How to Build a B2B Lead List From a CRM Export

Your first list should not start with a database subscription. At a company under fifty people, the fastest way to build a B2B lead list is to export the accounts you already closed and work outward from what they have in common.

That export is not a list yet. Contacts are stale, the industry column is free text, and revenue data is missing entirely. Four passes fix it: enrich the rows, qualify them against your ICP, rank the result, then push the top slice to your CRM and a sequence.

Build a B2B lead list from lists you already have

Most b2b lead list building advice opens at a provider database. Set your filters, pick a headcount band, export a hundred thousand contacts. That approach works and plenty of teams run it well. It is also the point where a founder without a data budget closes the tab.

There are four or five lists sitting in systems you already pay for, and each one carries more signal than a filtered database query.

Start with closed-won. Not because those companies will buy twice, but because they define the pattern everything else gets measured against. Closed won lookalike accounts are the rest of the market filtered by the traits your paying customers turned out to share, which beats the ICP written on slide nine of the seed deck.

We built our own first list this way, out of a sign-up table nobody had opened in months. The qualification pass cut it by roughly two thirds before anyone wrote a single email, which felt wasteful right up until the replies started coming back from the third that survived.

While we are here: the TAM spreadsheet needs to stop. Forty thousand rows, eleven filter tabs, a headcount column nobody trusts, and no evidence anyone ever worked past row 300. It is planning theater. A ranked list of 200 accounts with a written reason attached to each one is worth more than the whole file.

What the export needs before you build a B2B lead list

Match rates and score quality are both decided by what you bring, not by which provider runs. Pull these columns out of the CRM:

• Company domain. The single most important column. "Acme Inc." is a string thousands of companies share a version of; acme.com is an identifier that providers can actually match on.

• Outcome and close date. Won, lost, or churned, plus when. This is the evidence the AI pass reads later.

• What they paid. Contract value or MRR if your CRM holds it. If it does not, connect your billing tool as a second source and the number lands next to the row anyway.

• Any contact, even a stale one. A name from three years ago still tells you which job title bought.

Then the load itself:

1. Connect your CRM as a source and filter to the outcome you want, or import the export file directly if you would rather work from the CSV. The same path covers a conference list, and the mechanics are the same ones in how to enrich a CSV of leads.

2. Set the dedupe key to company domain so a company that appears as three deals arrives as one row.

3. Keep the outcome columns visible. Most people drop them at import and then wonder why the scoring pass has nothing to reason from.

Enrich every account with firmographics and contacts

A lead list from CRM data is mostly holes at this point. Old contacts, missing headcounts, no current tech stack. Filling those is the pass that turns rows into something addressable.

No single data provider covers everyone, so the waterfall tries them in order and stops at the first match. You do not build the chain. Providers are ordered by cost and hit rate, the cheapest one likely to answer runs first, and a miss costs nothing.

At $0.02 a credit, and those figures assume every row matches, which no list ever does. The total appears before the run, so an expensive column can be dropped while dropping it is still free.

Two sources most people forget on this pass: your billing tool, which puts real contract values on the closed-won rows and makes the ICP definition quantitative instead of vibes, and your support desk, which knows which of the churned accounts left angry and which just ran out of runway. Both connect the same way the CRM did.

Qualify the lead list against your ICP with AI

Now the pass that changes the list rather than widening it. You describe your ideal customer once, in a sentence or two, and the AI column reads the entire table and scores every account against that description.

Whole-table reasoning is the part that matters. A per-row lookup can tell you a company has 80 employees. It cannot tell you that this account resembles your three best customers more closely than the 400 rows underneath it, because ranking is relative and it can only see one row.

Write the description from evidence, not aspiration. Something like: "Seed to Series A B2B SaaS, 20 to 150 employees, sells to sales or marketing teams, already runs a CRM and a billing tool, and pays us above $8,000 a year." Every clause in that sentence should trace back to a column in your table.

What comes back is a score and a reason for each row. The reason reads like a note from a junior analyst: this one matches on size and category but has no CRM in its stack, that one is a category fit but its funding round closed in 2021. Reps skim scores and act on reasons, which is also how you catch a scoring rubric that has quietly gone wrong.

Rank the list and cut the accounts not worth working

Sort by score and the shape of the list appears immediately. There is usually a clear top band, a long middle, and a tail of accounts that got into the export because someone once scanned a badge.

Cut the tail. Then read ten reasons from either side of wherever you drew the line, because that is where the model's judgment is weakest and where you will find out whether your ICP sentence said what you meant.

Back to the domain column for a second, because it shows up again here. Rows that missed enrichment tend to sink to the bottom on score too, not because the account is bad but because there was nothing to score. Filter for empty firmographics before you cut anything, fill the domains by hand, and re-run just those rows. It is the one manual hour in this whole guide worth spending.

Build a B2B lead list your CRM and sequencer can use

A ranked table nobody works is a nicer version of the spreadsheet you started with. The last pass gets it back out:

• To the CRM. Map score, reason and tier onto properties. Properties that do not exist yet are created with the right field type, and two-way sync keeps them current rather than frozen at export time. The same write-back pattern powers HubSpot lead scoring.

• To a sequence. Enrol the top slice in email and LinkedIn steps on one canvas. Every enriched column is available for personalization, so the research that ranked the account also feeds the message that reaches it.

• On a schedule. Re-run enrichment monthly. People change jobs, companies raise rounds, and the list decays whether or not you look at it.

Export your closed-won accounts and run them through. The free plan covers 200 credits a month and tables of up to 100 rows, which is enough to score a real sample end to end, and credits never expire. Get started free.