To find lookalike companies, start with your best past and current clients. Write down what they have in common, then search for other companies that share those traits. The traits that matter are industry, headcount, geography, the title of the person who signed, and signs of current activity such as recent LinkedIn posts or an open role.
The order matters more than the tool. My rule is short: look first at who already paid you, then find the companies that resemble them.
Most list building runs the other way. Someone opens a database, ticks filters that sound sensible and exports thousands of contacts. That is general B2B lead list building, and it has its uses. This guide is narrower. It starts from a handful of clients who already paid you and works outward to the companies that look like them.
What is a lookalike company in B2B?
A lookalike company is a business that matches the buying pattern of your best clients. The same kind of person signs. The company is a similar size. It sits in a region you can serve. And it has a similar reason to buy. A shared industry label is the weakest part of the match.
Two firms with the same industry code can sell to different markets, run very different teams and buy through different people. One of them may look exactly like your best client. The other may never need you.
This is different from ideal customer profile work. A profile describes a type of company in general terms. A lookalike list names real companies and real people you can write to next week.
The output is a spreadsheet. Each row holds a company, a named decision-maker with a matching title, and a column saying which seed client the row resembles. If you cannot name that client for a row, the row does not belong on the list.
Which clients should you use as your seed list?
The seed list is the small group of clients the whole search copies. Pick it badly and every filter after it copies the wrong thing.
Start with everyone you have worked with, past and current. Then cut hard. A client stays in the seed only if it passes all 5 of these checks:
- It was profitable once you count the hours your team really spent.
- It bought your core offer, rather than a side project you took on once.
- It is the kind of work you want more of next year.
- You know who signed and what they were trying to fix.
- It came through a route you could repeat, rather than a personal favor.
Most companies have one client that arrived by a strange route and paid well. Put it in the seed, and your filters start chasing a market that exists in that one account.
I keep target lists of ideal clients only, and that discipline starts here, before any tool is open. A weak seed produces a list nobody can explain.
Skip this step and you get the usual pattern. Founders connect with random people on LinkedIn, wait for referrals, book zero sales calls a month and conclude the channel does not work. Usually the list was the problem.
How do you turn those clients into matching filters?
Write down what the companies share
Put the seed clients in rows. Add columns for industry, headcount, country and region, and the situation each client was in when it bought, for example a new site, a hiring push or a failed supplier. Fill each cell from what you know about the client. A database estimate does not count.
Add one more column for how each client describes its own work. Those words often become company keyword filters later, and they catch firms that an industry dropdown files in some odd place.
Find the person who actually signed
Company matching is half the job. Go back over each past deal and name the person who made the call. Write the exact title from their LinkedIn profile or the signed order, and ignore the title you assumed they had.
Use those titles on the new list. Every niche has its own names for the buyer, and a generic "manager" title pulls in many people who cannot approve anything. I strip Assistant, Manager and Associate from a target list unless the niche really uses them for the buyer.
If someone else replied first and passed you to the signer, note both titles. A lookalike account needs the same kind of person reachable inside it. Otherwise it only looks similar on paper.
Apply the filters in this order
My team builds targeted lists in this sequence:
- Job title, copied from the people who signed for your seed clients
- Geography
- Connection degree (people who share a connection with you), 2nd-degree first when LinkedIn is the channel
- Industry, limited to the relevant ones
- Company headcount, set by your targeting
- Activity, limited to people active on LinkedIn within the past month
- Extra signals, such as an open vacancy for the role you serve or keywords in the company description
Title comes first because the person is the scarcest part of the match. Industry and headcount shape the field, but the title decides whether anyone on the list can buy. I keep industry to the relevant ones and let headcount follow the targeting, never a default range.
What does a good lookalike filter stack look like?
Each filter should trace back to something your seed clients actually share.
| Seed client trait | Filter to apply | Why it matters |
|---|---|---|
| Most signers share a title | Exact job titles from past deals | Every row has someone who can buy |
| Clients sit in 1 or 2 countries | Geography | Keeps the list where you can deliver and sell in your language |
| Clients cluster in a few industries | Relevant industries only | Removes companies with no use for the offer |
| Clients fall in one size band | Company headcount | Matches budget and how decisions get made |
| Signers were active on LinkedIn | A LinkedIn post within the past month | Raises the chance a LinkedIn message gets read |
| Clients were hiring for a related role when they bought | Open vacancy for that role | Points to a current need |
| Clients describe their work in specific terms | Company keywords | Catches firms that an industry code files elsewhere |
Keep a filter when it describes most of your seed clients. Drop it when it describes only 1 of them, because then you are copying an accident again.
Start the search with title, geography, industry and headcount. Add activity and hiring signals only after that base list exists. On a small base, a hiring filter can wipe out almost every account, and an empty list teaches you nothing.
You need volume as well as quality. Each row should look like a client, and the list should be wide enough to keep outreach going for months.
If 7 filters on day 1 return a few dozen people, one of those filters probably came from a single client. Remove filters one at a time, weakest first, until the list is big enough to work and still resembles the seed.
How do you build a lookalike account list step by step?
- Write the seed list: the best-fit past and current clients, their shared traits and the titles of the people who signed.
- Turn the shared traits into filters, in the order above.
- Pull matching companies and add a "looks like" column with the name of the seed client each one resembles.
- Find the named person with the matching title at each company. An account with no person is not on the list yet.
- Validate every row against the seed pattern.
- Remove current clients, open deals and duplicates.
- Check the data: the person is still in the role, the email is valid and the company is still trading.
I run step 3 in Sales Navigator, UpLead or a similar tool, working only from what the past and current clients showed.
Step 5 is the one most teams skip. Check each account against the seed on industry, headcount, geography, signer title and buying situation. If it matches on only one or two, remove it. A company that matches on industry and country alone is a guess.
Copy cannot fix a bad list. So no message gets written until every row has passed.
How do you adapt the method for industrial and construction suppliers?
Industrial and construction suppliers are the companies we focus on most, and this method suits them. Our page on outbound for industrial and manufacturing companies covers the wider picture. Many have a small client base, large orders and a market full of companies that look alike in a registry but buy very differently.
The buying side is rarely one person either. Gartner's 2025 survey of 632 B2B buyers found buying groups range from 5 to 16 people across as many as 4 functions. So copy the titles that signed at your seed clients, not just the company type.
So for a supplier into construction, civil engineering or industrial maintenance, match on the job site before the industry label. Ask which of your clients run sites, at what scale and in which countries, and build the filters from those answers.
Expect several people per account. A construction purchase can involve someone who specifies the product, someone in health and safety, someone in procurement and a project lead. Gordian notes that the general contractor typically oversees procurement and often hands tasks to subcontractors who already have supplier relationships. On smaller projects, the same source adds, one person is usually responsible for procurement.
Take a 40-person manufacturer of temporary site fencing selling to contractors in the UK and Ireland. On a big site, it may need the general contractor's procurement lead and the subcontractor who installs the product. On a small job, one buyer may decide alone. Map 2 or 3 of those titles per account from your own past deals, and plan for a sales cycle of 6, 9 or 12 months before the first order, which is common in B2B.
A UK supplier does not have to stop at the UK. Choose countries where you can deliver, support the product and meet local standards. I would take accounts with the right sites and the right titles over a longer list that only shares a code.
What tools and data sources should you use?
I would rather build a list in Sales Navigator, Apollo, UpLead, ZoomInfo or RocketReach than work from a scraped list. These tools let you filter by company and by person, which is what the method needs.
The tools run the search. They do not know which of your clients matter. A tool returns whatever matches the filters, so the seed list still has to supply those filters.
Classification codes help fill the industry column. In the US, NAICS is the standard that federal statistical agencies use to classify business establishments, and the Census Bureau site lets you search it by keyword or by a 2 to 6 digit code. NAICS replaced the older SIC system in the US in 1997.
In the UK, Companies House asks companies to state their nature of business using a condensed set of SIC codes drawn from the full ONS list. You can look up what a seed client filed and search for other companies under the same code.
Treat either code as a first cut. A code tells you what a company registered as its main activity. It tells you nothing about who signs, how many sites the company runs or whether a project is live.
What should you do after the list is built?
The list is the input, and two checks come before outreach. First, run a quality audit on the finished list so bad rows never reach a sequence. Then qualify each lead before it reaches your calendar, so the calls you book are with people who match the seed.
When you write the first message, sell the meeting rather than your services. If you would rather have a senior team run the outreach, that is what our cold email agency work covers, and our appointment setting service books the sales calls with the people who reply.
A good lookalike list makes that easier. Each row already tells you which client the company resembles, which title you are writing to and why the account fits.
Use that as your test. Open a random row and try to write the first 2 lines of the message from that row alone. If you can, the list is ready. If you end up writing something you could send to anyone, go back to the seed.
Got your best clients on a list and want to turn it into matching accounts? Book a call and I will walk you through my method.
Questions founders ask about lookalike lists
How do you find lookalike companies in Apollo?
Use the company lookalikes filter in the search and enter a few reference accounts. Apollo's documentation says the filter accepts up to 5 companies at once, so pick five that resemble each other closely.
On a saved account, Apollo also offers a widget that shows companies matching that account's size, industry, technologies and growth signals. Treat that output as a company shortlist. You still have to add the people who approve purchases yourself.
Do lookalike audiences still work for B2B advertising?
They still exist, though the feature is changing. Google says Lookalike segments in Demand Gen campaigns are moving from fixed similarity thresholds to audience suggestions, and the old 100 user minimum for the uploaded list no longer applies. That lowers the bar for a firm with a small customer base.
My team runs no paid ads, so I would never build a pipeline on ad lookalikes. At most they warm up a market while a researched outreach list does the real work. Ad audiences do not give you named contacts you can write to.
Is there a free way to find lookalike companies?
Yes, for small volumes. nRev says every run of its similar companies finder is free, while bulk export and automation need a paid plan. A manual route costs nothing either. Copy the exact phrases your reference clients use on their websites, search for them, and scan trade association member pages and event exhibitor lists. Free methods suit testing one segment. They get slow once you need hundreds of verified contacts.
How many clients do you need before lookalike matching works?
You need enough clients to show a pattern that repeats across more than one account. With a single strong client, you are copying one story and its quirks. Salesforce warns that assuming what makes an ideal customer without proof can cost you, and points you to real customer data in your CRM instead.
If your base is thin, run a small first batch, see which matches reply, and add each new signed customer to the reference group.
Increasing the volume is the last step, not the first. Hold back until two or more customers clearly share a buyer type and a size band.
What if your best clients are spread across different industries?
Build a separate list for each cluster instead of one blended search. Group your customers by what they bought and who approved it. Each group then gets its own filters and message, and you measure its results separately. If a group holds only one account, park it until a second similar customer appears. Blending unrelated clusters produces filters so loose that they match almost any company.
Should your clients' direct competitors go on the list?
Often yes, because rivals tend to share the same need.
Check two things first. Does your contract with the current customer include exclusivity or confidentiality terms? Would working for a rival unsettle an account you rely on? Never name the existing customer in outreach to its competitor without written permission. If a rival matters to one of your most important accounts, keep it off the list until you have checked.