Data is the new oil – a proclamation by a British mathematician in 2006 sets the stage. As consumers and companies alike adopt new technologies, use new systems and rely on new information, the hunger for data exponentially grows with no sign of stopping. In the perception of a businessman, data is /the/ next step to innovate their companies and improve the bottom line. If you aren’t analysing data yet, you’re behind. If you’re using data solutions, you’re not doing it enough. People would consider it a mania if it just wasn’t so profitable.
Yet, opposing views are seeping into the discourse. Ethical concerns surrounding privacy, data sovereignty and AI make the knowledgeable consumer increasingly apprehensive about this new age of data. Furthermore, the integrity and applicability of data itself is increasingly up for debate. To what degree are we acting upon a ‘paper’, or ‘data’ reality if you will? How capable are we to use data correctly, without falling into its many traps?
THE CHALLENGE OF USING DATA WELL
From market trends and product failure percentages, to employee productivity and payroll: many departments have pivoted to a data-driven approach for their work. In some fields, such as market intelligence research or data analysis, this seems logical if not inevitable. When these roles were first conceived: it was not the methodology, but simply the type of data available, that defined these fields. We have grown our data-gathering needs greatly, and rightfully so. Data, when used correctly, can be a massive power to wield for any business.
But we see this data-oriented approach enter more ‘human-oriented’ domains as well: domains where data is more complicated to quantify and measure correctly. Companies deploy ‘productivity monitoring software’ that translates the usefulness of an employee into specific, tangible metrics such as ‘minutes per client’. We declare markets as healthy when specific graphs go up and others go down. We express our satisfaction with a service or bathroom through ‘green’ or ‘red’ smileys. We simplify the process so much that there’s no room for a larger context. What do minutes or a smiley really mean, after all?
The problem with that development is simple: humans, and the systems they create such as organisations or markets, are too complex to be quantified into concrete data. This goes both ways, because data also needs to be intrinsically simple in order to be compatible with our systems. Here’s how it works through a simple example:
Human Resources uses a productivity monitoring service that keeps tabs on how long paralegals work on cases. The software uses an excel-esque system to display specific metrics for HR to look at. These metrics, usually, are based on money and speed. How much are they earning or costing, and how long are they taking?
Because of the way this works, those questions are answered by simple numbers. How fast are they? The system the company uses cannot author paragraphs of analysis that takes into account circumstances, colleagues, complexity of cases, emotional state, error-margins, et cetera. Having a system be complicated like that, would defeat the point of the system: to gather data simpler, quicker and cheaper than a conventional Human Resources department. It is furthermore preferable to make data compatible with conventional platforms such as Excel, Basenet, Clickup, et cetera: the results they churn out should be understood by the statistically illiterate, too. This makes it so that there’s always an incentive to simplify further.
This culminates into ‘productivity’ or ‘quality’ of the paralegals to be expressed in:
- How many minutes do they take per e-mail?
- How many minutes do they take per dossier?
- How many billable hours do they take per client?
- How many worked hours (paid) do they need per client?
- How long are their breaks?
Can the system identify the substance per e-mail, per dossier, per client? Not really. If paralegal A happens to work on a more difficult client than paralegal B, the data-gathering software will already start adjusting the averages. In the eyes of the software, paralegal A will from that moment be less productive than paralegal B.
Maybe paralegal A is trusted more, or unlucky enough, to gain another difficult dossier on his desk. Maybe afterwards he gets sick while paralegal B stays healthy, a loved one dies or a car repair stresses him. There the data is gathered: slow on thursday, absent on monday, slow on tuesday again?! The longer this data is being gathered, the more of these little discrepancies happen, the more you are building a ‘data reality’ versus ‘reality’.
Because when the company gets in troubled waters and needs to decide who to lay off: what will they use? Will the committee in charge take in and read hundreds of pages of evaluations, testimonies and interviews about all employees within the company? Or will they use the software that neatly ranks all employees based on their productivity and earnings? In larger or more international corporations: the oversight and ‘intimacy’ towards an individual employee is harder to maintain, and extensive knowledge about the employee and the circumstances will be limited. Therefore, in this scenario, paralegal A will be let go instead of paralegal B. Any other indicators about the person or his performance are not taken in. The data-driven approach already limited the scope of what is being looked at.
Of course the gathering and application of data varies greatly across companies, countries and continents: but small discrepancies are bound to happen in any data-driven system. And it is the presence of these small discrepancies, of any type, that lead to the problems mentioned above. Problems that can end up life-altering for people if they don’t get a promotion or even get fired over such data.
It even happens at larger scales, involving more people spanning the globe. It was the credit rating agencies that, through removing the integrity of their rating data, brought us into the 2008 financial crisis. The entire South Korean government trusted economic numbers based on promissory notes and trickled down debt obligations, suddenly finding themselves a victim of the 1997 Financial crisis almost overnight. Then there were many bubbles, such as the Dot-com bubble or even the ancient Tulip mania, where the numbers spoke against the naysayers at the time. Yet in hindsight, it is the naysayers that are vindicated. It happens to the best and brightest too. There are many events such as these, sometimes with grave consequences for people, that find their origin in the misuse or overreliance on data.
DON’T TRUST YOUR LYING EYES
The issue with our over-reliance on data goes further than the error margins or human factors skewing them: it also undermines the confidence in the human ability to sense and understand. While intuition or ‘gut feelings’ rarely translate well into tangible business proposals: it is our human nature, something that we’re using and trusting the most. We intrinsically know this when we sense something is ‘off’ about someone, or when everybody around you disputes a seemingly established consensus. It is our canary in the coalmine that allows for us to survive and navigate increasingly complicated networks of people and organisations.
When we abandon this in favour of numbers or graphs completely, we lose the ability to grasp a reality that often goes beyond raw statistics. We know for example that ‘open’ or ‘flex’ offices are horrible to the productivity and intimacy of colleagues: but because the older data is more popular than the newer, landlords and consultants keep bringing them to the market, and corporations and governments alike keep adopting them. Now they are a norm, because ‘everybody has them’. The end result of bad data became authoritative data in itself. On a smaller scale it happens too: the less productive employee can be a big hit among clients, earning his place unconventionally. The more productive employee can bring the productivity of others down, but the data will never question if he is worth the cost. If data was used correctly, or applied in a proper context more regularly: maybe these issues wouldn’t be as prevalent. But we can expect people, especially those that are not data scientists, to always correctly utilise data in every circumstance.
It becomes a situation where well-marketed and popular solutions tell you to not trust your own lying eyes. That what you feel and see inexplicably is wrong: because companies bigger than you, or people smarter than you, are making these decisions too! But when we look at history and consider all the flops, bubbles, crises and bad habits we still maintain: it doesn’t become so far-fetched that it’s easy for us to lose sight of the ball. That’s how you get small boutique firms to obsess over ‘Customer Acquisition Costs’, visualising data over numbers below a hundred or small software firms obsessing over compliance with gas plants and nuclear reactors, even though neither of these things will ever apply to them in the foreseeable future. It is in a way a standard, if not a new anxiety, that as a responsible X, Y or Z: you must be doing these things, because everybody else is doing these things.
Plaza is therefore more cautious when it applies data solutions for its clients. Yes, data is definitely a great instrument to visualise and simplify. But no, it is not the one-size-fits-all solution that we are treating it to be in the workfield. Data should be complementary to empower the human element within companies, to further identify and correct the market trends that we go through. It should not replace the human element, for it creates the risk of important people missing the point. And when important people miss the point, important decisions are made poorly.
That’s why the most important question about data is to us: how do we use data correctly?