How to Find More Companies Like Your Best Customers: 5 Steps From Public Registers (2027)
To find more companies like your best customers, read 10 in a register, keep the traits 7 share, test them on 3 that left, then filter on the rest.
Contents7 sections
Companies House lets anyone filter UK companies by SIC code, company type, registered office address and incorporation date, and download the first 5,000 results as a CSV. Which code, type and age to enter is left to you. To find more companies like your best customers, read 10 of them in the same register and search on the traits most of them share.
This is for the owner or export manager of a company with 10 to 30 customers and no customer database to feed a lookalike tool. After five steps you'll have a few tested filters and a way to tell which matches deserve a first email.
How we put this together. In November 2026 we read the lookalike guides that rank for this question and each register's documentation. The 7-of-10 rule, the test against 3 customers and the worked example are our own.
Step 1: Which customers count as your best?
Choose 10 by one measure you already trust, and write the measure down before you choose: revenue over the last three years, margin after the work each customer takes, or how often they reorder. Writing it first keeps the list from turning into the customers you like most. If you have 30, take the top 10.
Three choices make the list better:
- One measure for all 10. A big one-off order and a steady reorderer are two kinds of customer, with few shared traits.
- Customers the market brought you. One that came through a family contact or a single tender tells you little about who else would buy. Leave it out, or mark it.
- Three that didn't work out. Add 3 companies that left, or took quotes and never bought: your misfits. They're the test in Step 3.
Keep the 10 and the 3 in one sheet, one row each. Step 2 adds the columns.
Step 2: What does each best customer look like in a register?
Look up each of the 10 in its country's company register and note six fields: industry code, size band, region, age, legal form and what it sells. The register gives the first five. The sixth comes from the company's website, because an industry code only records what the company registered as its main activity.
| Field | Where to read it | Write it as |
|---|---|---|
| Industry code | SIC in the UK, SNI in Sweden, NAF in France, industry in CVR | The full code, with its version |
| Size band | Employee bands in Sweden, Norway and France; monthly headcount in CVR; the website in the UK | The register's own band, such as 20–49 |
| Region | Registered office address, municipality, county or postcode | The level you'd search by |
| Age | Incorporation or creation date | The year |
| Legal form | Company type or legal form | As the register writes it |
| What it sells | The company's own website | One line, in your words |
Denmark's CVR lists each company's industry and any secondary industries, its number of employees where applicable, and its owners and managers. The free CVR company page shows headcount month by month, so you can see whether a customer was growing when it first bought.
Codes differ by country and change between versions, so write each with its version; how to build a company list by industry code sets them out country by country.
Bring the traits your best customers share to a demo, and we'll show you how HAL, gethal.ai's AI sales agent, sets up its ideal customer profile with you. Book a demo
Step 3: Which traits are shared by most of your 10 best customers?
Count, field by field, how many of the 10 share each value. A value at least 7 share becomes a filter; anything fewer stays a note. Then hold each filter against your 3 misfits. If 2 or all 3 of them share it too, it doesn't separate buyers from the rest, so drop it.
Shared by most is our rule for a seed of 10. A filter on a trait 5 of your 10 share would have missed the other 5 of your best customers, and it misses their twins in the register the same way. At 7 it misses 3, a loss small enough to see and decide on.
Count at the level you'd search on:
- Codes at the 4-digit class first; if no class reaches 7, try the level above.
- Size in the register's own bands, such as Statistics Sweden's 10–19 and 20–49. Two neighbouring bands you'd search together count as one.
- Region at the level the register filters by: county, municipality or postcode.
- Age as a cut-off year.
Keep the industry code as the base of every search even when the misfits share it, since it defines the market. Every other trait faces the test.
In the invented example below, a manufacturer of testing equipment reads its 10 best customers and 3 misfits:
| Trait (invented example) | Of the 10 best | Of the 3 misfits | Result |
|---|---|---|---|
| One 4-digit industry code | 9 | 3 | Base of the search |
| 20–49 employees | 7 | 1 | Filter |
| Founded before 2005 | 8 | 3 | Dropped: the misfits have it too |
| Limited company | 10 | 3 | Dropped: everyone has it |
| Head office in one county | 4 | 0 | A note |
| Sells to hospitals (website) | 7 | 0 | A website check in Step 5 |
Two register filters survive, code and size band, plus one website check.
Step 4: Where can you search for companies with that fingerprint?
Enter the surviving filters in a free register first. Companies House, Statistics Sweden, Brønnøysund and France's Export Sirene each filter by industry code, and all but Companies House by size band too. Export the result, and pay for a lookalike tool only where a register can't filter by code, as in Germany.
| Country | Where | Filters | Cost and limit |
|---|---|---|---|
| UK | Companies House advanced search | SIC code, status, company type, registered office address, incorporation and dissolved dates | Free; the first 5,000 results as a CSV |
| Sweden | Statistics Sweden's business-register API | Main SNI code, municipality, county, employee size class, turnover class, legal form | Free with a personal API key; 5,000 rows a call |
| Norway | Brønnøysund's open API | Industry code, employee range, municipality, legal form | Free; 10,000 a query, or the whole register as a spreadsheet |
| France | Annuaire des Entreprises, Export Sirene | NAF sub-class, headcount range, location, legal form, creation date | Free CSV, no account |
All four cost nothing.
Two limits to know before you export:
- Size bands have gaps. France's export page says about half of establishments have no headcount band, and Norway's register gives none for entities with 0 to 4 employees. A size filter drops them, so run one export without it too.
- The UK has no size filter. Filter Companies House by code, type and age, and judge size on each website in Step 5.
For a country that isn't here, our guide to company registers in Europe says what its register gives you and whether it's free.
Lookalike tools run the same match inside their own database, from a seed list you upload. Before you pay, ask which fields a tool matches on, and whether its industry and size come from a register or its own estimate.
Show us the filters you'd search on, and we'll show you how HAL checks the companies it finds against CVR, proff.dk and LinkedIn. Book a demo
Step 5: How do you tell a lookalike from a company that needs you?
A register match tells you a company resembles your customers. Need shows elsewhere: in a score on public checks, in a recent dated event such as a new hire or a board change, and in a short read of the company itself. Write first to the matches that show need now, and look at the rest again in three months.
- Score the list. The free target account scorecard (Excel) runs nine checks from public data, with weights you set, and sorts companies into three tiers.
- Look for a dated event. A job ad for the work you'd help with, a new director or a won tender says the company is changing now; free buying signals shows where to find each.
- Research the top of the list. Five checks in about 15 minutes per company, with the company research checklist.
- Feed the result back. When a match becomes a customer, add it to the sheet and recount. When one turns out wrong, add it to the misfits.
Check the email rules for each market with your own lawyer before the first email goes out.
Use HAL when you'd rather hand over the finding and checking: HAL, gethal.ai's AI sales agent, finds companies that fit the profile set up with you and checks them against CVR, proff.dk and LinkedIn. You choose who gets an email.
Where lookalike lists usually fail
A lookalike list breaks at the seed or at the end: a skewed set of best customers gives a skewed fingerprint, and a company that fits may have no reason to buy this year. The table sums up the five steps.
| Step | Decision | Tool or source | Cost | Where it fails |
|---|---|---|---|---|
| 1. Best customers | Which 10, by which measure | Your own sales records | Free | One customer dominates the list |
| 2. Read the register | Six fields per customer | The register; the website | Free | A customer sits under the wrong code |
| 3. Shared by most | Which traits become filters | Your sheet; the 3 misfits | Free | A trait fewer than 7 share slips in |
| 4. Search | Where to enter the filters | A register; a lookalike tool | Free; a tool costs money | Size filters drop companies with no band |
| 5. Need | Who gets a first email | Scorecard, signals, 15-minute check | Free; your time | A good fit with no reason to buy now |
One big customer as the seed
If one customer brings in a large share of your revenue, it pulls the fingerprint toward its own size and region, and companies like it are few. Count it as one of the 10 and never take a trait from it alone.
Traits a register can't show
What a company buys, which machines it runs, who decides and where it exports sit in no register. Leave them for Step 5 and the company's own site.
A market with 200 companies
When your code and country hold a few hundred companies, read them all. A filter that cuts 200 to 30 can drop buyers who differ from your 10 in one field. In a market that small, score every company and compare what each route to them costs per right buyer, as in outbound agency, AI sales agent or doing it yourself.
Tell us about your best customer, and see the outreach brief and suggested first email HAL writes for a company that fits. Book a demo
Who this is for
- The owner who still sells: Steps 1 and 3, which need someone who knows why each customer buys. If most of your new buyers still come from trade fairs, our weekly routine for finding buyers between fairs pairs with this one.
- The export manager opening a new country: that country's row in Step 4, then Step 5 before anyone writes. Codes and size bands differ abroad, so recheck the fingerprint there.
- Whoever keeps the customer sheet: the six fields of Step 2 as columns, the misfits as rows of their own, and a recount each time a customer is won or lost.
FAQ
What are lookalike companies in B2B?
Companies that share the traits of your best customers: the same industry, a similar size, region, age or legal form. Lookalike tools match your customer list against their own database; a public register lets you run the same match yourself. Whether a lookalike needs what you sell is a separate check.
How many customers do I need to find lookalike companies?
Ten is enough if you read them yourself rather than handing them to a model. Keep a trait as a filter only if at least 7 of the 10 share it, and test it against 3 customers who left or never bought. With fewer than 10, use every customer you have and treat the result as a first cut.
How do I find similar companies for free?
Read your best customers in the company register, note the traits most of them share, and enter those as filters in a free register search. Companies House exports up to 5,000 results by SIC code and Brønnøysund up to 10,000 a query; France's Export Sirene and Statistics Sweden's API, with a free key, filter by code and size band.
Do lookalike tools work when I only have a few hundred possible buyers?
In a market of a few hundred companies you rarely need one: export every company in your industry code and country from the free register and read them all. A filter that cuts it to a few dozen can drop buyers who differ from your best customers in one field.
How do I find companies like my customers in Germany or France?
In France, Export Sirene on the Annuaire des Entreprises filters by NAF sub-class, headcount range, location, legal form and creation date, as a free CSV with no account. Germany's Handelsregister search has no industry field, so filter there by region and legal form, or buy a list by industry code and confirm each company in the register.
Start with your 10 best customers
Write down your measure, pick the 10 and the 3 misfits, and read them in the register this week. To see how HAL finds and checks companies against a profile set up with you, book a demo, or see what it costs.
Keep reading: Finding buyers between trade fairs · Target account scorecard · Company list by industry code


