How to buy data: a step-by-step guide for first-time buyers

Buying external data for the first time feels harder than it should. Providers describe similar products in different words, prices are often hidden, and the licence matters as much as the rows. This guide walks through the process step by step, from the business question to a signed contract and a working feed.

Step 1: Start from the decision, not the dataset

Every successful data purchase starts with a decision someone will make differently once they have the data. "We need company data" is not a requirement. "Our sales team needs to prioritise 4,000 mid-sized manufacturers in Germany and Poland by size and growth" is.

Write down three things before you talk to anyone:

• The decision or workflow the data will feed (lead scoring, a pricing model, an investment screen, a training run).
• The unit of analysis: companies, people in a business role, products, places, shipments, days, tickers.
• What "good enough" looks like: the coverage, accuracy and freshness below which the project fails.

This keeps you from buying a large, impressive dataset that does not answer the question.

Step 2: Write a one-page data specification

A short specification makes providers' answers comparable. Keep it to one page.

Coverage

List the countries, industries, entity types and time period you need. Be explicit about edge cases: does "Europe" include the UK and Switzerland? Do you need subsidiaries or only parent companies?

Fields

Separate must-have fields from nice-to-have fields. For each must-have field, note the format you expect (for example ISO country codes, NACE or NAICS industry codes, ISO dates).

History and freshness

How far back do you need the data, and how quickly must changes show up? A churn model may need five years of monthly history; a sales list may only need current records refreshed every month.

Delivery

Name the systems you will load it into: a warehouse share, files on cloud storage, an API, or a spreadsheet for a one-off analysis.

Use and licence

Describe who will use the data and how: internal analytics only, inside a product you sell, for outreach, or for training a model. This decides which licence you need, so it belongs in the spec rather than in a late-stage negotiation.

Step 3: Find candidate providers

Search by data type and country rather than by provider name. On DataStoa you can search listings and filter by category, coverage, delivery method and price model, or browse a category such as firmographic data or B2B contact data.

If nothing fits, or you want providers to come to you, post a data request with your specification. Providers reply with proposals you can compare side by side. Searching, contacting providers and posting requests is free for buyers.

Aim for a long list of five to eight providers, then cut it to three for detailed evaluation.

Step 4: Shortlist with a few sharp questions

Send each provider the same short set of questions:

1. Where does the data come from, and how is it collected?
2. What is your coverage for my exact scope, and where are the known gaps?
3. How do you measure accuracy and freshness, and when was it last measured?
4. Which licence would cover my described use, and what does it cost?
5. Can you run a sample on records I choose?

Answers that are specific, include numbers and admit limitations are a good sign. Vague answers to the first question are a reason to drop a provider, however good the price.

Step 5: Test a sample

Never buy on a description alone. Ask for a sample that matches your scope, ideally built from a list of entities you already know well, so you can check it against ground truth.

At minimum, measure:

• Match or coverage rate: of the entities you care about, how many are present?
• Fill rate for each must-have field.
• Accuracy on a hand-checked subset of 50 to 200 records.
• Freshness: how recent are the last-updated dates, and do they match reality?

Run the same tests on every shortlisted provider so the comparison is fair. Our guide to testing data samples covers this in detail.

Step 6: Understand the pricing

Data is priced in a handful of ways, and the same dataset can cost very different amounts depending on the model:

• Per record: common for contact and place data; check whether you pay for records delivered or only for records that match.
• Subscription (monthly or yearly): common for feeds and firmographics; price usually scales with coverage, history, refresh rate and number of users.
• Usage-based: API calls or lookups, often in tiers.
• One-off: a historical extract or a custom collection project.

Ask what changes the price. Adding a country, extending history or allowing use in a customer-facing product can each move the quote. Asking early prevents surprises later.

Step 7: Negotiate the licence and contract

The licence defines what you can do with the data, which is often more important than the price. Check:

• Permitted use: internal analytics, commercial products, outreach, AI training or retrieval.
• Users and entities: which teams and affiliates may use it.
• Derived data: whether your models, scores and aggregates remain yours after the contract ends.
• Termination: what you must delete when you stop paying.
• Personal data: who is the controller, what legal basis the provider relies on, and how opt-outs are passed on.

Our guide to data licensing explains these terms in more depth. Have your legal team review the final contract.

Step 8: Onboard, monitor and review

Treat the first delivery as a test. Load it, rerun your sample checks at full scale and confirm that the schema and documentation match what was promised. Then set up simple monitoring: row counts, fill rates and freshness per delivery, with an alert when they move outside the expected range.

Put a review date in the calendar about two months before renewal. By then you will know whether the data changed the decision you bought it for.

Common mistakes first-time buyers make

• Buying the biggest dataset instead of the most relevant one. Coverage of your segment matters more than total record count.
• Testing only the provider's chosen sample. Curated samples look better than average data.
• Ignoring the licence until the end. A great dataset you cannot use in your product is the wrong dataset.
• Skipping integration effort. Mapping identifiers and cleaning fields can take longer than the purchase itself.
• No owner. Someone must own the data relationship, the monitoring and the renewal decision.

Frequently asked questions

How long does it take to buy a dataset?

A simple one-off purchase can close in a week or two. Recurring feeds with legal review, security questionnaires and integration work usually take four to ten weeks, depending on your procurement process.

Do I need a data team to buy data?

Not always. Many providers deliver spreadsheets or dashboards for smaller projects. For ongoing feeds you will need someone who can load, check and monitor the data, even if only part time.

Should I tell providers my budget?

Sharing a range usually helps. Providers can then propose the right scope, such as fewer countries or shorter history, instead of quoting their full product or declining.

Is it free to find providers on DataStoa?

Yes. Buyers can search, contact providers, request samples and post data requests without paying DataStoa. Contracts, payment and delivery are agreed directly between you and the provider.