B2B contact data comes from four places: contact databases (the Apollo, ZoomInfo, and Lusha class), enrichment waterfalls that query several providers for each contact (the Clay class), LinkedIn Sales Navigator plus a human researcher, and intent or trigger data that tells you who is active now. None is accurate enough on its own. The list that reaches the inbox is built from two of them and verified before every send.
This is for founders and sales leads deciding where the names for an outbound programme should come from. It compares the four sources on coverage, accuracy, and cost shape, gives a decision rule by list size and deal size, and explains the refresh rule we apply to every list. Tool names appear as examples of a category, not as recommendations; coverage differs by market and title, so test any vendor against your own ICP before you buy. Where data sits in the wider stack is in our guide to outbound sales infrastructure.
Where does B2B contact data actually come from?
Every vendor draws on the same few upstream sources: public web pages and company sites, professional profiles, address books and inboxes contributed by users of the vendor's own tools, and email patterns guessed from a company's format and checked against its mail server. Nobody owns the truth. The differences are freshness, how each vendor verifies, and which markets and titles it covers well.
That has two consequences. A US mid-market SaaS title is covered everywhere; a plant head in Pune or a finance director in Riyadh is thin in most databases and usually needs LinkedIn plus a researcher. And because roughly 2% to 3% of any list goes stale each month as people move, a record's age matters as much as its source. An accuracy claim on a vendor's page describes its best market, which may not be yours. Stick to business addresses; personal emails and mobile numbers raise consent questions that vary by market, and nothing here is legal advice.
Contact database vs enrichment waterfall vs manual research: how do they compare?
A database gives the widest coverage at the lowest accuracy. A waterfall improves the hit rate by asking several providers in turn and charges per contact. Manual research from Sales Navigator gives the most current titles and needs the address found and verified separately. Intent data adds timing on top of any of them and is not a source on its own.
| Source | Coverage | Accuracy (what we see) | Cost shape | Best for |
|---|---|---|---|---|
| Contact database (Apollo, ZoomInfo, Lusha class) | Wide; thin on niche titles and outside the US | 8% to 12% bounce when loaded raw | Seat licence plus export credits | Volume outbound into common titles; the first pass |
| Enrichment waterfall (Clay class) | Better; each miss is retried at the next provider | 3% to 5% bounce with built-in checks | Per credit; you pay for what you enrich | Mid-size lists, hard-to-find titles, layering signals |
| Sales Navigator plus manual research | Whatever is on LinkedIn; strongest for current role | Under 2% bounce after individual verification | Seat licence plus researcher hours | Account-based lists under 500, senior titles, high ACV |
| Intent and trigger data | Narrow by design: the slice of your ICP active now | Signal quality varies widely by provider | Annual contract; often the priciest line | Ranking and timing on top of another source |
The bounce figures are what we see when a list goes straight from the source to the sequencer. A separate verification pass, with catch-all domains handled, brings any of them under the 2% to 3% line that keeps sending domains healthy; how to run it is in how to scrub lists to avoid spam traps and invalid emails. The source decides how much you throw away; verification decides what you send. The chain we plan on is 1,000 raw records to about 880 verified to about 840 safe to send.
Which data source fits your list size and deal size?
Match the source to how many contacts you need a month and what a meeting is worth. Under 200 accounts with deals above $25,000: manual research, every address verified by hand. 300 to 1,000 contacts a month at $5,000 to $25,000: a database or Navigator export enriched through a waterfall, then verified. 1,000 or more contacts a month under $5,000: a database filtered hard by ICP, a waterfall for the misses, then verified.
| Your situation | Primary source | Fill the gaps with | Before sending |
|---|---|---|---|
| Under 200 accounts, ACV above $25,000 | Sales Navigator plus a researcher | A waterfall for the address only | Verify every address by hand; confirm the role on LinkedIn |
| 300 to 1,000 contacts a month, ACV $5,000 to $25,000 | Database or Navigator export | A waterfall for misses and phone numbers | Bulk verify; handle catch-alls; spot-check 5% by hand |
| 1,000+ contacts a month, ACV under $5,000 | Database filtered by ICP, not by firmographics alone | A waterfall on the highest-value segment only | Bulk verify; drop catch-alls or route them to LinkedIn |
| Any size where timing matters | Whichever row applies above | Trigger and intent data to rank the list | Re-verify anything older than 60 to 90 days |
The trap sits in the middle: buying 5,000 records 'just in case', filtered by size and industry. That is a filter, not an ICP, and the budget goes on people who were never going to buy; why a headcount range is a filter and not an ICP makes the argument in full. Data typically runs $500 to $1,500 a month per rep in the US and ₹15,000 to ₹40,000 a month in India. Spend it on fewer, better records. The step-by-step build is in how to build a B2B prospect list that does not bounce.
Sourcing and verifying the list is the first two weeks of every programme we run, and the list belongs to the client afterwards. If you would rather have it built than build it, that is part of our outreach service.
How fast does B2B contact data go stale?
Roughly 2% to 3% of a B2B list goes stale every month as people change roles, companies restructure, and mailboxes close. After three months, 6% to 9% of a list is wrong. After a year, between a fifth and a third. That is why anything older than 60 to 90 days is re-verified before it is sent, whatever the source and however clean it looked when it arrived.
Re-verification is more than an email check. Syntax and domain, mailbox existence, a catch-all flag, a role check on LinkedIn for the top tier, and a suppression pass against opt-outs, current customers, and open deals in the CRM. A bounce rate above 5% is broken and starts burning domains, and verification costs far less than a replacement domain and its 3 to 4 week warm-up.
What turns good data into bounces?
Five habits: buying by filter instead of ICP, loading a list without a verification pass, treating catch-all addresses as valid, sending to data older than 90 days, and using one source for every market and title. Each is fixable in an afternoon. Together they are the reason a raw bought list bounces 8% to 10% or more while a verified list bounces under 2%.
- Filter, not ICP. '200 to 1,000 employees, SaaS, US' describes thousands of companies that will never buy. Define who buys and why before you export anything.
- No verification pass. Every list, every time, including the one the vendor says is verified. Their check was run on their schedule, not on your send date.
- Catch-all domains treated as valid. A catch-all accepts every address, so a checker cannot confirm the mailbox exists. Route those contacts to LinkedIn, or send to them at low volume from a separate domain.
- Stale data. The 60 to 90 day rule. A list built for a campaign that slipped a quarter is a new list.
- One source everywhere. US-centric databases run thin in India, the Gulf, and much of Europe. Use LinkedIn plus a researcher where the database shows three contacts for a 500-person company.
When does a contact database not make sense?
When your list is under about 100 accounts, when your buyers sit outside the markets and titles the database covers well, or when you sell into regulated or enterprise accounts where one wrong contact costs you the account. In those cases a researcher with Sales Navigator, a verification tool, and a spreadsheet beats a licence.
A founder in the first months is in the same position: 20 hand-researched accounts a week teach more about the ICP than 2,000 exported rows, and the ICP has to exist before any tool can filter for it, so how to define your ICP in one afternoon comes first. The opposite edge is also real. If the motion needs 1,000 or more new contacts a month, manual research cannot keep up, and the answer is a database plus a waterfall plus verification, with a researcher on the top tier only.
How MarginSales approaches data sourcing
MarginSales provides sales outreach services for companies that want to extend their outbound capacity without building the entire sales development function internally. Every programme starts with ICP research, and data sourcing sits inside the pod's weekly work rather than being bought once at the start. We use at least two sources per list, verify before every send, re-verify at 60 to 90 days, handle catch-alls separately, and hold bounces under 2% to 3%. Weekly reporting shows bounce and reply rate by source, so a client can see which source works for their market, which matters in India and EMEA where US-centric databases run thin. The list lives in the client's CRM, whether that is HubSpot, Salesforce, Pipedrive, or Zoho, and belongs to them.
Enrichment, verification, and formatting are the simple parts, and they are automated. Deciding who the buyer really is, judging whether a title on a profile matches the person who signs, and reading a company's situation before it goes on a list are human work. Automate the simple, keep humans on the meaningful.
Frequently asked questions
Where do you get B2B contact data for cold email?
Four places: contact databases (the Apollo, ZoomInfo, and Lusha class), enrichment waterfalls that query several providers for each contact (the Clay class), LinkedIn Sales Navigator plus manual research, and intent or trigger data for timing. Most programmes combine two: a database or Navigator to find the people, then a waterfall or a researcher to find and verify the address.
How accurate is B2B contact data?
Less accurate than the sales page suggests. In our campaigns a list loaded straight from one database bounces 8% to 12%, a waterfall with its built-in checks 3% to 5%, and a manually researched, individually verified list under 2%. Roughly 2% to 3% of any list goes stale each month, so accuracy depends on the age of the record as much as its source.
Should you buy a contact list or build one?
Build one, using tools. A bought static list is already ageing when it arrives and is usually filtered by headcount and industry rather than by fit. Building from your ICP with a database or Navigator, enriching through a waterfall, and verifying before every send takes more effort and produces a list that reaches the inbox.
How often should you re-verify a prospect list?
Before every send, and always when the data is older than 60 to 90 days. At 2% to 3% decay a month, a list that was clean in January has 6% to 9% bad addresses by April, which is past the 5% bounce line where sending domains start taking damage. Re-verification costs far less than a replacement domain and its warm-up.
Get a bounce forecast on your current list
Send us a 200-row sample of your prospect list and tell us where it came from. We will tell you the bounce rate to expect, which rows are catch-alls, how much of it is older than 90 days, and which source would close your coverage gaps. Book the list review. You keep the findings whether or not we run the programme.