What Is B2B Data? Types, Sources, Quality Checks and What a Good Record Looks Like

B2B data is information about companies and the people who work in them. A sales team uses it to pick accounts, reach the right person and keep its CRM current. It comes as two record types, company and contact, plus the signals that say when an account is worth a call.

The hard question is not what it is but how good it is. In Spona's own checks, 98.10% of 250,251 US profiles returned a work email. Of those addresses, 96.42% passed verification, and 0.64% of sends to verified addresses bounced hard. This post lays the data out record by record, then shows how to run the same test on any vendor before you pay.

Get thousands of leads in 4 minutes.

Stop stitching tools together.
Start with verified sales data, delivered in minutes.

SCALE SALES NOW

No contracts · No setup · No onboarding

What is B2B data?

ZoomInfo's definition is the shortest: "information about businesses and the people who work in them". Cognism widens it to "any information that benefits B2B companies", above all their sales, marketing and revenue teams. Coresignal lists the company half: "company identifiers, funding, revenue, and more". The three agree on the core: records about companies, records about the people inside them, and events that link the two.

A company record answers "which accounts fit us": industry, size, location, revenue band, the software in use. A contact record answers "who do we talk to, and how": name, title, seniority, a work email and a phone. A signal answers "when": a funding round, a hiring spike, a new office.

Your own sales data is a different thing. It records what your deals did and lives in your CRM, not in a vendor's file.

One point the vendor pages leave out. In the UK, a named person's work email is personal data. The ICO says details of someone at work "will constitute personal data even if they are acting in their business capacity". That shapes how you may use the contact half of any file, and the legal section below covers the rules.

Anatomy of a B2B data record

A good record carries more than a name and an email. The record schema on Spona's Expert API holds 21 fields in four groups. They cover who the person is, where they work, how to reach them, and where each fact came from.

The fourth group is the one buyers skip. A source URL tells you where the vendor saw the fact. A last-checked field tells you how old it is. Without both, you cannot judge a record, and you cannot tell a UK contact where their data came from when the law asks you to.

Presence is not validity. A filled field means the vendor holds a value, not a correct one. Fill rate measures how complete a file is and says nothing about accuracy.

With that caveat, here is how Spona's own records fill, field by field. The US file holds 21,181 companies and 55,293 contacts. The UK file holds 26,632 companies and 52,028 contacts.

Read the gaps, not the full rows. Revenue band keeps its estimate flag, so a filled cell may hold a modelled figure rather than a filed one. Technology evidence sits on 77.18% of US companies and 71.98% of UK ones: for roughly a quarter of companies, the file shows no stack.

Direct dials fill about two thirds of contact records in both countries. The suppression flag is present on every record, and a present flag is not a suppressed record, since its value can be false.

Five kinds of B2B data

ZoomInfo splits the field into five kinds: contact, firmographic, technographic, intent and signal data. Cognism uses the same list, where signal data goes by "chronographic", a name ZoomInfo's list uses too. What matters is what each kind lets you filter on, so the table pairs each kind with the share of Spona searches led by it. Spona classifies every search once, by its main filter, across 32,999 US searches and 43,016 UK searches.

The firmographic and signal rows are our own sums, and each search still counts once. Firmographic adds two filter groups, industry with headcount and headcount with revenue; signal adds hiring, funding and new-location searches. Firmographic filters lead a third of UK searches, and the mix shows why vendors bundle the kinds. A technology filter is useless without the company behind it, and a hiring signal is useless without a contact to call.

The remaining searches lead with a metro and owner filter for local businesses, a department, an account list or another mix. Local-business searches take 9.70% in the US and 9.79% in the UK. Intent data comes from a different pipeline altogether. Cognism, for one, attributes its intent data to Bombora's "cooperative of 5,000 B2B sites".

B2B data sources: where it comes from

Three sources feed every vendor file, and you can tap two of them yourself.

First-party data is what your own systems record: CRM entries, web forms, event lists, support tickets. It is the most accurate data you will ever hold about your customers, and it says nothing about anyone else.

Public registries are the backbone of company data. Companies House counts 5,485,994 companies on the UK register. Its free company data product is a downloadable file of every live company, refreshed within five working days of each month end.

In the US, SEC EDGAR gives free public access to the filings of listed companies, with full-text search across more than 20 years. Registries hold legal entities, not contact details. They give you the company half only.

Official business counts also set the ceiling on what any vendor can hold. The SBA counts 36,207,130 small businesses in the US, and 82.3% of them have no employees. The UK has about 5.7 million private sector businesses, of which 1.4 million employ anyone. A vendor who claims every business in either country is mostly counting firms with no employees.

Third-party collection fills the contact half. ZoomInfo lists typical sources: public records, company registries and company websites. Some providers add contributor networks and human research teams. Every vendor mixes these in its own proportions, which is why two files of the same segment rarely match.

Spona's source register holds 47 source families, deduplicated by original publisher; that register is the basis of its "40+ sources" line. How a vendor turns those inputs into a product is a question of its own, and the data vendors post answers it.

How accurate is B2B data?

B2B data accuracy has no single number. A vendor can be right on names and wrong on emails, so measure each field the way it fails. For emails the test is mechanical: does the mail server accept the address, and does a sent message arrive?

Here is Spona's own funnel over three months, from profile lookup to hard bounce, split US and UK. Each country's figures sum 30 cells, one per role family and industry.

Three things to read in it. First, the bounce rate covers verified addresses only. Spona sent nothing to catch-all domains, and a vendor that quotes a bounce rate without naming what it sent to is hiding the denominator.

Second, a hard bounce is a permanent failure. The receiving server answers with a 5.X.X code, and code X.1.1 means "the mailbox specified in the address does not exist". A soft bounce is temporary and belongs in a different count.

Third, about 2% of found addresses sit on catch-all domains. Those domains accept every address, so no verifier can confirm them. A careful vendor labels them instead of counting them as valid.

ZoomInfo's own advice holds for any vendor: emails should be "verified at the mail server as well as checked for format". A format check catches typos. Only a server check catches a person who has left. The validation methods post walks through each one.

How fast B2B data goes stale

A record that passes today can fail within a quarter, because people move. The BLS puts median tenure with a current employer at 4.1 years. For workers aged 25 to 34 it is three years.

In a single recent month, 5.1 million US workers left their employer, 3.2% of all jobs. Companies change too: Companies House logged 214,754 dissolutions in one quarter, against 193,789 new companies.

Spona measures what that does to a file with a stable panel of the same 10,000 records per country. Each record sits in one role family, and Spona checks the panel again at the end of a quarter, before any refresh.

Over that quarter, 7.04% of US panel records changed title or employer, and 4.91% saw their email change or go invalid. In the UK the figures were 6.37% and 4.50%. The two measures can hit the same person, so you do not add them. A quarter's rate is not an annual rate either.

The role split is where the chart earns its place. UK executives moved most, with a job change rate of 8.09%, and UK finance staff least, at 4.37%. In the US, IT and marketing roles turned over fastest and finance slowest.

Email change does not always track job change. US finance contacts changed jobs least of any US role, yet changed emails most, at 6.69%.

What to do with the figure: re-verify before every send, and refresh active contacts on a schedule. The data decay and data hygiene posts give the method and the cadence.

How to test B2B data before buying

Six steps, in order, before you sign anything.

1. Write the target down. Industry, headcount band, countries, roles, and the fields every record must carry. A vendor cannot pass a test you have not defined.
2. Ask for a sample you choose. ZoomInfo suggests 500 to 1,000 accounts with at least three verified contacts each. Name the accounts yourself, so nobody can cherry-pick.
3. Count fill rate per required field. A record with no phone is useless to a calling team, whatever its email score. Use the fill table above as the yardstick.
4. Verify at the server, then send. Run the sample through a verifier, send to the addresses that pass, and count hard bounces against sends. Compare the result with the funnel above and with acceptable bounce rates.
5. Check 20 records by hand. Open the source URL and the profile URL, and compare title and employer with the record. A record with no source URL fails this step by default.
6. Price the usable records. Divide the quote by the records that passed steps 3 to 5, not by the records delivered. That number is the price you are paying.

On Spona, every search starts with a free sample of leads, and the price per lead shows at checkout before you buy.

Is B2B data legal?

Yes, under rules that differ by country and by channel. This section describes those rules; it is not legal advice.

In the US, the email rule is light and the data broker rules are getting heavier. The FTC is blunt: the law "makes no exception for business-to-business email". You may email a stranger at work, but you must say who you are and stop when asked.

California adds two layers. The CCPA's exemption for "business-to-business transactions" has expired, so the law now covers California residents who are contacts for business customers or vendors. It applies to firms with revenue of $26.625 million or more, among other thresholds. Under the Delete Act, data brokers must access the state's deletion mechanism at least once every 45 days and process its deletion requests.

In the UK, the two halves of the data pull in opposite directions. PECR's rule on marketing email "does not apply to corporate subscribers", so a cold email to a limited company's employee needs no consent under PECR. Under UK GDPR Article 14, you must still tell the named person "from which source the personal data originate". The deadline is one month, or your first message if you use the data to contact them.

Two UK traps. Sole traders and some partnerships "are actually classed as individual subscribers". So emailing a sole-trader plumber needs consent or the soft opt-in, which emailing a company does not. And for calls, the ICO tells you to screen against both the TPS and the CTPS registers before you dial.

Spona's B2B data: fit and limits

Spona fits a team that wants to describe its customer in a chat and get a verified list fast, paying per lead, not per seat. You describe your ideal customer, and the AI mines its source register, verifies every lead and delivers a scored list.

Spona timed 83 completed Standard Search exports of 1,000 contacts each. The median time from submitted filters to a downloadable file was 223 seconds, verification included. Every search starts with a free sample, and each new search session costs a one-time 1,000 credits.

The Free plan carries 3,000 one-time credits. Pro costs $99 a month and Premium $499. Expert costs $999, and the full plan table sits on the pricing page.

Where it fits:

• Provenance on every record. Source URL and last-checked fill 100% of records in both countries, the two fields this post told you to demand.
• Signals in the same search. Filters for funding, hiring, technology, first-seen and new-location work on Pro and above, combined with the profile.
• Your own lists. Enrichment on every paid plan: company data for 3 credits, a verified email for 5. A mobile number costs 35 credits.
• HubSpot. A native integration on Premium and Expert.

Where it does not:

• No intent data. If your trigger is topic research, buy that elsewhere; ZoomInfo and Cognism both sell it.
• No visitor identification. Spona does not de-anonymize website traffic.
• Salesforce and the API. Salesforce means CSV import, not a native connector. The API runs on Expert only; an API key for Pro and Premium is announced, not live.
• Phones. Direct dials fill 66.14% of US contacts and mobiles 72.76%. A phone-first team should test the sample for its own roles. Cognism says a research team verifies its mobile numbers, so its sample is the comparison to run.
• Technology coverage. Technology evidence sits on 77.18% of US companies, so a technographic-led search cannot reach nearly a quarter of the US file.
• The law stays with you. Each record's source URL helps you name the source in an Article 14 notice, but sending the notice is your job.

Get thousands of leads in 4 minutes.

Stop stitching tools together.
Start with verified sales data, delivered in minutes.

SCALE SALES NOW

No contracts · No setup · No onboarding

B2B data FAQ

How much does B2B data cost?

Three pricing models exist. Per seat, where a yearly licence covers a user with capped or unlimited exports. Per record, where list sellers quote a price per thousand, and per credit, where you pay for what you export. Spona's public range runs from $0.008 to $2.00 a lead, set by search type and plan, with a free sample first.

Which provider should you buy from?

Match the provider to the kind you need most, then run the six-step test above on its sample. Our provider comparison sorts twelve of them by data type and price.

Where can I find a UK company database?

Start with Companies House. Its free data product covers every live company on the register and refreshes monthly. It holds companies, not people, so you add contacts from a vendor or build your own; the build vs buy post weighs both.

What is a B2B data list?

A B2B data list is a static file of records, exported once and usually sold per thousand. It starts decaying on the day of export, at rates like those in the panel above, so ask when each record was last checked.

What is the difference between B2B and B2C data?

A B2B record describes a company and a person's role inside it. A B2C record describes a consumer: age, household, purchases. The law follows the split. In the UK, PECR treats corporate and individual subscribers differently, yet a named work email is still personal data under UK GDPR.