I was recently talking to a lawyer about some future planning. As he was explaining the complexities of the law, I realized that I had not understood one word he said. Not only did he speak quickly, but he used language specific to his profession. I knew immediately that I was totally lost. At the same time, a little voice in the back of my mind kept saying, “I wonder if this is how business stakeholders feel when someone with specific expertise, like an IT developer, asks them questions using language that’s unfamiliar to them.”
Which is where we BAs come in. Our job is to be trusted advisers, and one area where we can establish trust is to help our stakeholders understand language that might be confusing to them. In order words, we can establish trust by translating technical complexity into business language. We BAs have always done this. We take customer requirements and translate them into something the technical folks can understand…and vice versa.
But what about translating in the digital world? We still need to translate, but it’s different. It’s more complex. Here’s an example. Let’s say that I’m a business stakeholder talking to a data scientist who wants me to make business decisions that will feed into a predictive model. And let’s say that that she asks me questions like–
- What ETLs or stored procedures need to be developed, if any?
- What is the frequency of updates required for the data to ensure currency?
- What is size of each data set and how much data will I need to get from each one?
- Will the data need to be standardized due to disparity?[i]
- Which scales apply to your different datasets, nominal, interval, ordinal, or ratio?
- Which statistical analysis techniques do you want to apply? Regression? Cohort? Predictive and prescriptive?[ii]
And let’s say that I have no idea what she’s talking about. As she’s asking me for answers, I’m thinking, “What in the world is an ETL? What’s currency? How would I know the size of the data set? What is disparity? What do these scales mean? Cohort statistical analysis—are you kidding me?!”
As a business stakeholder, what should I do? Here are some possibilities:
- I could pretend that I understood her. The advantage of this alternative is that I wouldn’t have to admit that I hadn’t understood anything she said. I would reduce the risk of her thinking I was stupid or being frustrated with me because I didn’t understand. Perhaps some long-term pain, but a great deal of short-term gain.
- I could wait until she was done speaking and then ask questions. The problem with this option is that by the time she is done asking the question, I’ll probably have forgotten what I didn’t understand. So I wouldn’t be able to formulate a coherent question.
- I could interrupt, signal for a time out, and to ask her to speak in business language. But interrupting is never comfortable. And it may irritate others in group settings if we do it too much.
Of course, I’d rather not be in that situation. I’d rather have someone in the room or on the phone who understood what was being said and could tell me what it all meant. Someone who could simplify it for me.
Someone like a BA.
As a business stakeholder, I don’t care about the details. I want to know why something is needed, and it sure would be nice to have someone I trust in the meeting or on the call with me who could help me understand what is being said and why it’s important. Someone who could tell me, for example, not only that an ETL is a process used to get data from one or more databases to another, but also that it’s needed to maintain the integrity of the data that will be used to train the models. And importantly, the ramifications of having inaccurate data. That means, of course, that we BAs need to do our homework.
We need to do prep work prior to any meeting between the data scientist and the business stakeholder.
We need to find out which questions will be asked. If we don’t understand any of the terms, we need to research them.
We need to find out which business decisions are needed.
And we need to make sure we think about the impacts of those decisions to be able to advise our stakeholders about decisions they’re being asked to make and the impacts of all possible responses.
We need to be sure that the stakeholder is protected, which means we need to figure out a way to stop the questions until the stakeholder has been able to digest them. Ideally, we would have a conversation with the data scientist about our role and this process in advance of any meeting.
If this seems like a lot of work, it is! But prep work is part of what we do to establish trust. It gives us credibility. It gives one more reason for our stakeholders value to us and one more way for us to provide value to them. Which, after all, is what we do.
[i] The first four questions are from: Kate Strachnyi, May 14, 2018, 20 Questions to Ask Prior to Starting Data Analysis, Towards Data Science, https://towardsdatascience.com/20-questions-to-ask-prior-to-starting-data-analysis-6ec11d6a504b
[ii] The last two questions are from Sandra Durcevic, Jan 8, 2019, Your Data Won’t Speak Unless You Ask It the Right Data Analysis Questions, https://www.datapine.com/blog/data-analysis-questions/