Power and economics around data monetization can be conflicting with ethical principles of the use of data. While the economics of monetizing data is progressing rapidly, there is relatively more minor progress in the space of paying the data subject for the data (also called ‘Pay for Data’).
Traditionally the approaches of using available natural resources have evolved in 5 stages (1) exploitation, (2) consent to extract, (3) approval from the regulator for access, (4) contribute to deprived societies that are dependent on such resources or region in which such resources exist and (5) re-establish ecology by reinstating or taking efforts to bring balance. Data ethics would essentially follow the same route. It is unlikely we will have a different outcome, as it’s a trend of human evolution.
User relationship with technology
Before delving into the data subject, an economic analysis must include understanding the layered nature of the relationship between people, services, platforms, and media owners. Lavattuer, the executive director of the Me2B alliance, provides a well-illustrated overview of the various relationships that exist, such as Me2P (the user’s experience with the platform or service provider) and Me2T (the user’s relationship with the platform or service provider as an experiential relationship) (User relationship with technology enabler). The basic premise is that multiple businesses may use the data subject’s data or behavior. Therefore, understanding these relationships better can help one consider the economic value of using this type of data.
Shifting perspective from ethics to economics
Given the rise of concerns around data use and issues caused by algorithms that use such data, we are at an inflection point of thinking about Pay for data. ‘Pay for Data’ model is to bring the conversation from ethics to equity to economics. A monetization mechanism for data subjects brings equity to the whole economic equation and presents a data subject choice.
In general, the ‘Pay for data’ model is non-existent in the mainstream economic context. In most cases, availing of a service for providing personal information appears to be the norm. However, this is expected to change in the coming years with the increased regulation/ activism on data use and privacy.
Future beyond data monetization
Data privacy regulations necessitate data subjects’ consent and demand appropriate data use. Data use should not be overlooked; in particular, data use may be generic, making quantification difficult (e.g., the data will improve user experience on the platform). Regulations such as the CCPA provide a starting point and an opportunity to monetize data, but we will need to go beyond that to meet future data monetization goals. People have different opinions on how much data is worth. This is primarily due to specific approaches, such as those used in the financial services industry, which offer cash incentives in exchange for the voluntary submission of personal information (typical in the context of customer data and preference acquisition).
It is necessary to understand that attributing economic value to data based on nature (e.g., protected categories) may not be the best way, as even a tweet (Bitcoin tweet by Elon Musk) may have more monetary value than a protected attribute. This aligns with a common understanding that data is worth the context and domain in which it is applied/ collated.
Pay for data model – economic benefits
When a person is wondering about the economic benefits of Pay for Data or the costs of complying with privacy and tracking the payment for data, it is equally essential to comprehend the processors. Therefore, businesses will seek out other business models if pressed with (a) an economic proposition to Pay for Data and (b) the economic costs of compliance and tracking payment for data. Alternative approaches would include the use of (a) Synthetic data and (b) advanced models that can perform with limited data, besides collecting necessary/ selective data through the ‘Pay for Data’ model (based on critical need). Gartner’s study mentions that by 2024, (1) 60% of the data used for the development of AI and analytics solutions will be synthetically generated, and (2) the use of synthetic data and transfer learning will halve the volume of accurate data needed for machine learning (here). These approaches can be applied in parallel with the use of privacy-preserving methods in model development.
Approaches that focus on addressing privacy and data use have significant benefits. By only gathering relevant user data instead of gathering all the data available, you have more space to delve into monetizing data. All these improvements also address the challenge of data insufficiency in specific environments (e.g., Financial crime prediction). For instance, Ealax, a company in the UK, is building synthetic data to support financial crime investigations.
Sustainability provides equity
This economic model will provide equity for the stakeholders and bring the conversation to economics. However, the model is not flawless. Some of the issues that arise from such an approach include (a) Synthetic data has a possibility of creating bias in the data if the diverse data inputs are not appropriately considered and (b) using synthetic data would mean there shall be close attention towards model monitoring for drift. In addition, to make such a ‘Pay for Data’ concept to work, there are two critical necessities. They are (1) regulations that mandate the Pay for Data model and (b) incentives for investments in social businesses which adopt such a model. These efforts would make the Pay for Data model sustainable.
Needless to say that establishing the ‘Pay for Data’ model is just a start. Regulations, governance, valuation, and other aspects will evolve to make the practices robust over time.