What Is B2B Contact Data?

What is B2B contact data?

B2B contact data is the professional information that identifies a specific person at a specific company and makes it possible to reach them: their name, job title, seniority, department, employer, work email address, direct phone number and public professional profile. It is business-to-business data by definition, which means it describes people in their working capacity rather than their private lives.

That scope boundary is the whole reason the category exists as a separate thing from consumer data. A B2B contact record holds a work email at a company domain, not a personal Gmail address. It holds a direct office line or a business mobile, not a home number. It holds an employer and a job function, not an income bracket or a household composition. Regulators treat those two bodies of data very differently, buyers use them for completely different purposes, and a vendor that blurs the two is worth walking away from.

The practical definition most teams work with is narrower still. B2B contact data, in the sense a revenue team means it, is the set of fields you need before you can do anything with a lead: enough to route it, enough to score it, and enough to contact the person. A name and a company alone is a research task. A name, a company, a verified email and a title is a workable record.

What makes the category difficult is that these facts are not stable and they are not owned by anyone. A person's work email is created by their employer and dies the day they leave. Their title changes on a promotion nobody announces. Their company's headcount moves quarterly. No vendor observes any of this directly, so every commercial B2B contact database is a reconstruction of a moving reality, assembled from public profiles, self-reported information, licensed feeds and inference, and always slightly behind.

What fields does a B2B contact record actually contain?

A complete B2B contact record has two halves: the person and the company they work for. In Cleanlist the stored fields come back spelled exactly as the API spells them, which is work_email, email_status, direct_dial, linkedin_url, job_title, seniority, department, company_name, company_domain, company_headcount, company_industry, hq_location and founded_year.

That literal list is worth more than any marketing phrase, because the only question a data buyer genuinely has is whether their own field is in the response. A term like complete contact profile does not answer it. Ask any vendor for the field inventory in writing before a trial rather than after one, and check specifically for the field your workflow breaks without.

The person fields. Full name, job_title as the employer writes it, seniority as a normalized level, department as a normalized function, work_email, email_status carrying the verification verdict, direct_dial, and linkedin_url. The two normalized fields matter more than they look: raw titles are chaos (VP Sales, Vice President of Sales, V.P. of Sales and SVP, Sales all describe roughly one job), and seniority plus department are what make a list filterable and a lead-scoring rule possible.

The company fields. company_name, company_domain, company_headcount, company_industry, hq_location and founded_year in the stored record, and up to 180 firmographic properties on an enriched company. The domain is the single most useful field on the whole record, because it is the only identifier in the set that is unique and machine-checkable. Company names collide, get abbreviated, and get written three ways across three imports. A domain does not.

The field that carries a verdict rather than a value. email_status is different in kind from everything else, because it reports what a live mail server said about the address rather than what a provider claims. In Cleanlist it comes back as verified, accept_all for a catch-all domain that accepts everything and therefore proves nothing, invalid where the server refused the address, disposable for a burner domain, or a role-address label for info@ and sales@. Collapsing all of those into a single green tick is the habit that produces bounce rates in the double digits.

What is not a stored field. Funding events, hiring signals, tech stack and recent news are produced at the moment you ask rather than read out of a record. In Cleanlist that is an AI research column priced at 5 credits, and the answer carries the date it was produced, because a produced answer without a date is worthless six weeks later.

Where does B2B contact data come from?

Almost none of it is collected from the people it describes. Commercial B2B contact data is assembled from four kinds of source: public professional profiles and company websites, self-reported information people give to third parties, licensed feeds bought from other data companies, and inference that derives a likely value from patterns rather than observing it.

Public professional sources. Company team pages, press releases, conference speaker lists, regulatory filings, patent records and professional networking profiles. This is where job titles, employers and profile URLs mostly originate. It is the highest-quality layer for who somebody is, and the weakest layer for how to reach them, because a public profile almost never publishes a direct dial.

Self-reported information. Webinar registrations, gated content forms, trade show badge scans, review-site profiles and directory listings. People type their real work email here more often than anywhere else, which makes it valuable and also makes provenance matter enormously: the lawful basis for reusing that data depends entirely on what the person was told at the point they entered it.

Licensed feeds. Data companies buy from each other constantly. This is the least visible layer and the one that causes the most confusion in vendor evaluations, because two vendors quoting very different coverage numbers may be reselling substantially the same underlying feed. It is also why running the same 200 test records through three vendors sometimes returns three near-identical result sets.

Inference and pattern matching. A large share of email addresses in circulation were never observed anywhere. They were generated from a company's known email pattern and a person's name, in the shape firstname.lastname@company.com. A guessed address passes a syntax check perfectly well, which means a vendor that fills gaps with guesses can report coverage close to 100%. The cost of that shows up later as bounces, and bounces are charged to your sending domain rather than to the vendor. The only defence is verification against a live mail server, which is a separate step from finding the address.

Cleanlist sits deliberately at one remove from all of this. It operates no web crawler and maintains no contact database of its own. Every contact record it returns is resolved through licensed commercial providers, disclosed by name in the privacy policy, currently Anymailfinder, Crustdata, Datagma, Findymail, Hunter, Icypeas, LeadMagic, Prospeo and Wiza, each of which is responsible for the notice and lawful basis applicable to its own collection. That structure is unusual enough in this market to be worth stating plainly, and it is the fact that most changes how the compliance questions further down this page get answered.

How fast does B2B contact data decay, and which fields go first?

B2B contact data decays at about 2.1% a month and 22.5% a year on Cognism's published figures, and the decay is not spread evenly across fields. Email addresses go fastest at 22.5% to 30% a year (Cognism and SparkDBI), phone numbers at roughly 18% a year (SparkDBI), and company-level data such as revenue, headcount and industry classification at roughly 15% a year (6sense). Underneath all of it, LinkedIn's Economic Graph puts annual job changes at 10.9% of professionals.

Compound those rates and the shape of the problem is clear. A contact database that was accurate two years ago and has not been touched since is not slightly stale. It is wrong on something close to two records in five, and the wrongness is concentrated in exactly the fields outbound depends on.

Decay is also unevenly distributed across your list rather than uniformly across every row, which is the part most decay statistics obscure. Industry drives it hard: SaaS and startup records decay at 40% to 50% a year on the industry breakdown Cleanlist publishes, technology at 35% to 45%, while manufacturing sits at 10% to 15% and government at 5% to 10%. Seniority drives it too, because senior people move more visibly and more often than individual contributors. A list of VPs at Series B software companies ages at several times the rate of a list of plant managers at industrial suppliers, and a single blended refresh cadence applied to both wastes money on one and under-serves the other.

The failure mode is quiet, which is what makes it expensive. A stale record does not throw an error. It sends an email that bounces against your sending domain, or routes a lead to a territory owner who no longer covers it, or scores an account at last year's headcount. The cost lands in deliverability and in rep time rather than in a line item anyone reviews at month end.

The practical consequence is that the useful question about any B2B contact database is never whether it is accurate. It is when each record was last confirmed, and against what. A vendor that can answer that per record is selling something different from a vendor that quotes one accuracy percentage for a database of 200 million people.

What is the difference between a contact database you license and a waterfall you query?

A licensed database sells you access to one company's stored records for a fixed annual fee, and you browse it. A waterfall spends nothing until you ask for a specific record, queries several providers in sequence at that moment, verifies the answer, and charges only when it returns something. These are the two ways B2B contact data is actually bought, and choosing between them is the decision that determines almost everything else.

How a licensed database behaves. You get a search interface over a stored corpus, usually described in hundreds of millions of contacts. You can explore a market, build a list from filters, and see the size of a segment before you commit to it. The cost is fixed and known in advance, and it is the same whether a seat used the tool heavily or not at all. The weakness is freshness: every record sits in that corpus between refresh cycles, decaying at the rates above, and the vendor's refresh schedule is not usually something you control or can see. This is the model Demandbase, ZoomInfo, 6sense, Cognism and Apollo all sell, in various shapes.

How a waterfall behaves. You bring a record, or a list, or a search result, and ask for the missing fields. The engine walks a pool of providers in cost order, stops at the first confirmed answer, and never calls or bills the providers after the hit. The result is verified before it is accepted. Coverage comes from the pool rather than from any single database, which matters because no one database holds every person, and the recovery from asking a second and third source compounds across a list. The weakness is the mirror image of the database's strength: a waterfall answers only the question you asked, and you cannot browse it the way you can browse a corpus you licensed.

The cost shapes differ more than the sticker prices suggest. A licensed database charges per seat per year, so data spend drifts away from usage over a twelve month contract and the marginal record costs nothing, which quietly encourages exporting everything. A waterfall charges per returned record, so the invoice tracks what the team actually looked up, and a lookup that fails costs zero. Cleanlist charges 1 credit for a verified work email, 10 for a direct dial, 11 for both on the same contact, 1 for a company record, and 0 for a row that comes back empty. People Search and Company Search themselves cost 0 credits, so finding the records to enrich is free and only the fields are billed.

Neither model is universally right, and this is the honest version. A team that wants to sit inside a UI and explore a market of several hundred million stored contacts, run territory planning off it, and see segment sizes before committing, is buying a database and should buy one. A team that already has the list, the CRM records or the search results, and needs the missing fields filled and verified, is buying a waterfall. Plenty of companies run both, using a database for discovery and a waterfall for the fields that must be right on the day they are used.

How much does B2B contact data cost?

B2B contact data costs between $19.60 and $582.35 per 1,000 verified email addresses at entry volume, a 29.7x spread for what every vendor in that range describes as the same unit. Cleanlist established that range on August 7, 2026 by fetching 21 vendor pricing pages first-hand and normalizing every published price to one denominator: 1,000 verified business email addresses, exportable, on the cheapest plan buyable without talking to a salesperson.

The more useful finding from that exercise was how few vendors can be priced at all. Only 8 of the 16 contact data vendors surveyed publish enough information to compute the number, and three publish no purchasable dollar price at any tier. Across the eight that can be normalized, the median of each vendor's own cheapest figure is $89.29 per 1,000 verified emails. So anything between roughly $20 and $125 per 1,000 sits inside the published band at entry volume, and anything above about $250 per 1,000 is at the expensive end of what vendors are willing to print.

The two endpoints of that range explain most of the spread, and neither is a trick. Apollo Basic on annual billing works out to $19.60 per 1,000 at its published 1-credit email reveal rate, which is genuinely the cheapest published rate in the market. UpLead Essentials works out to $582.35 per 1,000 because it bundles email and phone into a single credit, so you pay the phone price on every row whether you use the number or not. Comparing those two headline numbers without knowing that is how budgets go wrong.

Where Cleanlist lands by the same arithmetic. Starter is $79 a month for 1,500 credits, which is $52.67 per 1,000 verified emails, or $39.33 on annual billing at $59 a month. Pro is $229 for 5,000 credits ($45.80 per 1,000, $34.40 annual). Scale is $599 for 15,000 credits ($39.93 per 1,000, $29.93 annual). Annual billing takes 25% off across the board. Cleanlist is not the cheapest vendor in that index and does not claim to be. What the row demonstrates is that the number can be computed at all, which is true of fewer than half the vendors surveyed.

Phone numbers are priced separately and cost five to ten times an email everywhere. Cleanlist charges 10 credits for a direct dial, which is $526.67 per 1,000 numbers on monthly Starter and $393.33 on annual. That ratio is the market norm rather than a Cleanlist quirk, and it exists because direct dials are harder to source, harder to verify and faster to break than email addresses.

Verification is the cheapest thing in the category. Email verification runs $3.90 to $8.00 per 1,000 addresses at entry volume across the vendors surveyed, and falls below $1 per 1,000 at million scale. In Cleanlist, validating an address you already hold is 0.5 credits. If your problem is bounce rate rather than coverage, verification is a much cheaper first move than re-buying the list.

The costs that do not appear on pricing pages. Watch for annual minimum commitments, per-seat charges layered on top of credits, overage billed at a higher rate than the plan rate, export restrictions on the tier you are quoted, and credits that expire at the end of a month or a year. A published per-credit rate means very little until you know what a credit buys, because at least one major vendor prices an in-platform reveal and a bulk enrichment of the same email at different credit counts on the same page.

How accurate is B2B contact data, and how do you test a vendor's claim?

Accuracy claims in this market are almost never independently measured, so the only number worth acting on is the one you produce yourself against your own list. The Cleanlist 500-Lead Enrichment Benchmark (2026) ran 500 stratified B2B leads through identical input and found single-source providers returned a verified work email for 70% to 80% and a direct dial for 30% to 60%, while a 25+ provider waterfall returned 98% verified email and 85% direct dial on the same 500 leads.

The reason the gap is that wide is coverage rather than quality. No single database holds every person, and the second, third and fourth providers in a sequence recover records the first one simply does not have. Phone is where the spread is widest, and phone is also the field where the difference matters most commercially, because reaching a decision maker directly rather than through a switchboard changes the outcome of the call.

Match rate and accuracy rate are two different numbers, and vendors conflate them constantly. A provider may match 90% of your records, but if 15% of those matches are stale or resolve to the wrong person, your real accuracy is 76.5%. Ask for both figures separately. A vendor who reports only one of them is reporting the flattering one.

How to run a test that actually tells you something. Take 200 to 500 records from your real ICP, not from the vendor's sample. Include the segments you know are hard: small companies, non-English-speaking markets, non-technical job functions. Strip the fields you want the vendor to find and keep a copy of the truth. Then measure four things: fill rate per field, email deliverability on a real send to a subset, phone connect rate on 50 dials, and title accuracy checked by hand against public profiles. Run the same file through every vendor on the shortlist so the comparison is like for like.

Watch specifically for guessed data. Generate the pattern-matched version of each email yourself (firstname.lastname@domain and its common variants) before you run the test, then check how many of the vendor's returns are identical to a guess you could have produced for free. A vendor that fills gaps with patterns and does not verify them is charging you for arithmetic.

What a low match rate is usually telling you. If a quarter of a list comes back empty across several vendors, the honest interpretation is usually about the list rather than about the enrichment. The ICP may be aimed at companies too small to have discoverable contact data, or at a market where the provider pool is genuinely thin. Changing vendor rarely fixes that. Changing the target sometimes does.

Is B2B contact data legal under GDPR and CCPA?

Yes, and the lawful basis for B2B contact data in the EEA and the UK is normally legitimate interests under Article 6(1)(f) of the GDPR rather than consent. Business contact details are still personal data under the GDPR, so the whole regime applies, but the limited and professional nature of the data is what makes the legitimate interests balancing test survivable for business-to-business outreach in a way it would not be for consumer marketing.

Legitimate interests is not a blank cheque, and treating it as one is where most B2B data programs actually fail a review. It requires a documented Legitimate Interest Assessment weighing your interest against the individual's rights, records of where each data category came from and what it is used for, and a genuine, easy way for a person to object. An objection to processing for direct marketing purposes is close to absolute under Article 21, so the opt-out mechanism has to work rather than exist.

How this works in Cleanlist specifically. The privacy policy states the position directly: Cleanlist does not collect this information from the individuals it describes, does not operate a web crawler, and does not maintain its own contact database. Data is licensed from the commercial providers named earlier, each responsible for the notice and lawful basis applicable to its own collection. Cleanlist holds name and job title, employer name, company domain and firmographics, business email address, business telephone number, and public professional profile URL, and states that it does not knowingly collect special category data, government identifiers, financial account details or health information. It processes that data solely so customers can identify, contact and qualify prospective business customers, and explicitly not for consumer marketing, credit decisions, employment decisions, insurance decisions, or any other purpose with a legal or similarly significant effect on an individual.

Under the CCPA and the US state laws that followed it, the obligations that bite hardest for B2B contact data are disclosure of the categories and sources of personal information, and honouring opt-out-of-sale-or-sharing requests. California residents can direct a provider not to sell or share their personal information, and a B2B data vendor has to be able to act on that.

Under CAN-SPAM and its equivalents, every commercial email needs a working unsubscribe and a physical postal address, and opt-outs must be honoured within 10 business days. Canada's CASL is materially stricter than CAN-SPAM and consent-based rather than opt-out, and the EU member state implementations of the ePrivacy Directive vary, so a single global sending policy calibrated to US rules is a liability if you send into Europe or Canada.

What to ask a vendor before you buy. Where does each data category originate, and can you name the providers. What is the lawful basis, and is the Legitimate Interest Assessment documented. How does a person exercise a deletion or objection request, and what is the response time. Are the upstream data sources disclosed as subprocessors in a document you can read now. A vendor that cannot answer those four clearly is a vendor whose compliance posture you would be inheriting without seeing it.

None of this is legal advice, and the answer that matters for your company depends on where you send, what you send, and what your own counsel signs off. Treat the paragraphs above as the questions to take into that conversation rather than the conclusion of it.

What rights does a person in a B2B contact database have, and how are they honoured?

A person whose business contact details sit in a commercial B2B database has five practical rights, and they apply regardless of whether that person has ever heard of the vendor holding the data: access, correction, deletion, objection to processing, and opt-out of sale or sharing.

Cleanlist honours all five for people who are not customers, through a route that requires no account and costs nothing. A request goes to contact@cleanlist.ai with the subject line Data Request and the name, employer, and email address or profile URL to look up. Cleanlist acknowledges the request and responds within 30 days, and may ask for limited information to confirm identity and to be certain it is acting on the right record, using that verification information for no other purpose.

Suppression is the part most vendors handle badly and it is worth understanding. Deleting a record outright creates a loop: the next provider refresh reintroduces the same person, because the upstream source never heard about the deletion. Cleanlist's approach is to retain the minimum identifiers needed to keep that person suppressed across future refreshes, used for nothing else. A vendor that deletes without suppressing will re-add the same person within a refresh cycle, which is the mechanism behind the common experience of opting out of a database twice.

The limit of what any vendor can do. If a customer has already exported a record into their own systems, that customer acts as an independent controller of their copy, and the original vendor cannot delete it on their behalf. Cleanlist's stated position is to say what it can about the export so the individual can approach the customer directly, and to suppress the record on its own side. That limitation is inherent to how this market works rather than specific to any one platform, and a vendor claiming it can guarantee erasure everywhere downstream is describing something it cannot do.

For a buyer, this section is not a compliance footnote. The quality of a vendor's opt-out machinery is a direct signal of the quality of everything else, because building working suppression across refresh cycles is unglamorous engineering that only gets done by teams who take the obligation seriously.

How is contact data different from firmographic, technographic and intent data?

Contact data describes a person, firmographic data describes the company around them, technographic data describes the software that company runs, and intent data is an inference that a company may be in a buying cycle. They get sold together under one heading and they come from completely different places, with completely different verifiability.

Contact data is facts about an individual: name, title, seniority, department, work email, direct dial, profile URL. It is verifiable, because an email can be checked against a live mail server and a title can be checked against a public profile. It decays fastest of the four.

Firmographic data is facts about an organisation: name, domain, industry, headcount, headquarters, founded year, revenue band, funding history. It is also verifiable, resolves for almost any record arriving with a real domain, and decays at roughly 15% a year rather than 22.5%. Cleanlist returns up to 180 firmographic properties on an enriched company record, and Company Search costs 0 credits.

Technographic data is the software a company runs, usually derived from scanning public-facing pages, job postings and DNS records. It is inferred rather than observed, it is systematically better for public-facing web technology than for internal systems, and it goes stale on a migration nobody publishes. Cleanlist does not sell it as a stored field.

Intent data is a probability, not a fact. It is inferred from behaviour that a third party observed somewhere else, typically content consumption on publisher networks, review-site activity or search patterns, and there is no way to verify a signal against a second source the way an email can be verified against a mail server. Cleanlist does not sell it. If a buying motion genuinely depends on intent, a platform built around it is the correct purchase, and ZoomInfo, 6sense and Cognism are the three most teams end up evaluating.

The reason to keep these separated in your head during a vendor evaluation is that they fail differently. A wrong email bounces immediately and visibly. A wrong technographic tag or a wrong intent signal quietly misdirects an entire quarter of prioritization and never announces itself. The verifiable layers deserve to be tested; the inferred layers deserve to be discounted until they have proven themselves against your own closed-won data.

How do you keep B2B contact data current once you have it?

Treat it as a recurring process on a cadence set by decay rate rather than as a project with an end date. A single cleanup restores the database to correct on the day it runs and starts decaying again the following morning at roughly 2.1% a month.

Enrich on entry. Every new record gets resolved and verified at the moment it arrives, whether from a form, an import or a search. This is the cheapest point in the lifecycle to do it, because the record has not yet been routed, scored, sequenced or acted on by a human, so a wrong value has not yet propagated anywhere.

Reverify contacts on a schedule tied to segment volatility. A blended quarterly refresh is the common default and it is wrong at both ends. Records in SaaS and technology decay at 35% to 50% a year and justify monthly reverification. Records in manufacturing and government decay at 5% to 15% and are being over-refreshed at quarterly. Segment the database by industry and seniority and set two or three cadences rather than one.

Refresh firmographics less often than contacts. A founded year never changes, an industry rarely does, and headcount moves on a quarterly rather than a daily scale. A team reverifying contacts monthly and firmographics quarterly is usually spending in roughly the right proportion.

Verify immediately before a send, not just at enrichment time. An address enriched and verified in March is a March fact. If a campaign goes out in September against a list built in March, the cheap move is a verification pass over the list first, at a small fraction of the cost of re-enriching it.

Watch the leading indicators rather than waiting for an audit. Rising bounce rate, falling phone connect rate and increasing manual corrections by reps are all the same signal arriving through three different channels. Any of them crossing a threshold should trigger a refresh cycle without anyone scheduling a review.

Close the loop back into the system of record. Enrichment that ends in a CSV nobody imports has changed nothing. The last step is always the record landing in the CRM, the list or the sequencer the team already works in, which is why the direction and field mapping of a CRM integration matter as much as whether one exists.

How do you evaluate a B2B contact data provider?

Run the same test file through every shortlisted vendor and compare seven things, in this order: coverage on your ICP, verification method, field inventory, provenance, pricing unit, integration direction, and the compliance documents. Vendor-published accuracy percentages should carry no weight in the decision, because none of them are independently measured and all of them were computed on a sample chosen by the vendor.

Coverage on your ICP, not in general. A database of 300 million contacts is irrelevant if it is thin on 30-person agencies in Portugal and that is your market. Test with 200 to 500 real records from the segments you actually sell into, weighted toward the hard ones.

Verification method, stated explicitly. Ask whether emails are checked against a live mail server at the mailbox level, and what happens to a catch-all domain. A vendor that returns catch-all addresses as verified is reporting a coverage number that includes addresses nobody can confirm exist. The honest handling is to flag them and let you decide.

Field inventory in writing. The literal list of keys the response contains, before the trial rather than after it. This is where a workflow-breaking gap gets found cheaply.

Provenance. Where did each category originate, and can the vendor name its sources. A vendor that treats its sourcing as a trade secret is asking you to inherit a compliance posture you are not allowed to inspect.

The pricing unit, decoded. What exactly is a credit, and does the same field cost a different number of credits in different parts of the product. Does a failed lookup cost anything. Do credits expire. Are exports capped independently of credits. These five questions account for most of the gap between a quoted price and an invoice.

Integration direction. Bidirectional sync with your CRM, field-level mapping, duplicate handling and activity logging. One-directional export into a CSV is a data delivery, not an integration, and it will quietly become somebody's weekly manual job.

The compliance documents, read rather than skimmed. The privacy policy's subprocessor disclosure, the DPA, the stated lawful basis, and the opt-out route. If your security review requires SOC 2 Type II or ISO 27001, ask early: Cleanlist does not currently hold either, and vendors including Cognism and ZoomInfo do. If a certification is a hard requirement in your organisation, that is a legitimate reason to buy elsewhere and it is better discovered in week one than in week six.

When is buying B2B contact data the wrong move?

Four situations where more contact data will not help, and one of them is by far the most common.

When your ICP is undefined. Buying contact data before deciding who you sell to produces a larger list of people who will not buy. Every downstream metric gets worse, not better, and the spend obscures the actual problem for another quarter. Define the segment first, even roughly, then buy against it.

When the problem is deliverability rather than coverage. If emails are bouncing, adding more addresses accelerates the damage to your sending domain. The sequence that works is: verify what you hold, remove what fails, repair sender reputation, then expand. Verification at 0.5 credits an address is a fraction of the cost of enrichment, and it addresses the actual failure.

When you need consumer data. B2B providers hold work emails and business phone numbers, and are explicitly scoped not to hold consumer identifiers. If your buyer is an individual acting personally, this entire category is the wrong shelf and the compliance regime is different in ways that matter.

When the market is small enough to research by hand. For a total addressable market of a few hundred named accounts, direct research produces better data than any vendor, because you can confirm the current title and the current org chart on the day you need them. Contact data providers earn their cost through volume. Below a few hundred records, the arithmetic often does not favour them.

There is also a fifth case that is not about the data at all. If reps are not working the leads they already have, the constraint is process rather than supply, and buying more contacts moves the bottleneck without relieving it.