refreshed monthly and each record carries its own fetched_at date. Use that date to decide what to refresh.

jq -r '.[] | select(.last_enriched < "2026-07-23") | .domain' crm_export.json |
while read d; do
  curl -s -X POST "https://api.leadocean.io/v1/companies/enrich" \
    -H "x-api-key: $LEADOCEAN_API_KEY" \
    -H "Content-Type: application/json" \
    -d "{\"domain\": \"$d\"}"
  sleep 0.25
done

The sleep keeps you under 5 requests a second on Free. On Agency the limit is 200 a second. For a full list refresh, POST /v1/exports takes up to 50,000 rows a request and 500,000 a day. The Exports page in the app (app.leadocean.io) shows the record price before you download. Bulk CSV export is not on Free.

What you get

On 2026-10-21 the free leadocean_count_companies tool returned these totals for US companies in Software Development.

The NAICS breakdown on the bracket count shows how loose the label is. The top code, 541511, covers 5,890 of the 9,261. The rest spread over more than 15 other codes, down to single digits per code.

The free enum lookup for "software" on the same date returned seven industry values. They were Computer Software, Data Security Software Products, Desktop Computing Software Products, Embedded Software Products, IT System Custom Software Development, Mobile Computing Software Products and Software Development. Filtering on one label leaves the other six out.

A company enrich response has this shape (response shape from the API reference, placeholder values, fields abridged):

{
  "success": true,
  "data": {
    "company_data": {
      "company_name": "Acme",
      "company_domain": "acme.com"
    },
    "company_metrics": {},
    "meta": { "fetched_at": "2026-10-01", "schema": "leadocean.company.v1" }
  },
  "meta": { "credits": 1 }
}

Run Step 1 again after every refresh. If the segment total moves by a large amount in a month, find out why before you launch a campaign on it. A big move usually means the label, a bracket or a country changed, not that the market did.

Troubleshooting

FAQ