How to price your data product

Pricing is where many data products stall. Set the price too low and buyers assume the data is thin or stale; set it too high without explanation and they never reply to your quote. Data pricing follows a few recognisable patterns, and most of the work is understanding how your buyers use the data and making that visible in your offer.

This playbook walks through how to choose a pricing model, how to build tiers, how to price usage rights, and how to present prices on a marketplace listing so the right buyers contact you.

Start from the buyer's value, not your cost

Your collection and processing costs set a floor, but they say little about what a buyer will pay. Buyers compare your price with the alternatives they have and with the value the data creates for them.

What does the data replace?

List the options a buyer has without you:

• Building the dataset in-house: engineering time, infrastructure, legal review and ongoing maintenance
• Paying analysts to research the same facts manually
• Buying a broader dataset from another provider and filtering it down
• Doing without, and accepting slower or worse decisions

If a team would need two engineers for six months to approximate your feed, your annual price can sit well above your marginal cost and still be an easy decision for them.

What does the data enable?

Then think about the outcome for a typical customer. A sales team might win a handful of additional accounts. A lender might avoid a few bad loans. A retailer might choose a better site. You do not need precise numbers; you need a credible story that your price is a small fraction of the value at stake. Write that story down, because you will reuse it in your listing and your sales calls.

Match the pricing model to how the data is used

The model matters as much as the number. Pick the one that mirrors how buyers consume the data.

Model	Works best when	Watch out for
One-off purchase	Historical extracts, research projects, list builds	Buyers returning for refreshes at one-off prices
Monthly or yearly licence	Feeds that update regularly and sit inside workflows	Churn if buyers do not see the value of each update
Per record	Contact lists, enrichment, POI or address files	Buyers selecting only the easy records
Usage-based (API)	Lookups, verification, real-time enrichment	Unpredictable bills that worry procurement
Custom quote	Large, multi-country or redistribution deals	Losing buyers who want a number before talking

Many providers combine two models: a yearly licence for the full feed, plus per-record or one-off pricing for smaller buyers. That gives a startup a way in without discounting your core product.

Build tiers from the dimensions buyers care about

Tiers let you serve a small team and a global enterprise with the same dataset. Good tiers change along dimensions that genuinely change the value:

1. Coverage: number of countries, sectors or entities
2. History: one year versus ten years of backfill
3. Freshness: monthly files versus daily updates or real-time API access
4. Delivery: file download versus a cloud data share, an API or a managed pipeline
5. Usage rights: internal analytics only versus use in commercial products, redistribution or AI training
6. Scale of use: seats, legal entities, API calls or end customers

Avoid tiers that differ only by arbitrary limits that cost you nothing to lift. Buyers notice, and it invites hard negotiation.

A worked example

Imagine a provider of company registry data for Europe. The structure below is illustrative, not a benchmark.

• Single-country extract: one country, current snapshot, CSV, internal use only. Sold as a one-off purchase to consultants and small teams.
• Professional feed: all covered countries, monthly updates, five years of history, cloud delivery, internal use. Sold as a yearly licence.
• Platform licence: daily updates, full history, API access and the right to show the data inside the buyer's own software. Sold by custom quote with a minimum annual commitment.

The data is the same; the value to each buyer is not. Each step up adds something the next buyer segment actually needs.

Price usage rights explicitly

Usage rights are the most underpriced part of many data products. Internal analytics, use in a commercial product, redistribution to the buyer's own customers and AI model training are very different licences. A buyer who embeds your data in software they sell is building revenue on your asset; a buyer who trains a model may keep the benefit after the licence ends.

Write each right into your listing and price it separately or by tier. If you have not decided on a right, say it is available on request rather than leaving it ambiguous. Ambiguity costs you twice: careful buyers hesitate, and the ones who do sign may assume broader rights than you intended.

Publish a starting price

On a lead-generation marketplace, your listing is your first sales conversation. A clear "from" price lets buyers qualify themselves before they write to you, so the inquiries you receive come from teams with a matching budget. If your deals vary widely, publish the price of your entry tier and explain what moves the price up: more countries, more history, faster delivery or broader rights.

"Price on request" still has its place for complex, multi-country or redistribution deals. Even then, a sentence describing what drives the quote helps buyers judge whether to get in touch. You can see how plans and listing fields work on DataStoa on the seller pricing page.

Use samples, trials and discounts deliberately

• Samples reduce risk for the buyer and cost you little. A representative sample in your real schema is usually worth more than a discount.
• Trials suit API and feed products whose value shows up over time. Time-box them, agree success criteria up front and limit rights to evaluation.
• Discounts work best when they buy you something: a multi-year commitment, prepayment, a reference customer or a faster signature. A modest, published discount for annual prepayment is easier to defend than ad hoc reductions.

Common pricing mistakes

• Pricing by row count alone. A million stale rows are worth less than a hundred thousand accurate, current ones. Price the coverage and quality buyers need, not raw volume.
• One price for every use. The same licence for a two-person research team and for a platform reselling your data leaves money on the table and invites disputes.
• Hiding renewal terms. Buyers budget for year two. Say whether prices can rise, how much notice you give and what happens to data already delivered if they stop.
• Ignoring currency and tax. If you sell worldwide, state the currency, whether taxes are included and whether you can invoice in the buyer's currency.
• Never revisiting prices. Your coverage, history and quality improve every year, so review prices at least as often.

Review prices with evidence

Treat pricing as an experiment. Track how many listing views turn into inquiries, how many inquiries turn into deals and where deals stall. If many buyers drop out after seeing your quote, the issue may be packaging rather than price: perhaps they need a smaller entry tier. If almost every buyer accepts your first number, you are probably too cheap.

Buyer requests are another useful signal. Browsing open data requests shows the budgets buyers state for the kind of data you sell, in their own words.

Frequently asked questions

Should I publish prices or ask buyers to contact me?

Publish at least a starting price when you can. It helps buyers judge fit before they write, which tends to improve the quality of the conversations you have. Keep custom quotes for complex scopes, and explain what drives them.

How do I price a dataset that has no direct competitor?

Anchor on what it replaces: the cost of building it in-house, of manual research or of a broader dataset that needs heavy filtering. Then test a price with a few buyers and adjust based on how they react.

Should AI training rights cost more?

Usually, yes. Training can create lasting value from your data, and models may outlive the licence term. Price AI training as a separate right or a higher tier, and state clearly what happens to trained models when the licence ends.

How often should I change my prices?

Review them at least once a year and whenever your coverage, history or delivery changes materially. Give existing customers notice before renewal, and keep their current terms for a period where that protects the relationship.