Big, small, old or new – how to make the most of any dataset

When I was starting out in market research (many moons ago!) everyone was talking about Big Data – about its possibilities, but also about its threat to people like me, the humble researcher. It was around this time that I made the move to work for an agency focusing on social media analysis, using a range of AI tools to automate the process of insight extraction. I was excited to be making the shift to working with ‘bigger’ data, but all it did was reaffirm my belief in the value of ‘small’ data and the role of the humble researcher.

While working at this social media agency, a likeminded colleague and I fought to run manual analysis in favour of automation – giving us control over what we looked for, and how. I also penned a thought piece that was published in the International Journal of Market Research called The (un)changing role of the researcher. In this article I explained that the insights from Big Data are only as insightful as the questions asked of it, and for this we need skilled researchers. I also argued that while big data can often tell us what is happening, we need small data (captured through expertly designed primary research studies) to tell us why.

Over the years, as data has grown even bigger and more abundant, my belief in the human skills needed to effectively analyse it has only been strengthened. While the majority of our projects are still bespoke primary research studies, we’re increasingly being called upon to help clients unlock existing data. Rather than commissioning new primary studies, clients are realising that the data they already have can work harder when analysed by researchers with the right skills.

This is exactly the type of partnership we thrive on at Old Salt. Telling stories from data is our forte, and we have a vast array of advanced analysis techniques in our toolkit to dig deeper into a dataset and reveal more nuanced insights (check out our ‘How to’ series on the blog if you’d like to read about some of them!).

While the design of a primary study is obviously crucial to get high quality, relevant data, the methods and techniques chosen for analysis are equally important. Analysis can be approached in many different ways and is as much art as it is science – data simply cannot speak for itself. We like to start the process with a set of hypotheses so that we can ask targeted questions of the data that speak directly to the business objectives. Sometimes the client has a set of hypotheses already, but we always seek to build on these in our analysis plan by considering the market context and relevant learnings from other projects, bringing our own objective view.

As with our primary research projects, the exact approach we take is bespoke to the brief, but often begins with assessing and organising the data. It may be that the data spans many years and could potentially have holes that need to be accounted for, or the data represents a sample that isn’t accurate for the analysis objectives and requires weighting. By immersing ourselves in the raw data we get a sense of its limitations, and how far we can push it. We also seek to understand any existing analysis based on the data to ensure our approach will be additive, not repetitive. We then get started on the analysis in earnest, often with a number of client checkpoints to review the direction of the findings and the analysis decisions we’ve made (the ‘art’ part!).

The way the insights are packaged is also key, and it’s vital to get this right if the findings are going to grab attention and be remembered across a business. With existing data this might be about consistency with previous outputs or creating something fresh and succinct that gives the data a new lease of life. We don’t start the process of designing a report until all of the analysis is complete and can be storyboarded. This process is so important to us at Old Salt that we could publish a post dedicated to the topic – check back in a few weeks and we will!

We love the challenge of digging deeper into existing data and the satisfaction of using it to tell a deeper story. If you have a dataset that you think can work harder, or questions that you feel you may already be able to get the answer to (with a little help!), please get in touch ?. And if you’re interested in a case study where we’ve helped a client to do just that, take a look here.