Data Can Break You

By Agnel J D’Cruz, Manager — Data Strategy and Acquisition

There are a lot of organizations that think saying “data driven” or hiring a Chief Data/Analytics Officer will solve all their problems.

Anecdotally, the only good that comes out of this kind of thinking is a short-term bump in stock price. Organizations should plan their data strategy to be able to deliver business value along with having the agility to change to a data-driven culture.

It is important to use data and analytics with care so they deliver value and do not lead to the wrong decisions. Here are some mistakes that organizations can avoid:

• Asking the wrong questions: When organizations ask the wrong questions of the data, the analyses turn out to drive the wrong strategy. An example would be where a growing company needs to focus on growing market share as against maximizing profits. Using data can answer both questions but if the company models for maximizing profits its market share as a challenger may take a hit, and vice versa. The questions asked of the data in the case of a market leader looking to consolidate market share and maximize profits will be very different from that of a challenger looking to take on a Goliath.
• Being solely tool or role focussed: some organizations or architecture teams believe that having the latest tool or technology or even hiring a bunch of data scientists could work magic in transforming their organization. While tools, technology, and people can have a positive impact, it’s equally important to ensure that there’s a cultural shift along with the right practices. Some of the practices that Socure follows in the data space is outlined in our blogThe Journey of Data by the same author. Also it is important to note that each organization or team within the same organization can be different, so agility is also a key strength.
• Using the wrong data or forcing proxies: this is another mistake organizations make in their quest to become data driven; the wrong data may be used in the analysis leading to incorrect conclusions and decisions that are detrimental to the business. Using proxies is an important aspect of data analysis where direct data is not available (e.g., testing sewage water for covid to predict a rise or fall in cases). Proxies (when used right) are an amazing tool in the hands of a data scientist but if used incorrectly or out of context can lead to mistaken inferences.
• Lack of data management: We’ve all heard of “garbage in garbage out” in the context of data. A rigorous data-management process ensures what is fed into the analysis/system is not garbage and helps us make the right decisions. A typical example of this type of an error is currency conversionAssume a company has $1,000 of sales in the US and $1,000 of sales in Canada. It’s a mistake to assume total sales equal $2,000. A strong data management system will ensure the conversions are done right while looking for outliers, completeness, missing data, format, and other considerations.
• Not focussing on ROI/Value: Executing large-scale projects without understanding benefits and costs involved is another big no-go. Data projects can range from small code runs to massive undertakings involving multiple resources and processing power. Taking on projects without clearly defined costs and benefits or calculating ROI can be something that comes back to haunt organizations. There is also another mistake that is made here which is what I call “extreme accounting” in which the need for ROI is so much that it adds red tape to the process, either discouraging practitioners or rejecting projects that have high intrinsic value and are investments for the future but are unable to prove immediate ROI.
• Complexity: Another aspect to avoid is a high complex solution with multiple points of failure. The world around us is changing fast, and with it come new data and new requirements. Having a very complex system with multiple points of failure can lead to organizations being less flexible and more resistant to change — not because of the lack of intent but more because of switching costs of technology. It’s important to have simple architectures and solutions to the extent possible so organizations can evolve with speed and agility.
• Incorrect interpretation and depiction: Assume a situation where in the summer you notice both ice cream sales and home sales both going up. You could interpret that high temperatures cause people to buy ice cream and also the same high temperatures cause people to buy and sell houses. While we know only one of these is probably true, the other, though correlated, is probably dependent on the school year which happens to have a summer break when parents consider moving homes for minimal disruption to children’s education. The same goes with using averages that have outliers that skew interpretations. A strong data muscle within organizations can help avoid similar blunders in business.
• Feedback and tracking: In a truly data-focused organization, the execution of the project is only a milestone. Once necessary business decisions are made on the basis of what the data tells us. It’s important to track the outcomes of these decisions and see if they align with the initial goal or strategy.
• Agility or the lack thereof: This is related to the previous point where organizations should be willing to be agile not just in project execution/delivery but also in strategy, e.g., if an expected outcome is not being achieved then the organization should be flexible enough to change course in-flight with the help of data. There have been instances where marketing programs have not achieved their intended outcomes and strategies have been changed on the basis of feedback data. The ability of an organization to receive, track, and analyze feedback, along with the willingness to be elastic on strategy is a huge boon to stakeholders.

The purpose of the above section is to highlight some of the mistakes that organizations can make despite having the right intentions around being data driven.

[It is very important to understand the potential pitfalls in a data project]

Conclusion

Data can make you or data can break you. Data done right can help any organization beat the market and competitors, but if done wrong, it will be detrimental. When dealing with data, what you don’t do is just as important as what you do.

(Images courtesy: https://unsplash.com/)

About Agnel

Agnel J D’Cruz is a Data Acquisition Manager at Socure and helps drive data strategy for our KYC product. Agnel comes with a wide range of experience, from working in consulting, to business analysis, to analytics solution delivery. His writing includes an HBR article, a co-authored book, in addition to blogs on LinkedIn. He also hosts his own podcast.