Can a data scientist on your team balance the company books? What about put together a marketing campaign? These jobs may seem far removed from a data scientist's job description; however, don't overlook the advantages of stretching your data scientists far beyond their typical realm of algorithms and code.
I often espouse dropping a business user on your data science team to keep them engaged, but does it work when you flip the script and put a data scientist on a business team? You bet! To add more value to your business collaborations, give your data scientists a business job for a while. Actual business experience greatly reduces the nasty practice of gold plating.
Gold plating and the halo effect
Gold plating may sound like something desirable, but in the software development world, it carries a negative connotation. Gold plating is when developers (including data scientists) add extra features to a solution because they feel they know what the business users really need.
For instance, your team may be chartered with developing a set of specific analytic reports. Once the initial requirements are completed, your data scientists assume that an ad-hoc slice-and-dice environment would be a nice addition, so they build and deliver it as part of the solution — in an unsolicited and unannounced manner. The problem is that no one ever asked for an ad-hoc query tool, so it may not be received as well as anticipated. This happens all the time.
You're at a special risk of gold plating with the data scientist variety of developer. Data scientists are very, very smart; they have excelled in their academic and professional careers over a multi-disciplinary curriculum including computer programming, advanced mathematics, artificial intelligence, and data visualization. This level of excellence creates a halo effect, or a false confidence that they know more about every subject (including business needs) than they really do. Initial conversations with the business may be met with an open mind, and then, in very short order, an unhealthy arrogance sets in. This is the soil that germinates gold plating. The halo effect is essentially eradicated when the data scientist performs a business user's job.
They say experience is the best teacher, and nothing is truer when it comes to understanding business requirements. On several occasions, I've purposely inserted myself into a business situation to better understand what it looks like from their side.
Over my long and illustrious career as a consultant, I've been a finance professional, a product manager, a marketing analyst, a loyalty specialist, and even a facilities engineer on an oil field. In every single case, I had false assumptions about the job that were dispelled once I started doing the job. Your data scientists will have the same experience when they actually plug into a business team.
I recommend more of an internship than an auditing function. If your data scientist is just watching the pros do their job, they won't get a real sense for what it's like to be in the role, so try to get them a real job in a business function. The business — for good reason — will have reservations about this plan, so it must be approached properly.
Although your data scientists should be given a real job, it shouldn't be a critical job. Work with your business to assign them a role that adds incremental value, not something that makes or breaks the business; that way, your data scientist can learn the job at his or her own pace, without violently disrupting the business function. After a month or two, they should have a pretty good sense for what the business needs. They still shouldn't gold plate, but they'll have better insights and relationships to make a real difference instead of just assuming they can.
It's great to have data scientists contribute business value to a solution, but make sure it's the right contribution. Although gold plating is universally discouraged, it still happens on a regular basis because data scientists assume they know what business users need. The only way to understand it from the business perspective is to live their life for a month or two.
Give your data scientists a real business job for a while so they can appreciate what it's like to be a business user. Then, when they have a great idea, it's not gold plating — it's just another business user with a great idea.
- Mini-glossary: Big data terms you should know
- Don't scare off data scientists on day one
- Two programmers are better than one for your data science team
- Job description: Chief data officer (Tech Pro Research, a sister site of TechRepublic)
John Weathington is President and CEO of Excellent Management Systems, Inc., a management consultancy that helps executives turn chaotic information into profitable wisdom.