Data Product Pricing
Data Product Pricing is how you set the price for datasets, feeds, APIs, and analytics products that you sell externally (or charge internally as chargeback). Five common models: (1) Flat subscription (Bloomberg Terminal: ~$30K/seat/year), (2) Volume-based per-query/per-record (AWS Data Exchange), (3) Tiered access (basic free, premium paid — Refinitiv), (4) Outcome-based (% of revenue lift attributed to the data), (5) Value-share (clean room or syndication deals where buyer and seller split incremental value). The right model depends on whether your data's value is concentrated (one-and-done insights) or recurring (operational dependency). Recurring use → subscription. One-time use → per-query. Everyone defaults to subscription because it's easy; many leave 30-50% of value on the table.
The Trap
The trap is pricing the data based on what it cost you to produce instead of what it's worth to the buyer. Cost-plus pricing for data products is structurally wrong because data has near-zero marginal cost — you're not pricing manufacturing, you're pricing utility. The other trap is single-tier pricing: when 90% of buyers want the cheap version and 10% would pay 10x for premium features, a single price either underprices the whales or excludes the long tail. Tiered pricing isn't optional for mature data products. Finally: never give a sales-led discount that can't be justified by deal size — discounting destroys reference pricing fast in data markets.
What to Do
Build a pricing model in three steps: (1) Quantify buyer value: interview 10 prospects about decisions the data informs and revenue/cost impact. Get a value-per-customer range. (2) Set list price at 10-25% of buyer value (your 'value share'). Below 10% you're underpricing; above 25% you'll lose deals. (3) Build at minimum 3 tiers — usually Starter (limited records/queries, free or $1-5K/yr), Pro (full data, $25-100K/yr), Enterprise (custom integration, SLA, $250K+/yr). Test annually with willingness-to-pay surveys and analyze deal-loss reasons monthly.
Formula
In Practice
Bloomberg Terminal has charged ~$24K-30K/seat/year for decades with minimal price changes. The price is set based on value to a trader (who generates millions in P&L), not on Bloomberg's cost to deliver bits. They refuse most discounts even at scale, preserving reference pricing. Result: ~325,000 terminals × ~$30K = ~$10B+ in annual revenue, 80%+ gross margin, and a near-monopoly in financial data despite open competition. The pricing discipline is the moat as much as the data itself.
Pro Tips
- 01
If your data product has fewer than 3 pricing tiers, you're leaving money on the table. The price gap from Tier 1 to Tier 3 should be at least 5-10x — this is what enables you to capture both SMB and enterprise buyers.
- 02
Run an annual willingness-to-pay survey using the Van Westendorp Price Sensitivity Meter. Ask 4 questions: at what price is it (a) too expensive to consider, (b) starting to feel expensive, (c) a bargain, (d) so cheap you'd doubt quality. The intersections show your defensible range.
- 03
Per-query pricing seems modern but creates buyer anxiety: customers throttle usage to control bills, which kills adoption. If your data is operational (used daily), default to subscription. Reserve per-query for one-and-done research use cases.
Myth vs Reality
Myth
“Lower prices grow data products faster”
Reality
False for B2B data. Buyers infer quality from price; underpriced data products are perceived as low-quality and excluded from RFPs. AWS Data Exchange data showed providers who priced in the bottom quartile had 35% lower conversion than mid-tier despite having competitive data quality. Pricing too low signals 'commodity' and limits TAM.
Myth
“Usage-based pricing is always more fair”
Reality
Usage-based pricing transfers revenue volatility from buyer to seller. It's only 'fair' if your costs scale with usage (rare for data — costs are mostly fixed). Snowflake's per-query model works because compute IS variable; for pure data feeds with fixed production cost, subscriptions are healthier for both sides.
Try it
Run the numbers.
Pressure-test the concept against your own knowledge — answer the challenge or try the live scenario.
Knowledge Check
You're pricing a B2B dataset that delivers ~$500K/year in value to a typical buyer. Your direct cost is ~$15K/customer/year. What's the most defensible list price for the standard tier?
Industry benchmarks
Is your number good?
Calibrate against real-world tiers. Use these ranges as targets — not absolutes.
B2B Data Product Value Capture %
B2B subscription data products with quantifiable buyer valuePremium (Bloomberg-tier)
20-30%+
Healthy
10-20%
Acceptable
5-10%
Underpriced
2-5%
Commodity Trap
< 2%
Source: OpenView Partners SaaS Pricing Benchmarks 2024 / Simon-Kucher Data Pricing Studies
Real-world cases
Companies that lived this.
Verified narratives with the numbers that prove (or break) the concept.
Bloomberg LP
1981-Present
Bloomberg Terminal has held its ~$24K-30K/seat/year pricing for decades, with strict no-discount policy even for the largest customers (Goldman, JPM). The price reflects buyer value: a single trader can generate millions in P&L, so $30K is < 1% of value created. Bloomberg refused volume discounts that competitors offered, preserving reference pricing. With ~325,000 terminals globally, this discipline produces ~$10B+ in annual revenue at 80%+ gross margin. Competitors who tried to undercut on price (Refinitiv, FactSet) gained share but never displaced the Terminal — proving that pricing discipline is itself a moat.
Price per Seat (2024)
~$30,000/year
Price Increases Over 40 Years
Modest (~CPI)
Active Terminals
~325,000
Estimated Annual Revenue
~$10B+
Premium pricing held for decades creates a moat. The price itself communicates indispensability. Discounting once destroys reference pricing forever — Bloomberg understood this from day one.
AWS Data Exchange
2019-Present
AWS Data Exchange enables data providers (Reuters, Foursquare, Dun & Bradstreet, etc.) to sell datasets via the AWS marketplace using subscription, free trial, or private offer pricing. Internal AWS data showed providers who priced in the bottom quartile had ~35% lower conversion despite competitive quality — buyers inferred low quality from low price. Providers offering 3+ pricing tiers had 2.1x higher annualized revenue per listing than single-tier providers. The platform proved that B2B data buyers reward signaled premium and tiered access.
Active Data Providers
300+
Conversion Lift: Tiered vs Flat
~2.1x
Bottom-Quartile Price Penalty
~35% lower conversion
Most Common Successful Model
Annual subscription, 3 tiers
Pricing tiers and price level are themselves quality signals. Underpricing a data product on a marketplace doesn't grow it faster — it gets it filtered out of consideration sets.
Decision scenario
The 'We Need to Hit Q4 Number' Pricing Decision
You run a 4-year-old B2B data product priced at $60K/year. 45 customers. ARR $2.7M. The CRO says you'll miss the $3.2M Q4 number unless you offer a 40% discount to close 5 'on-the-fence' enterprise deals. Each deal would be $36K instead of $60K — adding $180K and hitting plan.
List Price
$60K/year
Customers
45
ARR
$2.7M
Q4 Gap
$500K
Discount Requested
40%
Decision 1
The 5 deals would close at $36K each. But your existing 45 customers are at $60K, and at least 8 are up for renewal in Q1. Discount disclosure is hard to prevent in enterprise sales.
Approve the 40% discount — hit the Q4 number, deal with consequences laterReveal
Hold price at $60K. Offer a 3-month proof-of-value pilot at $5K to the 5 fence-sitters with conversion to full annual at $60K if value is proven.✓ OptimalReveal
Related concepts
Keep connecting.
The concepts that orbit this one — each one sharpens the others.
Beyond the concept
Turn Data Product Pricing into a live operating decision.
Use this concept as the framing layer, then move into a diagnostic if it maps directly to a current bottleneck.
Typical response time: 24h · No retainer required
Turn Data Product Pricing into a live operating decision.
Use Data Product Pricing as the framing layer, then move into diagnostics or advisory if this maps directly to a current business bottleneck.