The Problem with Data Monetisation

Why some data companies succeed while others struggle.

Photo by Sara Kurfeß on Unsplash

Background

I have several friends who founded, or have managed, businesses focused on “selling data”. Actually, that sounds crude. What they do is sell the information contained within the data. This information can take the form of processed data feeds as inputs into business decisioning, or inputs into reporting or benchmarking. Some of these data businesses existed in the pre-digital days when the founders had to unlock information in analog datasets. The exponential rise in the production of digital data has tempted many to consider a start-up in data monetisation. But how do you know if you have a viable business model? What are the considerations that “make or break” the business model? And so I dedicate my 25th weekly article to discussing the challenges with monetising 3rd party data.

(I write a weekly article on bad thinking and bad practices in data analytics / data science which you can find here.)

Examples of Monetised Data

Allow me to give you 2 examples of data monetisation from my past life as a CAO in banking to set the stage for unpacking the “make or break” considerations. During my early career with Citibank in the 1990’s to 2000’s, I was exposed to Experian’s Mosaic solution for marketing segmentation. In this earlier iteration, it was a geo-level database on likely (probability-based) customer profiles such as lifestyles, affluence, preferences. The data was sourced from surveys and various raw data purchases and intelligently stitched together. The end data product was useful as an overlay to shape direct marketing activities.

Another classic set of monetised data is the famous credit bureau used in the financial services industry. The data is essentially sourced from participating financial institutions (either through regulatory mandate or mutual agreement) to achieve a holistic view of a person’s borrowing profile and behaviour across each of these participating financial institutions. While none of the data in the credit bureau is proprietary, the ability to combine data across the industry (which no single financial institution could do on their own) was essential to managing credit risk.

Now consider the following generalised example of data monetisation that’s tapping into the digital era. Suppose you have collated data on digital banking user experience across all financial institutions in your country. Your data was legally extracted from social media comments, apps download and review sites, official and unofficial complaints filings, etc. Can this data be successfully monetised? What are the “make or break” considerations?

Viability Heuristics

To understand whether your data monetisation idea has ‘legs’, I would like to suggest the following decision tree:

1. Does your data play a vital role in either or both the decisioning input and decisioning outcome for the organisation that’s using it? Playing in the decision input space is more valuable than playing in the decision outcome space. But being able to play a role in both spaces increases your monetisation value significantly.
2. Within the decision input and decision outcome spaces, what is the level of uncertainty that will be reduced through the use of your data? The higher the level of uncertainty reduction, the higher your monetisation value.
3. Within the decision input and decision outcome spaces, what is the effort for the organisation to achieve a similar level of uncertainty reduction via internal replication? The more effort it requires, the more your monetisation value. The best scenario is that your data cannot be replicated at all.

Let’s look at the Mosaic and credit bureau examples again. The Mosaic data allows you to define go-to market, sharpen your marketing selection criteria. The credit bureau data allows you to enhance your internal credit scores, monitor the wider performance of your customers’ credit behaviours over time, track market share. Mosaic plays in the decision input space only, while the credit bureau plays in both the decision input and decision outcome space, making it naturally close-looped. The credit bureau data beats out the Mosaic data in terms of the monetisation value in point (1) above.

Let’s now consider the uncertainty reduction that is achieved through the Mosaic vs the credit bureau data. Let’s compare on the decision input side for both. If you are an organisation looking to expand your target universe through direct marketing activities, then the Mosaic data significantly reduces your uncertainty. The credit bureau data doesn’t have that same level of uncertainty reduction because the organisation would logically already have some information signals pertaining to underwriting and credit monitoring (because these are their applicants / customers). So for point (2) above, I would say that the Mosaic data wins out against the credit bureau data.

Finally, let’s consider the effort required to internally replicate either the Mosaic data or credit bureau data solution. The Mosaic data can arguably be replicated but at a high cost. The credit bureau data is almost impossible to replicate. So the credit bureau data solution wins out in this comparison.

While the 3-point decision tree is a simple heuristic, it helps to ‘explain’ why the credit bureau data is generally perceived as more useful, and thus leading to a wider market adoption and higher data monetisation value.

Viability of Your Data Monetisation Idea

Let’s now apply this heuristic to the made up example described in the earlier paragraph where you are looking to evaluate the data monetisation value of your digital banking user experience data. Your data allows an organisation to understand if your (likely anonymised) customers are unhappy with your digital banking solution, having insights on the features and functionalities that they care about across the financial industry, and benchmarking your organisation is a leader or a laggard in this domain.

On point (1), I would argue that your data is skewed towards the decision outcome space. While it contributes to decision input (e.g. “insights on the features and functionalities that customers care about”), sentiment monitoring and industry benchmarking operate in the decision outcome space.

On point (2), your data does reduce uncertainty, although it may not be too significant. Organisations would already have some internal information about the negative user experiences of their customers. They also have a sense of where the market is heading in terms of “needs and wants” when it comes to digital banking, and have a sense of their relative leadership in the industry even though they don’t exact rankings. Overall, your data ‘tops up’ on internally existing data or triangulates certain operating assumptions.

On point (3), my assessment is that your data can be replicated and that replication isn’t going to be cost prohibitive. While they may lose out on information fidelity, organisations can somewhat replicate your data through surveys (I’ve talked about my dislike for surveys in previous articles!) and mystery shopping.

The combined assessment across points (1) through (3) should indicate that you may have limited data monetisation value in your data. You would need to strategise how your data can be more vitally important in the decision input space and in the areas where significant uncertainties exist.

Conclusion

Data monetisation and Infonomics is a growing area of interest. While many business folks get caught up with the hard economics of data valuation (i.e. how much to charge for their data), what is critically more important is whether the broad principles of monetisation exist. I recognise the simplicity of my 3-point heuristic, and hope to see more dialogue in this area.