About the Industry

Data Collection Methods


Projects can often be divided into an exploratory or pilot stage and a main stage - the former is used to better define the objectives, get some idea of what data to expect and who might best provide it - and adjust plans for the main stage accordingly. It can also provide some of the most interesting insights for the project as a whole - for example if it involves a small number of face-to-face interviews while a main stage is conducted by telephone or online, clients often speak of the first stage putting 'flesh on the bones' or adding colour to the statistics and other arguments. It is not unknown for a main stage to be cancelled after a surprising pilot stage either answers the initial question, casts doubt on the usefulness of the second element, or suggests another direction entirely.

Obviously, although common it's by no means de rigeur to have two stages - many projects have only one, some have 3 or 4, some have lots of stages. Tracker projects generally involve one 'wave' of fieldwork every 3 months, 6 months or year, for example, and exist to compare numbers and spot trends over time - they can run for many years.

Qual and Quant Research

Methods can be qualitative or quantitative - the latter looks to get sufficient responses or data records (whether completed interviews or some other form of feedback) to analyse numerically, giving results in percentages and often scaling up results from this fixed but fairly large number of answers to represent an entire market or population. The former places emphasis on weighing up the significance of individual respondents and answers. Quantitative researchers will generally try to decide in advance what answers are possible to a question, and make respondents choose one or other of them; while 'quallies' will sometimes work with a very loose 'topic guide', designed to prompt respondents to start talking, and pick up on whatever is of interest for further questioning.


There's a lot of ground in between these two - quant interviews often contain scope for people to suggest their own answers - 'open-ended questions' which are then coded up afterwards by researchers who group them according to their content or general emphasis; while qual interviews can also contain lists of prompts and precodes in among more open-ended content. I would argue there's no hard-and-fast line between the two, and that researchers who don't restrict themselves to one or the other are often the best, but some in the industry would be up in arms about this - there are passionate specialists who draw a very firm line and who believe you can't do both and be good at either.

According to ESOMAR's Global MR report covering trends in 2015, quant research accounted for 70% of all studies in 2015 (down 3% from 2014), qual for 16% (no change from 2014, but with focus groups down from 11% to 8% within that, online communities up to 5% and face-to-face (ftf) static at 2%) and 'Other' formats including desk research and secondary data collection up 3% to 14%. To an extent the rise of the last of these reflects changes in the scope / definition of research, as well as the vast new availability of secondary data.

Other Approaches

As above, a custom survey project is really just a good place to start talking about what is now an immense variety of other possible project types and approaches. The data collection part of a piece of research might involve:

  • agreeing a questionnaire and then pushing it out to a panel of pre-recruited respondents, by email
  • drawing up a detailed technical spec of requirements and then running it off against a huge pre-existing database of consumer purchasing information, audience data or mobile phone usage records collected direct from participants' phones
  • tracking a campaign via social media response, analysed via a live dashboard, and pulling out key results for further analysis and interpretation
  • using the telephone to recruit a number of groups of people (for example 4 groups of 12 people each), to come and discuss a client's products and those of competitors, in a venue where clients can watch from behind a one-way mirror (focus groups)
  • logging into an online exchange or automated research store, following instructions to choose target audiences and set up a questionnaire, paying with a card and hitting the Go button, and waiting for the results to land in your inbox a day or two later, in some cases complete with graphs and commentary
  • combining any of the above with each other or with desk research, for example scanning the Web, specific business databases, the press or even libraries, or...
  • driving to Southampton with a list of which ships are in port, finding them and walking up the gangplanks, asking to speak to the radio operator for half an hour, then struggling through their views of various satellite navigation services, with the help of scribbled diagrams, cartoons and various dictionaries as only one out of five speaks much English.

Yes that last one is a personal experience, and it's from a while back. Variety is the spice of life, and hopefully this illustrates the variety that's out there.

Note also the word 'custom'. I've talked about projects which start with a brief and a choice of supplier, but most of the world's research is in the form of continuous / ongoing waves, tinkered with between waves and generally very closely supervised, but following much the same process each time. Perhaps the two highest profile examples of this are audience research and polls of voting intention.


The company that's normally considered the largest in the global research industry, Nielsen, divides itself into 'Watch' and 'Buy' divisions, of which the former tracks audiences for TV and radio and other media using waves of data from huge panels of respondents, who both report their habits manually and allow data on their choices to be fed back automatically by devices installed in their homes (actual diaries are being phased out in the US, sometime after many would have assumed they had disappeared - they've been running alongside automated metering devices for many years). Many other companies run smaller panels (some of them still very large) in the US and elsewhere with similar intent, and increasingly these combine viewing, listening and reading / browsing behaviour across mobile devices, PCs, television and radio.

Political polling is arguably the highest profile kind of survey work, and hasn't covered the industry in glory over the past few years - it's a tiny proportion of the research sector's income, but has led to companies like Harris, Ipsos MORI and YouGov being household names - all of these companies, for example, make a lot more money from commercial research than they do from talking to voters. Other kinds of opinion research are also conducted by public sector and not-for-profit organisations on a rolling basis - for example large scale surveys of social trends and attitudes to health, education or society.

String Theories

What else do we include in the definition of MR? As above there's no hard and fast answer how long is a piece of string? Some of the other sectors seen to be 'colliding' with MR can be seen in our section on 'MR's Neighbours'.

Back to Research Objectives or Typical Approaches
On to Analysis - Business Outcomes and Implementation

Email this article to a friend

All articles in the 'About the Industry' section written by Nick Thomas of MrWeb during July - October 2017, and © MrWeb Ltd 2017. Thanks to GRIT, ESOMAR, the AMA and the MRS for stats and tables used.

About the industry - contents...

1. What is MR and how is it changing?

A comprehensive guide to what we do, how it's changing, and who else does similar stuff.

2. The global industry today

The supply side: revenues, rankings and company info, full listings.

3. Careers in MR

An introduction to the career ladder, activities, salaries and choices.