An Iterative, Outside-In Approach to Pricing Your Data

This article was written by Jennifer Belissent and originally appeared on the Snowflake Blog here:


When asked, “What’s my data worth?” I often provocatively answer, “Nothing.” But that’s only true if no one is using the data. Data has no intrinsic value, despite valiant efforts to define it. However, there are literally a million (or more) ways of deriving value from data—and that’s how we determine what data is worth. Data’s use determines its value, and that value establishes price.

You might know the characteristics of the data; for example, the data quality, the frequency of refresh, or the uniqueness of the data set. These do suggest potential value. If the data is ubiquitous and can easily be substituted, it might not have as much value. However, if the data is truly unique and could provide insights that few others have access to, then the price point might be higher. But, still, what is that price?

The challenge remains. When a unique data set comes to market, commercial value is unknown and price is elusive. Most companies starting to monetize their data find themselves in this scenario. Some set a price based on an estimated value and then adjust based on customer interest. If a flood of positive responses comes quickly, subsequent buyers might face a higher price. If prospects hesitate, negotiation might lead to a lower price. Through a process of trial-and-error, or target practice, they hone in on the “real” price, based on the anticipated value for the buyer. Trial-and-error pricing might work but isn’t particularly efficient, nor effective at building positive customer engagement.


A more deliberate, outside-in approach to data pricing starts with three questions:

  • What are customers willing to spend?
  • How have others priced similar data?
  • What value could customers derive from it?

Unfortunately the answers to these questions are often unknown. However, other parts of the equation might provide a starting point. In solving a math problem, you often start with what you do know, the “given” (for example, you know the length of one side of a triangle and the relationship with the other sides). In pricing of a data product or service, ways to solve for the unknown by identifying the “given” include:

  • Start with an internal use—revenue generation or cost avoidance. With a vast fleet of trains under management, Siemens Mobility knew that predicted and scheduled maintenance was cheaper than dealing with an unforeseen breakdown. An analytical model can predict those maintenance needs accurately, eliminating costly, unscheduled downtime. That cost of unscheduled downtime becomes the starting point for the price (and value) of the insights service. Better customer experience due to fewer breakdowns and higher availability of trains adds to the estimated value of the data service. When a customer can leverage data for use cases similar to those adopted internally, the internal value generated can be used as the “given” in the pricing exercise.
  • Test product offerings through proofs of concept (POCs) and value (POVs). GE Aviation had used its data to better understand the engines it built, and it realized that customers could benefit from that data to better understand engine operations. A POC tested the tangible benefits, and co-creation with interested customers refined the offering. GE Aviation now offers flight efficiency services, fuel management, and fleet management. Ultimately, the price of the product is based on value delivered.
  • Commission third-party research. A provider of revenue management software for healthcare providers allows customers to compare its key performance indicators to industry benchmarks, such as indicators from Medicare or Medicaid in the United States. To determine its starting price point, the company had a third party perform an ROI study. The methodology used customer input to estimate value and made assumptions about the product uptake to determine overall value to the company.
  • Leverage your partner ecosystem. When partners already leverage your data to improve their operations or better understand their customers, you can tap into their estimated value to enhance the value-prediction models. Retailers share sales with CPGs to improve demand forecasting. What forecast improvements have the CPGs seen? Waste eliminated? Fewer markdowns?
  • Adjust for uncertain outcomes. In the first instance, you need to adjust the results—the savings, or revenue generated—by the likelihood of achieving that outcome. Simply put, if the cost savings is $100 but the likelihood of the outcome is 80%, the expected value is only $80. Once you have a pool of customers using the data, the model for estimating the value in a given use case will be more robust. You can identify average savings, or estimate the likelihood of achieving those savings.


The next step is to demonstrate that the price reflects the value delivered. A couple of options exist to enable exploration:

  • Show-and-tell. If the prospective buyer isn’t familiar with the use case or doesn’t know the potential business value to be delivered, a demonstration might be necessary—a POC to illustrate how to use and derive value from the data. GE Aviation developed its predictive maintenance solution pricing by having the data team work with existing customers. This type of engagement can lead to product co-development with the client. Similarly, Amadeus worked with Qantas Airlines to develop and deliver a disruption management solution to mitigate the effects of weather or operational disruptions.
  • Try-and-buy. A variant of show-and-tell provides a brief trial option to demonstrate the value of the insights. This try-and-buy model gives a potential buyer the opportunity to build a business case for the data acquisition. Currently in private preview, the Try-And-Buy feature of the Snowflake Data Marketplace enables potential customers to test the impact of data sets from Crunchbase, Knoema, SafeGraph, and others.


There is no reason to obsess over theoretical valuation models requiring a laundry list of assumptions about ill-defined data attributes. Savvy executives will abandon that futile exercise and focus instead on identifying the best use of their data and the insights derived from them. Your use cases and those of your partners and customers will determine the value, and ultimately the price of the data product or service.

Whether the “given” in this pricing equation starts with your own cost savings or revenue generated or is based on a POC, the process is interactive and iterative—a user-centric, agile approach to pricing. Test the price in the market, review and revise, and repeat. Once you have a pool of customers using the data, the model for estimating the value in a given use case will be more robust and the price more accurate.