I Took the Liberty of Ordering You Spaghetti; A big mistake too many analysts make

Feb 27, 2023

Imagine arranging to pick up food from a popular restaurant to share with your work team. You’ve read wonderful reviews about the accomplished chef, and you’ve been looking forward to treating the team to a great meal. You arrive to place your order but cannot find a menu. The server happily informs you that your meal is all packaged and almost ready. You explain that you had some preferences and special requests you were hoping to make. She smiles condescendingly.

The chef appears and dismisses your concerns. He explains that he knows the ingredients much better than you do, and has exceptional skills in food preparation, developed over a decade of training. He insists that because you initially indicated you wanted Italian food, he took the liberty of creating a full array of dishes, presented in Tuscan styles, that best represent traditional cuisine.

When you try to describe your goals for this shared meal among your team members, and their particular inclinations, he resists. He has created these sorts of meals for hundreds of people, who are just like you. Incredulous, he asks: How could you know better what to prepare when you have no culinary expertise?

Well, here’s the problem.

While he may indeed be an expert in food preparation, he doesn’t know about your favorite flavors, one coworker’s allergies, or another coworker’s adherence to a vegan diet. Further, he didn’t know if your team has a tradition of sharing many, different small plates, or perhaps had a monthly occasion to celebrate with a variety of desserts. He assumed he knew best. He didn’t ask.

Most people would agree that--- unless the client specifically signed up for a total “surprise-me,” chef-selected menu--- it is unlikely that this meal will meet expectations better than a meal created in collaboration with input from that client. It makes logical sense.

So, what does this have to do with analytics?

Recently, I listened to an analytic team brag about how many predefined reports their system could produce. They were thrilled about their ability to spit out a plethora of insightful findings! Their implicit message was that they would choose what their clients needed to know.

Put another way, they suggested that clients were often not smart enough to know what they need. So, in their opinion, it was the analytic team’s job to define what is necessary. Not once did anyone mention a need to gather input from clients receiving their report, although analysts did express some frustration that clients were less appreciative than analysts expected. Why aren’t they more excited?

I’m not kidding.

Is it any wonder that 85% of analytic projects provide no business value? Or that 67% of business professionals are dissatisfied or completely frustrated by what they get from their analytic teams? Or that only 50% of business decisions are made with data? We can do better.

Although not a perfect analogy, the meal example above illustrates the flaws inherent in analyst-driven content. No matter how extensively trained and talented the data science team, they do not know what the business team needs better than that team knows themselves.

I repeat: The person requesting information ALWAYS knows best what they really need. Maybe they need help explaining it, defining it, or understanding options of how to analyze or show it. But they know best what they need, why they need it, how it will be used, and by whom.

Like a chef in love with his own signature ingredients, techniques, and presentation, data teams often become enamored with specialized analytic methods and wedded to traditionally accepted ways of expressing results.

Here, have some ravioli. You’ll like it.

It’s not uncommon for analysts to begin planning how they will do an analysis and what datasets they will need before the requestor has completed her first sentence. Then, they run off to produce what THEY believe is important to know. To that point, these are three—completely avoidable—attitudes I notice from analytic teams that interfere with delivering successful results:

  1. An assumption they know what the client needs (better than the client does).
  2. A greater interest in the methods used to answer a question than in getting the question right.
  3. An arrogance that clients are not smart enough to understand what they need.

What we can do instead.

In my work training analytic translators, we devote extra time and effort defining a project. Specifically, translators learn to explore a client’s request through extensive discovery. This step in the translation process is the most critical series of conversations a translator has—after all, getting this part wrong means the rest will be wrong, too.

We approach these interactions with appreciation that analytic teams serve the business, and everyone brings unique, important knowledge to the collaboration. We also assume that an analytic request almost never comes to us fully formed. Through conversations, the request will almost certainly evolve—sometimes dramatically. That evolution is not a “problem,” it indicates useful clarification.

For example, we have discussions that clarify background such as:

  • How the idea for this project came about,
  • What they hope to learn,
  • What they hope to decide from this new knowledge,
  • Criteria they have for those decisions, and
  • Who (else) will use the information.

Through this dialog, the person making the request inevitably becomes clearer about his or her own objectives and, as such, the project definition morphs too. As the business and analytic team members reach a better understanding about necessary project components, everyone’s confidence increases. Plus, the eventual results are more likely to meet business objectives the first time.

In the case of analytic teams who believe they know best (without input), it will require a shift in their thinking and approach. Too many companies suffer from a serious disconnect between business and analytics, stuck in a dynamic that doesn’t serve either team. By learning translation skills, teams learn what matters to each other and invite input.

That way, when the meal arrives, those who hoped for linguini alfredo don’t end up with fried calamari.



Wendy D. Lynch's headshot photo.

Wendy D. Lynch, PhD

Wendy Lynch is an experienced sense-maker and data scientist with over 35 years of research experience, primarily in business settings. She has played the role of Analytic Translator for hundreds of companies, from start-ups to Fortune 100 corporations. Her expertise is both in data analytics and effective communication, combining the two into a framework for optimizing the value of analytics in a business setting.  Connect with her through LinkedIn or email.

Newsletter: A Matter of Translation

Making sense where business and analytics meet.
Training professionals to speak both languages.

We won't send spam. Unsubscribe at any time.