Framing questions: Marketing data integration Part 1

How do you get the best data output from your surveys?
Get the report
Question mark
Alex Wheatley

Director of Digital Innovation, Europe

If you've seen the introduction to our Marketing Data Integration Series, you understand the extent new technology and mobile plays in modern research. So now you're on a mission to modernize your research (bravo!). You understand how to reach the representative audience you need (include mobile!), and you’ve identified the right tools available to get the answers you need (responsive programming!).

So what now? You need to know how to get the best data output from your survey questions; you need a Modern Survey. But what does that entail?

Be pointed

We know that both the attention of respondents and screen space are finite resources, therefore the first exercise is to ensure that only the essential questions are being asked. It’s best to be pointed; meaning if questions overlap or are unclear, don’t ask them.

Here’s a tip: If the survey was run previously, you can use past data to help steer your choices. Time permitting, you may also consider a pilot study (or a soft launch) to help identify questions that can be trimmed.

Redundancy indicators include:

  • Dropouts
  • Satisfaction Scores
  • Correlating Answers
  • Skewed Answer Balances
  • Straight-lining
  • Click Counts
  • Text Analysis
  • Answer Timings

Create a narrative

With a core question set identified, the next step is turning them into something people will want to answer. Too many surveys are a random collection of questions, covering various topics, with no rhyme or reason. This is not a Modern Survey.

A Modern Survey has a clear beginning, middle and end. It values respondents by giving them a reason to take part, and you’re rewarded with stronger insights as a result. Narrative is what sits at the heart of this approach. As a researcher, look at your questions and think “how do I pitch this to a survey taker?”

This is often an easy task. Simply think a bit less like a researcher and a bit more like a storyteller. Surveys are built on the stories we want to tell, but we don’t leverage this as we should.

Instead of asking a random collection about people’s shampoo usage, for example, why not build the survey around the central question we all want to know: “What is the secret of a great shampoo?” With some introductory text and a bit of imagination, you can ask the same questions but with more intrigue. Questions on packaging, smell, features, product usage, branding etc. all can be asked, but with a central narrative and signposted structure, your survey is no longer a random collection of questions. It’s an enjoyable problem for the respondent to solve.

Here’s a tip: If the story you want to tell is a long one with multiple topics to cover, consider making it a magazine rather than a novel. Instead of one long survey, try serializing your research. Get people to opt into your research piece; give them a reason to return to each wave and they will. We have found that making people feel valued in content driven longitudinal research is a fantastic way to draw them through larger volumes of content.

Strengthen your question set

Once you have your story, make your questions as valuable as they can be. This means identifying the questions people want to answer, not using the easiest questions to get to the data. The biggest detriment to insight is questions written with the researcher, not the respondent, in mind.

What does this mean? No more repetitive banks of Likert scales! If you want valuable views, ask questions that allow respondents to give them; don’t formulate questions for the purpose of an easy PowerPoint slide. We see repeatedly that standardised questions give standardised answers. Instead, think about (and test!) your questions and scale choices to ensure they resonate with respondents.

Here’s a tip: Think about what you would find interesting to answer. No one likes long, boring memory exercise. Arduous recall tasks are where you should be looking for data, not survey responses. Consider what panellists signed up for: opinion sharing. Develop your questions around that. Don’t ask people what adverts they remember; ask who makes the best and the worst and get recall as a bonus.

Here’s another tip: Think of how you can challenge and reward respondents. Don’t ask them to list their favourite brands, ask them what stores their perfect shopping mall would have. Don’t ask them to rate brand values, tell them they are a CEO and ask what they would prioritise for their brand.

Finally, gut check

Utilise innovative and enjoyable methods, like implicit testing, or feedback, to draw respondents in and gain deeper insight. A Modern Survey may be hard to define, but it is easy to see.

Take you survey and ask yourself: “Was that enjoyable and easy to answer?” Or, “was it arduous, unclear and a chore?” If the answer is the former then you are ready to field, well done. If it is the latter, then consider reaching out to your modern survey experts for support.

Are your projects sufficiently "modernized"?



Over 1500 scientific and peer-reviewed articles to further understanding in health research.