In early September, I had the opportunity to travel to Berlin to attend the Data Natives conference. This three-day event focuses on all things data science and Internet. The event has 4 stages with start-up pitches, revolutionary ideas about the impact of data on society and in-depth technical presentations. There was something for everyone. There was even a full day dedicated to the ethics and technology of Web3. I learned a lot there and would like to share some surprising (and philosophical) insights here. I wonder; will data continue to be the new gold?

The cost of data

In our work, we collect huge amounts of data every second. Every click, scroll, impression, exit intent, view, input or error is recorded. We capture petabytes of information with the grand ambition of understanding visitor behavior and ways to manipulate our websites' features or layouts to increase conversion rates and revenue. The data science revolution has shown that data, when used in the right way, can contribute immensely to business success. Data-driven work is the new gold standard. Yet obtaining this data is not a cheap or easy task. We need experts to set up website tracking to measure the behaviors we want to investigate. Moreover, we need to store all this data. Most of us will use the cloud to do this. This means storing all this data in huge server centers somewhere in the world.

But do we really need all this data?

This question came to mind during one of the presentations at Data Natives. In an interesting presentation, Lior Barak explained the need for a Return Of Investment (ROI) model on data points, and not just on ‘data’ in general. He argued that now that we have demonstrated that data has a high ROI, we can become more critical and start optimizing this ROI model. Do we really need all the data we store? Tracking and storing data is expensive.

Barak therefore argued that we need to determine the ROI of each measurement rather than the data as a whole. If we use the measurement in one of our predictive models or if it contributes significantly to our insights, it is more valuable than if it is just sitting on a server, not really contributing to anything and just taking up space that you pay for. He therefore proposed a value-based scheme in which each measurement that was stored was given its own ROI. This way it is easier to decide which data we want to store and which we don't. This makes us more efficient in our data storage and data use. Moreover, if the data collected are personal data, this approach would also be more in line with the General Data Protection Regulation (GDPR).

I think we need to be more critical about how we collect and store our data. Do we really need to store everything Google Analytics collects? Can we be more data efficient? I think our measurement plans should include not only what to measure, but also what not to measure. If this is still too rigorous, as an alternative, a measurement plan can already include a proposal of how long we keep data. Think about it, is 6-year-old data still going to give us insight into the experiences of our current visitors?

Reaping the benefits of data

A second major topic at the conference was about the privatization of data. Several presentations addressed the next step in the data revolution. Discussions about data governance, monitoring the use and misuse of data, and ensuring that profits benefit the right people prevailed during the conference.

The last day was entirely devoted to the ethics and technology of Web3. Web3 is generally seen as the third major iteration of the World Wide Web. In his talk, Jan Sell (Coinbase) described these iterations as follows:

• Web1: Read. This is the original worldwide web in which websites were built by professionals to be consumed by the public read-only.
• The second iteration, Web2: Read & Write, is the social revolution where platforms that allow the public to not only read but also produce content dominate (Write).
• The future third iteration, Web3: Read, Write & Own, is the decentralization revolution. Web3 shifts from the large centralized platforms on the Internet to private platforms in which all users own their own data.

Web3

Web3 empowers the public to benefit from their data. The industry has shown that user data is of incredible value. In addition, the public is becoming increasingly aware of the unfairness in the distribution of profits. As an Internet user, you leave your data on almost every website you visit (once you accept cookies). A transaction takes place in which user data is collected in exchange for (free) use of the platform or website. Using this data, BigTech and e-commerce companies have successfully increased their revenue by huge amounts. However, this wealth is not shared directly with the people who contributed their data. Consequently, there is a growing sense that this original deal is unfair. The profit generated from a user's data far outweighs the cost incurred by the user to use the platform or website.

Web3 provides a fairer distribution. The technology allows for better ownership of your data and to have more control over who uses it and at what price. The panel discussion between Gilbert Hill, Robin Lehmann and Merlene Ronstedt showed how Data Unions are already taking the first steps in teaching people to become more skilled at taking ownership of their data. They argue that as a user who generates data, you should be able to share in the (financial) gains that this data generates. This is also in line with the AVG, which provides a legal ground to have a greater say in how your data is used.

The future

If we extend this trend into the future, the introduction of Web3 will have major consequences. When companies have to charge all users for the use of their data, user data becomes an expensive commodity. This will also have major implications for online experimentation. User data is the basis of our online experimentation, and with rising costs, we will have to become more sophisticated in the use of this data.

I believe that, with the Web3 era upon us, using the ROI model described above becomes even more attractive. It gives us a first step in determining what data points/types of data we would be willing to pay for in the future, in addition to the direct benefit of lower costs. Yet, we can also consider other methods that can reduce the amount of data we use. For example; targeted segmentation, which collects data only from the types of users you are interested in, thereby reducing the amount of data we collect. Alternatively, the field of data science has shown greater skill in generating accurate insights and predictions from smaller and smaller samples. Exploitation of these techniques can also help us become more sophisticated in our use of data.

As an experimentation industry, I think we should treat data as the new gold. Data, like gold, is an increasingly expensive commodity with a stable value proposition. Thereby, the value of data, like gold, will remain a social construct: it is valuable because we all agree it was and will be in the future.