We had a chat with our co-founders Julian Simpson and Bjorn Gronli about their 25-year journey within market research.
Q.1 What was your first role in Market Research?
I joined Quantime UK (now SPSS/IBM) in 1985 as a member of their Quanline consultancy team. My first role was to support/manage the reporting and analysis end of a massive event driven research project conducted by BT plc. Over 100,000 CATI interviews were conducted each month with business and consumer customers – 4 research agencies managed the interviews. At the end of interviewing each month 4 mag tapes were biked over to the London Quantime office– this was always on a Friday evening. The tapes were loaded and then over the weekend it was my job to run the FLIP process (on a UNIX Vax 750) that converted the data into 3 Quanvert databases. On the following Monday 50 BT offices around the UK dialed (initially using 300 baud acoustic couplers) into the Quantime time-share server. The BT users would then use Quanvert (at that time a question/answer user interface – it was several years before Windows would come along!) to run/download their monthly report data.
I first came in contact with MMI Norway (now Ipsos) as a System Engineer from IBM in 1987. MMI used IBM mainframe with punch cards to register and tabulate survey data. I was contracted together with an IBM colleague, to create a “Worlds first” integrated solution for Market Research. By creating a questionnaire in SGML (later become XML), you were able to both print a paper questionnaire and automatically produce a data entry program that would follow questionnaire logic, making it unnecessary with any post data cleaning. We also created a new analysis system that would use the same SGML questionnaire as the means to do all analysis and tabulation, greatly improving operating efficiency and reducing manual programming errors to virtually zero. Our motto was:
“What you get is maybe not what you want, but it is guaranteed 100% accurate”.
Q.2 You both co-founded MarTech companies. What was the main deriver of you to start the business and what did you set out to accomplish?
After leaving Quantime in 1993, I worked for 7 years as a self-employed consultant – during this time I moved away from reporting and over to developing and managing sampling systems. Interestingly my main client during this time was again BT plc. Not only was I working with RDBMS SQL database systems – which managed the customer lists for contact management purposes – but also Windows based VBA systems (starting with Superbase before moving onto MS Access) which automated the sampling process. BT were continuing to run large Event Driven studies (if not as big as in 1985) which needed regular and timely sample feeds. Regional based monthly/quarterly targets were managed automatically by the system. This was achieved by feeding in interview outcome data and so any shortfalls in completes were compensated by an increase of sample in the quota which needed it. Sample was issued daily and sent directly by FTP to the interviewing agencies. No human intervention was required. The sampling and panel management company MARSC was born out of selling a similar solution sold into Barclays Bank UK who set out on their own Event Driven research project in 2000. We then saw the opportunity to develop a panel management solution from the initial BT and Barclays bespoke systems.
I transition from IBM to MMI (Now IPSOS) the biggest Norwegian Research Company at the time. I continued developing the internal software, going from IBM Mainframe, via IBM OS2, to Windows Client/Server platform. We added automatic scanning of paper questionnaires, CATI with dialer integration and then WEB interviewing and analysis including tables, charts , target-group analysis, factor analysis, multiple correspondence analysis. Together with 2 other cofounders in 2000 we agreed with MMI to create a tech spin off in order to offer the system developed internally (called MarkIT), to the global market. It was our strong belief that we had a state–of–the–art survey and analysis software solution and that we would be successful in marketing the solution to the global Market Research Industry. MI Pro was born.
Q.3 How has our industry evolved since you started out in market research? In particular, how has technology impacted the way research is done?
It has been fascinating to see how time–share services (used by MR Agencies because they could not afford their own servers) have switched over to Windows based networked solutions (via for a small time Unix based solutions that could run on 386 PCs) and then back again to a shared service, ie cloud solutions. In my own field of experience the big move (in Europe and North America) has been the move away from CATI to CAWI interviewing & the introduction of online panels. The industry continues to struggle with response/completion rates which advances in technology do not seem to have helped to any great extent.
As I mentioned, initially questionnaires were created on typewriter and later word processing equipment. Data was punched using punch cards and tabulated with programs that required programming skills, so the leap to today’s DIY solutions, where virtually anyone can create a survey and report the results are huge. Also, the built in “intelligence” in the systems, eliminate to a large extent, the need for both research experience/competence as well as programming skills. The time it takes from creating a survey to finished results has come down from months/weeks, to days/hours.
Q.4 Today we hear a lot about automation, agile research, AI and Machine Learning. What is your view on these concepts and technologies?
As indicated in my earlier answers – I have been involved in automation (to some extent) for the whole of my life in the MR industry. However, we are now seeing a clear move towards further automation driven by the need for cost savings in the research process. This push creates great opportunities moving forward. Automation and Machine Learning offers the potential to derive much more value from data already held.
I have been working with these technologies for over 20 years. We early on tried to use neural networks to “fuse” data from different surveys into one. The development of AI/ML solutions has skyrocketed the last few years and are deployed in solutions like coding of open-ends and video analysis. As the ease-of-use increases, I believe we will see many new areas where this technology will become very important. Things like using various data sources to predict results without the need to actually ask people for their opinion, predict future behavior, and create totally new types of analysis. Automation has made it possible for many enterprises to reduce their dependencies of Market Research companies to conduct the fieldwork their surveys. In many cases they can conduct their own surveys and are able to integrate survey data with their CRM systems and other corporate data for faster and more comprehensive insights.
Q.5 What it’s your advice to somebody starting out in the market research today?
Don’t be limited by your existing software, everything you see online or in another tool is available to you – use it. That said – ensure that the new technologies you choose are based on proven research technologies.Keep your finger on the pulse of this fast-developing industry and embrace change.
Be open minded, focus on giving value to clients based on insights and expertise. Be a consultant to our clients rather than just conducting fieldwork and basic reporting/analysis. Basically try to move up the value chain.