4 Things Business Leaders Need to Know to Stop Pissing off their Data Scientists
A Series to pay Homage to the amazing Data Professionals of my Class
Last week I graduated from my Masters in Analytics, which took two years of learning together with a group of amazing data professionals. Coming from a sales and business background, it was obvious this was going to be an uphill journey, so instead of doing a blog with fancy data tricks, I will do some blog posts as a non-techie, who somehow made it through, by the kindness of classmates and faculty!
Let’s kick it off with this post, hoping that it will stir greater love between business and data teams ;)
1) Data Scientists are not magicians
Get me some interesting data, so that I can make money with it.
Data scientists can build and test hypothesis and prediction models, but they typically cannot make magic, i.e. find an untapped profit centre in 2 days. It takes a clear Business intent working hand in hand with Data Science skills to bring out the latent potential of your data assets.
It hit me first hand when I tried to do my Final Year independent project. I wanted to do something different from Finance, and that got me stuck at the intent stage for a really really long time. There was tons of ecommerce and travel data sets in the internet, but the question to grapple with was - “Will this endeavour contribute to the business?” Eventually, my project was on online travel booking patterns of Airbnb users in North America. The intended benefit is to sharpen the use of time based incentives, for geographical segments.
Data can be valuable but there is also a ton of it. In that sense, it is like mining for Diamonds. However, if you spend too much time on irrelevant data then it becomes just, well, coal. This leads to the next point.
2) Data Scientists work with the Data they have
Data is like collecting firewood, if you have not been consistently collecting the right pieces of wood, don’t assume that everything can be ablaze in warmth whenever you want it. Damp wood cannot make you a bon fire, neither can data that has not been collected meaningfully with a purpose.
For example, if I wanted to know how many people bought Tesla stock after reading my Tesla Research Report, but yet, I do not have the ability to track reads, opens and downloads, then I would never know. Worse still, I have seen cases where management gets pissed with the Business Analysts (No Data Science team), because they can’t “figure it out”.
Its really pretty simple - If the company did not collect the right data in the past, then its going to be a cold Winter, till you start collecting in the next Spring.
Suck it up and give your Data Science team reasonable time to collect it.
3) Unsexy work takes the most time. The 3 Letter - E-T-L
The terms Data Analysis and Data Science, almost always conjures fancy images of charts and graphs to the management folks. Powerpoint Charts, Power BI, Tableau. and the infamous: “Why does it take so long?” Bosses go into reminiscent mode: "When I used to be a management associate, I could spit out powerpoint charts overnight”.
In reality, it does not work like that anymore. The massive scale of data points available today cannot be simply synthesised through a mental rumination exercise. Data needs to be cleaned, and transformed into a meaningful fashion before the final pitch consideration can be contemplated.
However, Technology has infinitely complicated this endeavour. Yes, the statement sounds strange, but it is true. Many of us who have worked in Big Corporate firms know that are so many of our internal systems that don’t “talk” to each other. Thus, a big portion of this effort, goes to getting the data points to be comparable “apple to apple”. Let’s use a simple example, so that it does not cut too close.
I used to have 2 phones- a Samsung Galaxy Note and an Iphone x (as in variable x), and I could not get the 2 phones to maintain the same contact list dynamically, albeit, this was in the mid 2015s. The solution then, which I still use today, is to install the Gmail app on both phones and use the Google contact list as the main database. Thus, it would Extract (and Clean), Transform into Google terms, and Load into the Google Contacts database. Once the data went in, it can be pulled from the Google Contacts, and delivered to any phone at any time after that.
Now, imagine we are doing this on website activity of clients of a Bank or an Insurer or a Broker; there are different systems from Google Analytics to Avaloq to Salesforce. It will take time to Extract the raw data from the various sources, clean it up to align the fields, Transforming by joining, filtering, and Loading into an accessible database. The E-T-L process takes the longest time and when that is done, the visualisation layer can be built relatively fast in comparison.
Now imagine all the different systems in your company, but of course again, you might think that they all talk to each other ;).
4) Data is Not Free for us to profit from
Lastly, Data is not free. Data comes at a cost. Take a look at the biggest names of the world - Google, Facebook, Zynga, or in South East Asia Grab, SEA (Garena); they spend billions creating platforms that gives them user data and engagement. With these data points, they build ways to monetise, through advertisements or product solutions in the engagement.
Thus, data that is relevant and necessary will cost because of data acquisition and data transformation to be useful. The next time that your Data Team needs to spend money on data. Don’t snap at them!
The better question is “What are my Data Focused Free Products/ Loss Leaders?” If a company does not have that, then it is harvesting today’s value, but not planting tomorrow’s seeds- the infamous “Low Hanging Fruit” addiction. The problem with this addiction is that there will be no more fruits within arm’s reach after a while, and it will be too late to plant new seeds!
Like what you read? Subscribe Here. It’s Free! You can leave anytime too.
Disclaimer: I write as a hobby on topics that I find useful to have a voice on. Nothing here represents the opinions of current or past employers, nor product recommendations or financial advisory in any form. I hope you find the writing useful.