# Find lookalike accounts from your best customers

> Turn your best customers into a profile of industry, size, location and tech stack, then search for more companies that match it.

Source: https://www.b2benrich.com/guides/lookalike-accounts  
Author: Dag Holmen, B2Benrich · Updated 2026-10-28

Lookalike companies share the traits of your best customers: a similar industry, size, revenue, location and tech stack. If your best customers have a lot in common, the companies that match them are the most likely to buy next.

Ad platforms will build a lookalike audience for you, but they don't show the logic, and you can't take the result anywhere else. Building the list yourself takes a few API calls. You get named companies you can read, explain, hand to sales and load into ads. This guide does it in five steps: pick the seed, enrich it, find the shared traits, search for matches and remove the companies you already have.

## Step 1: Pick the seed

The seed is the customer list everything else copies, so choose it with care.

- **Use your best customers, not all of them.** Best can mean highest revenue, fastest to close, longest retained or most expansion. Pick one definition.
- **Aim for 20 to 50 companies.** With fewer, the patterns are mostly noise. With many more, you start copying customers you'd rather not have more of.
- **Drop the outliers.** One famous logo that bought for unusual reasons will pull the profile the wrong way.

You need one thing per company: its domain.

## Step 2: Enrich the seed

`POST /v1/enrich/companies/domain-to-company` takes a domain and returns `{ status, company_id, data }`, with the firmographics in `data`. This script enriches every seed domain and tallies the most common values for each trait:

```python
import os, requests
from collections import Counter

BASE = "https://api.b2benrich.com"
HEADERS = {"X-Api-Key": os.environ["B2BENRICH_API_KEY"], "Content-Type": "application/json"}

def post(path, body):
    r = requests.post(f"{BASE}{path}", json=body, headers=HEADERS, timeout=30)
    r.raise_for_status()
    return r.json()

seed = ["acme.com", "globex.com", "initech.com"]  # your 20-50 best customers

profiles = []
for domain in seed:
    res = post("/v1/enrich/companies/domain-to-company", {"input": domain})
    if res.get("data"):
        profiles.append({"company_id": res["company_id"], **res["data"]})

for field in ["industry", "employees_range", "revenue_range", "hq_country"]:
    tally = Counter(p[field] for p in profiles if p.get(field))
    print(field, tally.most_common(3))
```

Keep the `company_id` for each customer. The [company enrichment API](https://www.b2benrich.com/guides/company-enrichment-api) guide covers cleaning domains and handling misses.

## Step 3: Find the shared traits

For a 30-company seed, the tally might come out like this. The numbers are made up.

| Trait | Most common values (customers) |
|---|---|
| Industry | Software Development (19), IT Services and IT Consulting (7), Financial Services (4) |
| Employees | 201-500 (14), 51-200 (10), 501-1000 (5) |
| Revenue | $25M-$50M (12), $10M-$25M (9), $50M-$100M (6) |
| HQ country | United States (22), Canada (5), United Kingdom (3) |

For each trait, keep the values that together cover most of the seed and drop the long tail. Here that might be two industries, three size bands, two revenue bands and two countries. That's your lookalike profile.

Before you search, check each kept value against the filter lists from `/v1/search/meta/industry_linkedin`, `/v1/search/meta/employee_count_range` and `/v1/search/meta/revenue_range`. Filters match exact strings, so if an enriched value isn't in the list, map it to the closest one. All values in this guide are placeholders.

## Step 4: Add the tech stack

Tech stack is often the sharpest trait, especially if your product connects to a specific tool. You may know it already: the integrations your customers turn on, or the tool your product replaces.

To check it against the data, use Prospector, because company search has no tech filter. Search `/v1/search/prospector` with the tool in the `technologies` company filter plus your top industry and size band, collect `company_domain` from the results, and see how many seed domains show up. If most do, the tool belongs in the profile. Exact tool names come from `/v1/search/meta/technologies`, and [technographic data](https://www.b2benrich.com/guides/technographic-data) covers the filter in detail.

## Step 5: Search for matches and exclude customers

Without a tech trait, search companies by the profile's firmographics with the [company search API](https://www.b2benrich.com/guides/company-search-api). With one, search Prospector and keep one row per `company_id`. This script takes the first route, loops over the kept values and skips every domain you already have:

```python
def norm(domain):
    return (domain or "").strip().lower().removeprefix("www.")

exclude = {norm(p.get("domain")) for p in profiles}
exclude |= {norm(d) for d in open("crm_domains.txt").read().split()}

# The profile from step 3. Placeholders: use exact values from the meta endpoint.
industries = ["Software Development", "IT Services and IT Consulting"]
sizes = ["51-200", "201-500", "501-1000"]

matches = {}
for industry in industries:
    for size in sizes:
        page = 1
        while True:
            data = post("/v1/search/companies", {
                "industry_linkedin": industry,
                "employee_count_range": size,
                "headquarters_country_name": "United States",
                "page": page,
                "page_size": 100,
            })
            for c in data["records"]:
                d = norm(c.get("domain"))
                if d and d not in exclude:
                    matches[d] = c
            if len(data["records"]) < 100:
                break
            page += 1

print(len(matches), "lookalike companies")
```

Add `revenue_range` the same way if revenue separates your best customers from the rest. The exclusion file should hold more than customers: open opportunities, churned accounts you don't want back, partners and competitors. Normalize domains on both sides, or "www.Acme.com" and "acme.com" won't match.

## Sort by closeness

Not every match is equal. Sort by how many profile traits each company shares, giving the top value in each trait the most weight. A company that matches your main industry, main size band, main country and the key tool goes first. One that matches only on industry goes last.

Then read the top 20 by hand. If they don't look like your customers, the seed or the profile needs work, not the search.

## Put the list to work

- **Outbound.** Find the people who buy at each match. [Buying committee mapping](https://www.b2benrich.com/guides/buying-committee-mapping) covers roles and contact details.
- **Ads.** Upload the matched companies or their buyers as a matched audience. Unlike a platform lookalike, you know exactly who's in it.
- **Account expansion.** When a segment of your target list runs dry, a fresh pull from your newest best customers adds accounts that fit.

Rerun the whole thing when your seed changes, or once a quarter. B2Benrich is unlimited API usage for $500 a month with no credits, so a rerun doesn't change your bill. To fold the matches into a full market map, see [TAM mapping with an API](https://www.b2benrich.com/guides/tam-mapping-api).

## Frequently asked questions

### What are lookalike companies?

Lookalike companies are businesses that share the traits of your best customers, such as industry, employee count, revenue, headquarters location and tech stack. They make a strong next target list because they resemble accounts that already bought.

### How do I find companies similar to my customers?

Enrich your best customers’ domains to get their firmographics, tally the most common industry, size, revenue and country values, then search for companies with those values. Remove your existing customers and open deals by domain before you use the list.

### How many customers do I need for a lookalike list?

Twenty to fifty is a practical range. Fewer and the shared traits are mostly noise. Many more and the profile drifts toward your average customer instead of your best ones.

### Can I include tech stack in a lookalike search?

Yes. Put the tool in the technologies company filter on the Prospector endpoint, along with your firmographic traits, and keep one result per company_id. Get exact tool names from /v1/search/meta/technologies.

### How is this different from a lookalike audience on an ad platform?

An ad platform builds the audience inside its own system and doesn’t show you who is in it. Building lookalike companies with an API gives you a named list you can check, explain and use in outbound, ads and your CRM.
