A version of this post appeared on LinkedIn.

Edelman recently hosted the sixth edition of Shop Talk event in Toronto. Shop Talk, for context, is our way of trying to contribute to the conversation and community in marketing, communications, digital and technology in Canada. Our events aim to bring smart people from a variety of sectors and disciplines together to have conversations on the opportunities and challenges facing our industry.

At our latest event, focused on Big Data, Tamer Abu Ramadan, senior director of Analytics at Edelman, facilitated a conversation with panelists representing business, tech and academia.

This was a timely conversation as it represents a topic of discussion I’m having with colleagues, clients and partners nearly every day in our business. What is Big Data? What isn’t Big Data? Is Big Data even a useful term? How do we use data, big or small, to help us understand what to do?

I thought I’d share a few themes, comments and reflections I took away not just from our Shop Talk event but also from some of those recent discussions I referenced. Since I consider myself conversational but not fluent in data at the level of my colleagues in analytics, I've boiled my takeaways down to people, questions, answers and stories.

Start with the question you’re trying to answer

All of our panelists at Shop Talk encouraged us to frame the question we were trying to answer as a way of identifying the data and insight required to answer the questions. Put another way, it is vital to have a clear business objective or challenge in mind when culling and organizing data. You may not need a big data set to answer the question, but if you don’t appropriately frame the question or challenge you won’t know where to start. At times, the most useful way to make “Big Data” useful is to parse it into smaller datasets that focus on the question you’re trying to answer. We spend a lot of time on our teams ensuring that we’ve appropriately framed the question we are trying to answer.

Understand the role of the human in getting to the answer

A lot of talk about this at our panel and in our business, as well. A few of my key takeaways: Automation and machine learning can also help turn smaller datasets into a larger dataset and make connections that humans might not otherwise predict, but a machine will also, at this point, make decisions about what is significant that might not take into account factors outside of the data at hand. Humans, ideally with a diverse set of backgrounds and experience, must keep the data honest and No. 1 understand how and whether, in the spirit of trying to answer a question, the data should be valued and weighted to account for additional factors (recency bias being one example that was brought up by a Shop Talk audience member) and No. 2 understand any bias that humans themselves bring to either the weighting or analysis of the data.

And understand the role of a story in translating the data

My colleague Nick recently said to a client, “A good report translates data well; a great report translates the ‘so what?’ well.” This was similar to a comment Jon Bromstein, one of our Shop Talk panelists and an analytical lead at Google in Canada, made when he talked about the gap that often exists between storytellers/telling and data owners. While it is incumbent on all of us working in marketing, digital, communications and, candidly, business to maintain a solid data fluency, not all of us will be data “experts” or have jobs exclusively dedicated to data extraction and interpretation. Despite this, there is a massive and important opportunity and requirement for those working immersed in and adjacent to data to help translate the “so what” well and be the bridge between data and the story it tells.

Tristan Roy is managing director, Digital and National Specialties, Canada & Latin America.