List building · accounts first, then people, then contacts
B2B list building: how to build a prospect list that doesn't bounce
The short answer
To build a B2B prospect list that doesn't bounce, keep every unproven address out of it. Pull the accounts that match your ideal customer profile, check that each one fits, find the decision maker in each, then add only the work emails and phone numbers returned as verified. List building from a contact export brings in guessed and stale addresses, and those are what bounce.
On this page12 sections
- What is list building in B2B sales?
- Why do prospect lists bounce?
- Should you start a prospect list from companies or from people?
- How do you build a B2B prospect list, step by step?
- Step 1: How do you pull the accounts that match your ICP?
- Step 2: How do you check that each account fits before you search for people?
- Step 3: How do you find the decision maker in each account?
- Step 4: How do you add work emails and phone numbers without guessing?
- How do you check a prospect list for bounces before you send?
- What did this prospect list cost in records?
- How do you run this chain with the BlitzAPI API?
- Sources and method
Most prospect lists are built backwards. Someone exports every contact with the right job title, uploads the file to a sequencer, and finds out what was wrong when the bounces come back. By then, the sending domain has already paid for it.
This guide builds the list the other way around, in the order the data is connected: the accounts first, then the decision maker inside each one, then the contact details, then a check before anything is sent. Every step runs on a real list, built on November 10, 2026, with the number of records each step cost.
What is list building in B2B sales?
List building, or prospect list building, is the part of outbound work that decides who you will contact, before anyone writes a word to them.
- Definition List building also called prospect list building, B2B list building
- Putting together the companies and the people you plan to contact, with a proven way to reach each person, before any outreach starts.
PreciselyList building is the stage of outbound prospecting that turns an ideal customer profile into a prospect list: select the accounts that match the profile, check that each one fits, find the decision maker inside each account, then add the work email and phone number that a data provider returns as verified. It ends where the sequence starts: copy and sending come after.
It sits at the very start of outbound lead generation. Sequences, copy and sending come after, and they can only be as good as the list they run on. B2B prospecting gives that list a particular shape: you sell to companies, but you write to people, so every row ties one person to one account.
What comes out of list building is a prospect list, the file your sequencer reads.
- Definition Prospect list also called lead list, target list
- A table of the people you plan to contact, one row each, with their company, their role and the details you need to reach them.
PreciselyA prospect list holds one row per person to contact: the account (company name, company page, domain), the person (current title, profile URL), the contact points (work email, phone number) and their provenance (which source returned each value, and when). Lead list is often used for the same thing, though strictly a lead has already shown interest and a prospect has not.
Each row has four parts, and each part answers its own question:
- The account: is this a company we want? Its name, its company page, its domain, its size.
- The person: is this who decides? A current job title, and how far down your list of titles you had to go to find it.
- The contact points: can we reach them? A work email and a phone number, each with a found or not found status.
- The provenance: can we still trust it? Which call returned each value, and on which date.
- AThe accountIs this a company we want?
- Company
- Acme Security
- Page
- linkedin.com/company/acme-security
- Domain
- acme-security.example
- Size
- 51-200 · 142 on LinkedIn
- DSThe personIs this who decides?
- Title
- Director of Sales, current job
- Found at
- title level 2 of 3
- The contact pointsCan we reach them?
- Work email
- verified, at the account
- US phone
- found: false, left blank
- The provenanceCan we trust it today?
- Built on
- November 10, 2026
- Each value
- the call that returned it
The fourth part is the one most lists skip. It is also the one that tells you, three weeks later, whether a row can still be sent.
Why do prospect lists bounce?
Because some of the addresses in them were never proven to work. An email bounce is a message that comes back undelivered, and a hard bounce means the mailbox does not exist. Three habits put those addresses in a list:
- Guessing. An address built from a pattern, like firstname.lastname@company.com. It looks right, and nobody asked the mail server.
- Exporting once, sending later. The address was real when the list was pulled. Three months later, the person has left, and the mailbox is gone.
- Sending the unproven. On a catch-all domain, the mail server accepts every address during a standard check, so a basic verifier can’t tell a real mailbox from a missing one. The address is marked unknown, and it goes out anyway.
- Guessedfirstname.lastname@acme-security.exampleBuilt from a pattern. Nobody asked the mail server.ResultUnknown mailbox
- Exported, sent laterreal in March, sent in JuneThe person left in between. The mailbox was removed.ResultMailbox gone
- Unproven, sent anywaycatch-all domain: status unknownThe server accepts every address during a standard check, so that check proves nothing.ResultCan bounce after the send
Three habits, one cause: the address joined the list before anything proved it works.
None of these is fixed in the sequencer. Building the list in order keeps guesses out: start from a contact export, and you take whatever addresses come with it; start from accounts, and every address has to earn its place in a row that already makes sense. Accepting only verified addresses, and sending soon after the list is built, takes care of the other two.
A list doesn’t bounce because of the sequence. It bounces because of how it was built.
Should you start a prospect list from companies or from people?
From companies, when you want one right person in every account. From people, when a single job title describes your buyer everywhere.
A people search with company filters, like BlitzAPI People Search, returns the people who match across many companies in one search, page by page. That is fast, and it is enough when the title is the same in every company. It can also return a dozen people in one account and nobody in the next, and each person returned counts before you have looked at a single company.
| Accounts first | People first | |
|---|---|---|
| Starts from | the companies that match the profile | the people whose title matches |
| Fit is checked | before anyone is searched | afterwards, row by row |
| People per account | as many as you ask for, best title first | whoever matches, from none to dozens |
| Works best for | account-based lists, varied org charts | one clear title across the market |
This guide takes the first route, because it lets you check each account before you pay for its people.
How do you build a B2B prospect list, step by step?
In five steps, each one deciding something the next one depends on:
- Pull the accounts that match your ideal customer profile.
- Check that each account fits, before you look for anyone inside it.
- Find the decision maker in each account that passed.
- Add the contact points: a verified work email, and a phone number where coverage allows.
- Check the list before you send it.
- 1Pull the accountsWhich companies match the profile
- 2Check the fitWhich of them you really want
- 3Find the decision makerWho decides in each account
- 4Add the contact pointsHow to reach that person
- 5Check before you sendWhich rows go to the sequence
A call that finds nothing counts nothing, at every step.
The list behind this guide follows those steps. The brief: a seller who wants to reach the head of sales at US cybersecurity companies with 51 to 200 employees. Before the first call, we stated the worst case: 101 records at most, if every account fit and every person had both an email and a phone number. The sections below show what each step returned.
Step 1: How do you pull the accounts that match your ICP?
Turn each line of your ideal customer profile into a filter, then size the market before you pull it.
Industry, headquarters country and headcount band are the usual first three. Funding stage, ad spend, web traffic and words in the company description narrow it further. Company Search takes all of them in one query, and every page of results also says how many companies match in total.
That total is the cheapest number in list building. Ask for a single result, and the response gives you the size of the whole segment for one record. Our three filters (the Computer and Network Security industry, headquarters in the United States, 51 to 200 employees) matched 1,504 companies. We pulled 25 of them.
Each company returned counts as one record, so a page of 25 costs 25 records whether or not the companies fit. Which is why the next step exists.
Step 2: How do you check that each account fits before you search for people?
Read what came back against what you meant, and set aside the accounts that only matched on paper.
A filter matches what a company page declares. Your profile means something narrower: a company that sells, with a sales team, in your market. Three fields in each result do most of the checking:
- The company type. A nonprofit or a government agency can sit in the right industry and still not be the buyer your profile describes.
- The employee count on LinkedIn, next to the size band. The band is what the page declares; the count is the number of employees found on LinkedIn. When they disagree, decide which one your profile means.
- The description. A sentence or two tells you whether the company sells the kind of product your buyer is responsible for.
Of our 25 accounts, 7 did not pass. Two were nonprofits, both above the band on LinkedIn as well. Five declared 51 to 200 employees while LinkedIn counted between 24 and 1,271 people. We kept 18.
25 accounts pulled · employees on LinkedIn
- 156Account 1: 156 employees on LinkedIn, kept
- 142Account 2: 142 employees on LinkedIn, kept
- 538Account 3: 538 employees on LinkedIn, set aside, nonprofit
- 138Account 4: 138 employees on LinkedIn, kept
- 390Account 5: 390 employees on LinkedIn, set aside, nonprofit
- 99Account 6: 99 employees on LinkedIn, kept
- 272Account 7: 272 employees on LinkedIn, set aside, headcount outside 51 to 200
- 1,271Account 8: 1,271 employees on LinkedIn, set aside, headcount outside 51 to 200
- 895Account 9: 895 employees on LinkedIn, set aside, headcount outside 51 to 200
- 171Account 10: 171 employees on LinkedIn, kept
- 24Account 11: 24 employees on LinkedIn, set aside, headcount outside 51 to 200
- 200Account 12: 200 employees on LinkedIn, kept
- 52Account 13: 52 employees on LinkedIn, kept
- 185Account 14: 185 employees on LinkedIn, kept
- 64Account 15: 64 employees on LinkedIn, kept
- 176Account 16: 176 employees on LinkedIn, kept
- 95Account 17: 95 employees on LinkedIn, kept
- 115Account 18: 115 employees on LinkedIn, kept
- 139Account 19: 139 employees on LinkedIn, kept
- 218Account 20: 218 employees on LinkedIn, set aside, headcount outside 51 to 200
- 83Account 21: 83 employees on LinkedIn, kept
- 101Account 22: 101 employees on LinkedIn, kept
- 172Account 23: 172 employees on LinkedIn, kept
- 160Account 24: 160 employees on LinkedIn, kept
- 175Account 25: 175 employees on LinkedIn, kept
- 18 kept Privately held, 51 to 200 on LinkedIn
- 2 nonprofits right industry, not the buyer you want
- 5 outside the band declared 51 to 200, counted 24 to 1,271
Those 7 had already counted 7 records. Both checks also exist as filters: the company type, and the employee count on LinkedIn as a range. Put them in the query, and the next pull won’t return the same misfits.
We then measured both filters on the whole segment, at one record per count. Of the 1,504 companies whose page declares 51 to 200 employees, 1,088 count 50 people or fewer on LinkedIn, 30 count more than 200, and 386 fall inside the band. Keep only the privately held ones, and 328 remain. Our page of 25 held some of the most followed company pages in the segment, and it made the gap look much smaller than it is.
Check the account before you pay for its people.
Step 3: How do you find the decision maker in each account?
Rank the job titles you would accept, and search them in that order inside each account, keeping the first match.
An outbound list needs one right person per account, not everyone with Sales in their title. Waterfall ICP takes a company page and a ranked list of title levels, searches the first level, and moves down the list while result slots are left. With one slot, as here, it stops at the first match. The decision makers guide shows how to write those levels.
Ours had three: VP Sales, CRO or Head of Sales first; Sales Director next; the founder or CEO last. Across the 18 accounts:
- 10 returned someone at the first level;
- 4 at the second level;
- none had to fall back to the founder or CEO;
- 4 came back empty, and counted nothing.
Fourteen decision makers, fourteen records. An empty account is an answer too: no one on file there holds a title that matches your phrases, and learning it cost nothing.
Step 4: How do you add work emails and phone numbers without guessing?
Ask for the verified address of each person you kept, and accept not found as an answer.
Email Enrichment takes a person’s profile URL and returns a work email only when it has verified one. Otherwise it returns found: false, and nothing is counted. It does not build addresses from a pattern, so a blank field in your list means exactly what it says.
Of our 14 decision makers, 8 came back with an email and 6 without. We left those 6 blank. Guessing them would have put six unproven addresses in the list, the first habit of the previous section.
Then comes one more check, about the company rather than the mailbox: does the address belong to the account? People hold more than one job at a time, such as a board seat, an association or a company of their own. The response lists every address it found, each with the company page of the job it belongs to, and the first one is not always the account on your list.
- job 1Board memberA security conference
- job 2Director of SalesAcme Security, the account on your list
Email Enrichment returns
all_emails[]
- alex@cyberconf.examplecompany_linkedin_urllinkedin.com/company/cyberconfAnother organization: not for this list
- alex.morgan@acme-security.examplecompany_linkedin_urllinkedin.com/company/acme-securityMatches the account: keep
In our run, 2 of the 8 needed that check. One person’s first address was at an association where they hold a second, current role; their address at the account came second, and that is the one we kept. Another person’s only address belonged to a different employer: verified, deliverable, and wrong for this list. We set it aside. That left 7 work emails, all on the account’s own domain.
Phone Enrichment works the same way, with one limit that decides who goes through it: coverage is the United States only, and it needs a paid plan. Of our 14 people, 7 were located in the US, and 4 of them came back with a number. Four more had no country on their profile; since a miss counts nothing, we tried them as well, and 1 number came back. The 3 people based in India, France and Australia were not looked up.
The phone response gives one number and does not say which job it comes from. The documentation says it is taken from the first job in the profile that has one, current or past, so a number can come from a former job: confirm it on the first call.
Show the answer key
- Write firstname.lastname@ on the company domain. Most companies use that format.That is a guessed address: nothing has shown that the mailbox exists. It is the first habit that makes lists bounce, and the bounce counts against your sending domain.
- Keep the row with the email blank, and reach the person by phone or on LinkedIn.The person is still the right one at the right account. A blank field is an honest answer; the row stays in the list without risking a bounce.
- Use the general info@ address of the company instead.It was not returned for this person, and it does not reach them. A shared inbox is a guess about who reads it, not a contact point.
How do you check a prospect list for bounces before you send?
Run five checks on the finished list. All of them read data you already have.
- Every address was returned as verified. No pattern, no info@, nothing typed in by hand.
- Every address belongs to the account. Its company page is the account’s page, and its domain is the account’s domain.
- Every person still holds the job the row describes. The title came from a current position, and the row carries the date it was found.
- No duplicates. One row per profile URL, and one decision maker per account unless you chose more.
- The list is fresh. Send it within days of building it. Before reusing a list older than a month, run the email step again from the stored profile URLs, and drop the addresses that no longer come back.
Addresses from anywhere else, such as an old export, an event list or a partner’s file, need their own verification before they join. BlitzAPI does not check addresses you send it: it returns the addresses it has verified, and found: false otherwise.
A bounce rate is measured after sending, not before. We did not send anything to this list, so this guide reports no bounce rate. What it shows is a list in which every address was verified when it was returned and belongs to the account it sits next to.
What did this prospect list cost in records?
BlitzAPI built this list on November 10, 2026: 25 US cybersecurity accounts with 51 to 200 employees, 18 kept after the fit check, 14 decision makers, 7 verified work emails at the account and 5 US phone numbers, for 53 records against a budget of 101 stated before the first call.
- Size the market1,504 matching companiesAnnounced1Used1
- Pull the accounts25 accounts, 18 kept after the fit checkAnnounced25Used25
- Find the decision makers14 people in 18 accountsAnnounced25Used14
- Add work emails8 found, 7 at the accountAnnounced25Used8
- Add US phone numbers5 foundAnnounced25Used5
Counted nothing4 empty accounts6 people with no verified email6 phone lookups with no number
The budget was the worst case: every account fits, every account has a decision maker, every person has both an email and a phone number. The run came in close to half of it, for two reasons. The fit check took 7 accounts out before any search for people, and a miss counted nothing: 4 empty accounts, 6 people with no verified email, 6 phone lookups with no number.
Nine of the 14 people can be reached by email or phone, and 3 by both. The other 5 are still the right people at the right accounts: they go to LinkedIn, not to a guessed address.
To budget a larger list, use the same arithmetic before you run it: one record per account pulled, one per decision maker found, one per email and per phone number found. The worst case is easy to state in advance, and the real figure only comes in under it. The five one-record counts of step 2, made after the list was built, brought the day’s total to 58 records, still within the 101 stated.
For builders
Everything above runs on four endpoints. This part is for the people who will wire them up: the calls of the run, a sample response, and what to check before you ship.
How do you run this chain with the BlitzAPI API?
Your ICP as filters in, matching companies out, page by page, with the total number of matches.
Send
company.industry, company.hq, company.employee_range- The first three filters of most lists. Industry values come from a fixed list of normalized names.
company.type, company.employee_count- The two fit checks of step 2, as filters: Privately Held, and a min and max on the employee count on LinkedIn.
max_results, cursor- Up to 50 companies per page. Send back the cursor of the previous response for the next page.
Read
total_results- Every company that matches, on every page: the size of the segment.
results[].linkedin_url- The company page: the input of Waterfall ICP.
results[].type, size, employees_on_linkedin, about- What the fit check reads.
A company page and ranked title levels in, the best-fit people out, with the level that found each one.
Send
company_linkedin_url- From the company search result.
cascade[]- 1 to 10 levels, each with include_title, and exclude_title when needed. Current jobs only, by default.
max_results- 1 for one decision maker per account.
Read
results[].icp- The level that found the person: 1 is your first choice.
results[].person.linkedin_url- The input of the email and phone calls.
results[].person.location- Its country_code decides whether the phone call is worth making.
A profile URL in, a verified work email out, or found: false and no record counted.
Send
person_linkedin_url- From the Waterfall ICP result.
Read
found, email- The primary verified address. In the run, it was the address of the first job in profile order, not always the account.
all_emails[]- Every address found, each with company_linkedin_url and email_domain: where the account check happens.
A profile URL in, one US phone number out, the mobile when there is one, or found: false.
Send
person_linkedin_url- Only for people in the United States: coverage is US only, on paid plans.
Read
found, phone- One number, in international format, from the first job in profile order that has one, current or past. The response does not say which job.
Each company, person, email or phone number returned is one record; a call that finds nothing counts nothing. Plans and rate limits are on the BlitzAPI pricing page. The four endpoints are also tools of the BlitzAPI MCP server, which is how the run in this guide was made.
The calls of the run
The account search with the two fit checks of step 2 as filters. Run on November 10, 2026 with max_results set to 1, it reported 328 matching companies, against 1,504 without them:
{
"company": {
"industry": { "include": ["Computer and Network Security"] },
"hq": { "country_code": ["US"] },
"employee_range": ["51-200"],
"type": { "include": ["Privately Held"] },
"employee_count": { "min": 51, "max": 200 }
},
"max_results": 1
}The cascade, the same for every account:
{
"company_linkedin_url": "https://www.linkedin.com/company/acme-security",
"cascade": [
{
"include_title": ["VP Sales", "Vice President Sales", "Chief Revenue Officer", "CRO", "Head of Sales", "Chief Sales Officer"],
"exclude_title": ["assistant", "intern", "operations", "enablement", "engineering"]
},
{
"include_title": ["Sales Director", "Director of Sales", "Director Sales"],
"exclude_title": ["assistant", "intern", "operations", "enablement", "engineering"]
},
{
"include_title": ["[CEO]", "Chief Executive Officer", "Founder", "Co-Founder", "[President]"],
"exclude_title": ["assistant", "intern"]
}
],
"max_results": 1
}An email response with two addresses, trimmed (fictional company and person, real structure):
{
"found": true,
"email": "alex@cyberconf.example",
"all_emails": [
{
"email": "alex@cyberconf.example",
"job_order_in_profile": 1,
"company_linkedin_url": "https://www.linkedin.com/company/cyberconf",
"email_domain": "cyberconf.example"
},
{
"email": "alex.morgan@acme-security.example",
"job_order_in_profile": 2,
"company_linkedin_url": "https://www.linkedin.com/company/acme-security",
"email_domain": "acme-security.example"
}
]
}The address for the list is the entry whose company_linkedin_url is the account’s page, here the second one.
Before you ship
- Read total_results with max_results set to 1 before a large pull: one record for the size of the segment.
- Put the fit checks in the query (type, employee_count), then still read about before you search for people.
- Pick the address from all_emails[] whose company_linkedin_url is the account, not the top-level email.
- Treat found: false as a result. Leave the field blank and keep the row for another channel.
- Building an address from a pattern, or sending to info@, when no verified email came back.
- Call the phone endpoint only for people located in the US, and store the date each value was returned.
Sources and method
- Written by
- The BlitzAPI team
- Published
- API calls run on
- Company Search, BlitzAPI documentation
- Waterfall ICP logic, BlitzAPI documentation
- Find Work Email, BlitzAPI API reference
- Find Mobile & Direct Phone, BlitzAPI API reference
- Data quality: how BlitzAPI verifies data, BlitzAPI documentation
- Plans, records and rate limits, BlitzAPI documentation
- List building playbook, BlitzAPI documentation
The BlitzAPI team built the list in this guide on November 10, 2026, through the BlitzAPI MCP server, with a budget of 101 records stated before the first call. The figures are the counts of that run, and the records are the sum of the records_used field of its 45 responses. Five more one-record counts, made afterwards, measured the filters of step 2.
The guide names no company and no person from the run: it gives the segment (industry, country, headcount) and the counts. No email was sent to the list, so the guide reports no bounce rate.
Acme Security, the security conference and the addresses in the diagrams and in the sample responses are fictional. They reproduce the structure of the real responses and the cases met in the run.