B2B Data for Outbound — buying it, verifying it, enriching it

Ask any room of people starting outbound what they need and the answer is a lead list. Where do I buy data, which provider is best, how many contacts can I get for the budget. It's the wrong first question, and it's the reason a lot of first campaigns fail in a way nobody diagnoses correctly.

Here's the pattern we see constantly, and it's worth stating plainly because it costs people months: bad records get deep into the system before anyone notices, the campaign underperforms, and the team blames the copy, the sequence or the reps. The subject line gets rewritten four times. The actual problem was that a chunk of the list was never real.

Start here: the list is the cheap part

Buying B2B data is close to a commodity purchase. The large providers assemble records from broadly the same places — public web, company filings, professional profiles, and contributed data from their own users' inboxes and CRMs. They then resell that at different prices with different coverage. Nobody has a secret population of decision-makers nobody else can see.

You're not buying accuracy. You're buying a snapshot, and the moment it's exported it starts going out of date.

Which means the interesting question isn't which vendor. It's what you do between exporting a list and sending to it. That gap — verify, enrich, decide who's actually worth contacting — is where campaigns are won, and no provider sells it to you.

One honest limit, before this post oversells its own subject. Better data does not fix a bad campaign. Teams switch providers expecting reply rates to move and nothing happens, because the real problem was a generic message or an ICP too broad to say anything specific to. Bad data guarantees failure; good data merely permits success.

1 · What you're actually buying

A record is a set of guesses of varying confidence. The company name and domain are usually solid. Headcount and revenue are estimates, often stale, sometimes wildly wrong for private companies. The job title is whatever was last published somewhere. The email is frequently derived — a pattern applied to a name and domain rather than a confirmed address — and the mobile number may have come from a contributed contact book years ago.

Three things follow. First, coverage is uneven in ways the sales demo won't show you. The same provider can be good for company data in a market and useless for contacts in that same market. Regional depth varies enormously, and so does depth by seniority — heads of department in mid-market companies are far patchier than VPs at large ones. Which is why coverage is the wrong number to compare on, and it's the number every provider leads with. Seventy million profiles tells you nothing about whether the four thousand companies you care about are in there and reachable. The metric that matters is the match rate: of the accounts you actually want, what proportion comes back usable. Nobody publishes it, because it depends entirely on who you're asking for — so measure it yourself on a sample.

Second, a provider that was the right answer two years ago may not be today. Data businesses lose access to sources, change how they infer emails, or grow so fast that quality slips. Reputations lag reality in both directions. Test rather than trust the ranking you read in a forum thread.

Third, and this one catches technical teams: a scraping platform is not an enrichment provider. Some of the best-known names in this space sell proxies and scraping infrastructure, which is a genuinely different product. Buy one expecting clean records and you inherit the work of normalising formats, handling retries and reconciling conflicting answers yourself. That's a reasonable trade if you have engineers and want control. It's a nasty surprise if you thought you were buying a list.

One warning before you go looking for advice: threads asking "which data provider is best" are the most vendor-infested corner of the internet. We read several while writing this — an unfamiliar tool named first and described at suspicious length, a founder disclosing halfway down, comments removed by moderators, and one commenter warning that vendors flood these threads, then recommending a product two sentences later. The telling detail: dozens of comments confidently ranking providers, and not a single match rate or price among them.

2 · Data decay: why every list goes stale

People change jobs. Companies get acquired, rebrand, consolidate domains, restructure departments. Every one of those events silently breaks records you already paid for. This isn't a flaw in a particular vendor — it's the nature of the thing. A list is a photograph of a moving object.

The cost isn't only bounces. Stale firmographics damage you in the conversation, and worse, silently — no report tells you it happened. Calling a company fifty people when they're eighty, or missing the round they closed last quarter, proves you did no real work at the exact moment you were trying to prove the opposite. Headcount running three or four months behind is routine.

The practical consequences are simple and mostly ignored:

• Pull close to send. A list exported and sat on for three months is materially worse than one exported last week. If your campaign starts in a fortnight, build the list in a fortnight.
• Don't reuse an old list without re-verifying. The one you cleaned in spring is not clean now, and re-sending to it is one of the faster ways to damage a sending domain.
• Job changes are a signal, not just decay. A contact who moved is a lost record and a new opportunity at the same time — they arrived somewhere with budget and something to prove.

With engineering capacity there's a better answer than pulling lists more often: look records up at the moment you need them instead of maintaining a static file. Pass a name and domain, hit a provider, verify, score, write to the CRM — nothing sits around going stale. It costs per lookup rather than per seat, so it suits continuous sending, and it changes the question from "how fresh is their database" to "how well does this fit how we work."

3 · Verification is its own line item

This is the section most people skip and most campaigns die on. Verification is not a feature you hope your data provider included. It's a separate step, with its own cost, that you run regardless of where the records came from — including from the expensive premium source.

Practitioners who run a proper chain — the provider's own free check, then a paid pass, then a dedicated verifier — commonly report culling 15–20% of a freshly purchased list before a single email goes out. And they still see some bounces afterwards. That's the honest state of the art, and it's a long way from the accuracy figures on the pricing pages.

The catch-all problem deserves its own paragraph, because nobody selling data volunteers it. Some domains accept mail to any address, real or not, so a verifier can't confirm whether a mailbox exists behind one — those records come back neither valid nor invalid. Bulk them in and your bounce rate is a lottery; drop them all and you lose a real slice of a niche market. Separate them, send at low volume from a domain you're willing to risk, and watch before scaling.

Where verification belongs in the process is genuinely contested, and both arguments are good. Run it right after enrichment and before anything reaches the CRM, and you stop bad records contaminating your reporting — otherwise you end up questioning your messaging when the real problem was coverage. Run it before enrichment and you avoid paying to enrich records that were dead anyway. If you're paying per enriched record, the second order saves real money.

There is also a cost nobody mentions: stricter verification cuts volume. With a narrow ICP you can clean your way down to a list too small to work, which is a real trade-off rather than a reason to skip verifying. Deliberately deciding where that line sits beats discovering it. The sending side of the same problem is covered in cold email deliverability.

4 · Enrichment worth paying for, and enrichment for its own sake

Enrichment has a simple test, and almost nobody applies it: does this field change what you send, or whether you send at all?

Fields that pass: headcount and growth, funding stage and date, tech stack, hiring activity, whether they already run the thing you replace. Each can become a first line or a reason an account moves up the list. Most of the rest fail. A record with forty attributes isn't a better prospect than one with four — it's the same prospect and a bigger bill. We've watched teams buy deep enrichment and then write "Hi {{first_name}}, I saw you're in {{industry}}".

The related trap is doing more research instead of better targeting. Per-prospect research should be short and rigidly scoped: does this company fit, am I talking to the decision-maker, is there a reason to contact them now. Three questions, a couple of minutes, then write. Open-ended research feels productive and mostly produces analysis paralysis — and it can't rescue a bad list.

5 · Signals sit above data, and matter more

Data tells you who. Signals tell you when, and when is the harder half. A perfectly accurate record for someone with no reason to care this quarter is worth less than a slightly stale one for a company that just raised, just hired, or just started migrating off the thing you replace.

That's why we treat signal monitoring as a separate layer rather than a premium data tier: funding, hiring, leadership changes, tech-stack moves, public deadlines. The tools for it are covered in the outreach tools breakdown, and the reasoning behind timing over volume in why cold email gets ignored.

One caution we've earned the hard way: a signal tells you when to write, not what to write. A list of companies hiring for a role you serve is worth very little if the email that goes out is still your standard pitch. The signal has to show up in the first sentence as an observation about them.

Be honest about how worn the obvious version is, though. "Saw you're hiring for X" now reads as automated rather than attentive — practitioners name that exact opener as one they ignore. The signal is still good; the lazy sentence built from it is dead. Say what the hiring implies for them, not that you noticed it.

6 · Ten markers of a list worth sending to

Before a sequence starts, run down this list — it's the fastest way to tell whether a B2B lead list is ready to send to. If you can't answer most of them, the campaign isn't ready and no copy will save it.

1. It was built for this campaign, not inherited from a previous one.
2. It was exported recently — weeks, not quarters.
3. Every record has been verified since export, not when it was bought.
4. Catch-all domains are separated into their own segment with their own plan.
5. You know the bounce rate you'll accept before you send, and where you stop.
6. The titles are real titles in that market, not a keyword guess that returns coordinators alongside directors.
7. Company-level fit is checkable — you can say in one sentence why each account belongs, and it isn't "they're in the industry".
8. Every enriched field is used somewhere in the message or the prioritisation.
9. There's a reason to contact them now for at least the top tier — a signal, not a hunch.
10. It's small enough to work properly. Forty accounts you understand beat four hundred you don't.

7 · Stack the sources — don't pick one

Everything above points at one conclusion, and it's what most first-time buyers get wrong: no single provider is enough, and choosing between them is the wrong exercise. One source is strong on company data and thin on contacts. Another has the contacts and misses the mid-market. A third is the only one catching funding and hiring the week it happens. Pick one and you inherit its blind spots as your own.

What works is layering. A primary source for breadth. A second called only on the misses, so you pay for it sparingly. A dedicated verifier on top of both. One or two signal feeds running separately, because timing data ages in days rather than months. This is what people mean by waterfall enrichment, and the reason it beats any single subscription isn't volume — it's that agreement between independent sources is itself information. When two unrelated sources give the same title, you can trust it. When they disagree, you've found a record to check rather than a record to send to.

Here's the part the pricing pages don't tell you: the subscriptions are the easy bit. The work is the glue between them — deciding which source is authoritative for which field, resolving conflicts, deduplicating across sources that format everything differently, handling retries when one API is having a bad day, keeping suppression lists so you don't email someone who already said no, and knowing from experience which provider is quietly unreliable in which market. None of that is a feature you can buy. It's accumulated judgement with some code around it.

Which is why agencies that have been doing this a while are usually running internal tooling you can't purchase. Not because it's being kept from you — because it isn't a product. It's built around specific campaigns, in specific markets, against specific providers' quirks, and generalising it enough to sell would strip out exactly the part that makes it work. Every serious outbound team ends up with some version of it. The question is whether you're going to spend a year building yours.

8 · Buy, build or outsource

So who builds it? Four routes, and the right one depends on volume and on whose time is cheap.

Buy a subscription when you're sending continuously and someone in-house genuinely owns the process — export, verification, hygiene. Seat-based tools are poor value used occasionally; the pricing assumes constant use.

Build by hand when the target list is small and high-value. For fifty accounts worth six figures each, manual list building beats any provider — you'll get better titles, better context and no bounces. This is the underrated option, and the one people skip because it feels unscalable. It doesn't need to scale if the deals are big enough.

Build the whole layer in-house if outbound is core to your business and you can fund what it really costs: subscriptions bought at list price before you know which you'll keep, underused seats, and a couple of quarters learning which source is reliable in your market by burning campaigns to find out. Then the part nobody budgets — someone owns hygiene every week, and when they leave the knowledge goes with them. Plenty of teams should still do it. Just don't start expecting a subscription and a fortnight.

Outsource when you want the layer without building it. Plainly: that's what we do, and this page exists partly because people arrive asking which provider to buy when the real question was the system around it. What you'd be buying isn't database access — it's the layer from section 7, assembled for your campaign and your market rather than for everyone. Different ICP, different stack.

Whoever you ask, judge them the same way: how do they verify, where does verification sit in the flow, what happens to catch-alls, how would they test coverage in your market, and which sources would they combine and why. Anyone answering with one provider name and a contact count is selling a list, not a pipeline. Cost comparison in how much B2B lead generation costs.

The takeaway

Buying B2B data is the cheap, commoditised part of outbound, and treating it as the main decision is why so many campaigns start badly. Every list is a snapshot already ageing. Coverage is patchy in ways no demo shows. Verification is a separate cost whatever you bought, catch-alls need their own plan, and enrichment earns its money only on fields that change the message. No single source covers a market, so the answer is a layer, not a purchase. And above all of it sits timing.

The good news is that this is a solved problem — it's just solved by a system rather than a subscription. And unlike most of outbound, it compounds. Once the layer exists, every campaign after it is cheaper and faster than the last: the sources are already chosen, the verification order already works, the suppression lists already exist, and you already know which provider to trust for which market. The first campaign pays for the setup. The tenth one runs on it.

So the question isn't which provider to buy. It's who's building your layer, and whether they've built one before. If you'd rather not spend a year learning this the expensive way, that's the part we run — tell us your market and we'll show you what the stack would look like for your ICP before you commit to anything.