Company contact database: build your own instead of renting one

Every outbound team eventually hits the same wall: a clean list of target accounts, and no way to reach a single human inside them. The reflex is to shop for a company contact database, pay for access to a few hundred million records, and hope the people you need are in there and still reachable. There is a second option that almost nobody writes about, because nobody sells it: assemble the database yourself, account by account, at the moment you actually need it. This guide covers what a company contact database is, what renting one really costs compared to building one, and the exact five layer workflow to assemble your own in a spreadsheet.

What is a company contact database?

A company contact database is a structured set of records that links organisations to the people inside them: the company on one side (domain, industry, headcount, location), the person on the other (name, role, seniority), and the reachable details that connect the two (professional email, phone, profile URL). What separates a real database from a spreadsheet export is the join. A flat list of names is a list. A database keeps the account and the person addressable as two related objects, so you can ask questions like "every head of finance at the 40 accounts that just started hiring" instead of scrolling.

It's also not a CRM. A CRM is a system of record for relationships you already have: deals, notes, owners, stages. A company contact database sits upstream of that. It's the raw material you feed into the CRM once a contact is worth tracking. Confuse the two and you end up with 40,000 rows nobody has ever emailed, which is expensive in seat licences and poisonous for reporting.

The third distinction is provenance. A usable record carries where the data came from and when it was last confirmed. Without that you can't tell a verified email from a guessed pattern, and you'll find out the difference when your campaign bounces. Independent testing of B2B providers puts real accuracy in a 63% to 91% range and coverage anywhere from 26% to 92% depending on the field and the region, as our 2026 provider benchmark documents. No vendor sits at the top of both. That gap is the single most important fact to hold in mind before you spend anything.

Buy a company contact database, or build one?

Renting means paying for a seat on somebody else's index. You get instant scale, a search interface, and the comfort of a big number on the pricing page. You also get a bill sized to the whole index rather than to your use of it. If your ideal customer profile is 300 companies in three countries, a 500 million record subscription means you're financing 499,999,700 records you will never open. That's not a criticism of large databases. It's a question of fit: breadth is only worth paying for if your targeting is broad.

Building means starting from your account list and resolving contacts on demand. You accept that it takes a workflow instead of a search box, and in exchange every record you pay for is a record you asked for, freshly resolved on the day you asked. It also sidesteps the coverage lottery. A global index is rarely global in practice, and match rates swing hard by market, which we break down in our guide to geographic coverage and match rates. Resolving on demand lets you see the miss immediately on the account that matters, rather than discovering it as a silent hole in a bulk export.

Most mature teams end up layered, and there's nothing wrong with that. The mistake is defaulting to a subscription before you know how many records you actually touch in a month. If you've never counted, count first. The criteria worth weighing before signing anything are laid out in our guide on how to choose a data enrichment provider.

The decay problem every contact database inherits

Whichever route you take, the data starts dying the moment it is written. Roughly 30% of B2B records become unusable every year: people change jobs, companies rebrand, domains move, teams reorganise. That figure is not a vendor problem, it is a property of the world the data describes, and we cover the mechanics of fighting it in our page on database enrichment.

For a rented index that's uncomfortable: what you're buying is a snapshot, and its accuracy peaks on the day the vendor crawled it, not the day you export it. For a self built database it's simpler but not free: whatever you assemble in January is materially worse by December unless you refresh it. The difference is that a build on demand workflow makes refreshing cheap and targeted, because you already know which accounts matter this quarter.

One practical rule falls out of this. Never treat contact data as an asset you own. Treat it as a perishable input you resolve close to the moment of use. A verified email pulled the week you send beats ten records that have been sitting in a sheet since last spring, and it costs less, because you're not paying to store and clean the nine you'll never use.

The five layers of a usable contact database

Assembling your own sounds vague until you break it into layers. Each one answers a different question, and each has a distinct cost, so you can stop at whichever layer your use case needs.

Layer 1, the account list. Which companies do you want to reach? This is a strategy decision, not a data one. It can come from a CRM export, a market map, a conference attendee list, or a single well fitting customer expanded into lookalikes. Layer 2, firmographics. What is true about each company: size, industry, location, domain, tech in use. This is what makes segmentation possible later. Layer 3, the people. Who inside each account holds the role you sell to, and are they still there. Layer 4, contact details. The verified email or phone that makes the person reachable at all. Layer 5, signals. What just changed at the account that makes this the right week rather than some week.

Layers 1 and 2 are cheap and durable. Layers 3 and 4 are where the money goes and where decay bites hardest, which is exactly why resolving them on demand beats stockpiling them. Layer 5 is what turns a static list into a reason to reach out. Where each of these layers can legitimately come from, and the tradeoffs between primary, secondary and aggregated inputs, is the subject of our page on B2B data sources.

Build a company contact database in Google Sheets

The whole thing fits in a spreadsheet. That surprises people more than it should. Derrick runs as a sidebar inside Google Sheets, not as a wall of formulas you have to learn. You pick a feature, point it at the column that holds your input, and it runs down the rows and writes the results into new columns. The sheet stays a sheet, so the person who owns the list doesn't need engineering help to maintain it.

Start with your accounts. Paste company names or domains into column A. If your list is thin, Find Similar Companies turns one good customer into a whole lookalike set at 1 credit per company, and it is available from the free plan.

Fill in the firmographics. Enrich Companies appends the available company information at 1 credit per company, billed per row processed. It is available from the free plan, so you can build and segment layer 2 of your database before spending anything on people.

Resolve the people. Find a company's people lists current and former staff for each account, optionally filtered by job function, at 1 credit per person. This is the step that most spreadsheets never get past, and it is the one that turns an account list into a contact database. It is available on paid plans.

Make them reachable. Email Finder returns a verified professional email at 5 credits per email, billed only when an address is actually found, so unsuccessful lookups cost nothing. It's a paid plan feature. Verification happens before the address comes back, so you're not pouring guessed patterns into a sequence and discovering the truth through your bounce rate.

Clean before you use. Find Duplicates is unlimited and available from the free plan, which matters more than it sounds: the fastest way to burn goodwill is to email the same person twice from two rows that were never reconciled.

The same five steps work on 30 rows and on 30,000. Nothing about the workflow changes with volume, which is the point. You shouldn't have to graduate to a different tool because your list grew.

What a company contact database actually costs to build

Nobody publishes this number, so here it is with real prices. Take a realistic quarter of outbound: 200 target companies, two decision makers to reach at each, so 400 people. Run the five layer workflow above and the credit ledger looks like this.

2,200 credits for 320 verified, reachable contacts. That is about 6.9 credits per usable contact. On the Mini plan at 9 EUR per month for 4,000 credits, the whole quarter of list building costs roughly 5 EUR and leaves you 1,800 credits of headroom. On the Pro plan at 175 EUR per month for 100,000 credits, the per credit floor is lower still, and unused credits roll over on paid plans rather than expiring. The free plan gives you 100 credits per month at no cost, enough to run the workflow end to end on a handful of accounts and see the output before committing to anything.

The number worth keeping isn't the total. It's the ratio. You paid for 320 contacts because you wanted 320 contacts. Compare that against any per seat subscription by dividing its annual price by the number of records your team genuinely opened last year. Teams that do this arithmetic honestly are usually surprised, and our breakdown of what enrichment tools actually cost walks through the pricing models that hide the real figure.

Keeping the database fresh after you build it

A built database is not a finished database. Given the 30% annual decay figure, a sensible cadence is to re verify the contact layer on any account before you touch it, and to refresh firmographics quarterly on the accounts still in play. Everything else can wait. Refreshing a record you won't use this quarter is spend without a return.

Signals change the shape of this work. Rather than refreshing on a calendar, you refresh on an event. Signal tracks leads and accounts for job changes, funding rounds, hiring sprees and tech stack moves, and fires an alert the moment something happens. It is available from the Standard plan at 20 EUR per month, and it is billed per signal that actually fires, so a quiet month is a cheap month. For a lighter touch, Company Hiring Signal shows which companies are hiring and for which roles, at 1 credit per company, and it is available from the free plan.

Your database stops being a static asset that rots and becomes a queue that reorders itself. A contact whose job just changed is both the record most likely to be wrong and the person most worth contacting this week. Handling that as one workflow rather than two is what separates a maintained database from an archived one.

API, MCP and automation for a database that maintains itself

The spreadsheet is the default surface, not the only one. Derrick exposes the same enrichment engine through a REST API, available from the Standard plan at 20 EUR per month, so the resolution steps above can run inside your own product, a nightly job, or a CRM sync without anyone opening a sheet. That is how a contact database moves from something a person maintains to something a schedule maintains.

For AI driven work, Derrick MCP lets Claude, ChatGPT and any MCP compatible assistant call the same enrichment tools directly, from 20 EUR per month. An agent researching an account can resolve the right person and their verified email mid task, instead of handing the job back to a human. And because there is a REST API underneath, Derrick connects to Zapier, Make and n8n alongside 3,000 or more other applications, so enrichment can be triggered by a form submission, a CRM stage change, or a webhook from anywhere in your stack.

Four surfaces, one engine: the sidebar for the person who owns the list, the API for the pipeline, MCP for the assistant, and no code automation for everything in between. A contact resolved by a nightly job and a contact resolved by a rep in a spreadsheet come back identical, which is what keeps a database coherent as more of your team starts feeding it.

Five mistakes that ruin a contact database

Building before defining the account list. Enriching a vague list produces a large, expensive, unusable file. Layer 1 is a strategy decision and it deserves an hour of thought before a single credit is spent.

Treating unverified addresses as contacts. A pattern guessed from a name and a domain looks like an email and behaves like a bounce. Bounces damage sending reputation across every future campaign, so the cost lands long after the mistake.

Hoarding. Storing 40,000 records against 300 you will contact is not thoroughness, it is a maintenance bill you signed without reading. Resolve close to the moment of use.

Refreshing on a calendar instead of on events. Quarterly refreshes spend evenly across records that matter and records that do not. Signals concentrate the spend where something actually changed.

Skipping deduplication. The same person arrives twice through two spellings of a company name, and both rows get emailed. Deduplication is unlimited and free on Derrick, so there is no excuse for this one.

Get those five right and a self built company contact database beats a rented one on the metric that matters, which is not record count but the percentage of your outreach that reaches a real person in a role you actually sell to. Start free with 100 credits per month and build the first layer of yours today.