American Marketer


Why conventional polling does not work. Hint: mobile

May 27, 2015

Parry Bedi is cofounder/CEO of GlimpzIt Parry Bedi is cofounder/CEO of GlimpzIt


By Parry Bedi

In the wake of the last United States presidential elections, the referendum on Scottish independence, the Israeli elections and now the United Kingdom general elections this month, the efficacy of political polling – and by extension, the whole market research industry – is once again being questioned. And in my opinion, very rightly so.

If we cannot trust these polls that are supposed to give us an answer to a simple question: who will people vote for? How can we trust them to provide us with answers to really critical questions such as should we ban fracking or how do you feel about the new healthcare system?

While we eventually know who won the elections when the votes are counted, and perhaps the only people “suffering” are political strategists and pundits, the impact of being lulled into this false sense of security that we know what people think, especially as an input to public policy, is frightening.

If this election is any indication, it is that polling, as conducted today, does not work. But why?

Mobile-first world
The industry continues to lag behind in adopting the technology advances sweeping our society.

Many firms still solely rely on landlines to gather responses, partly due to regulations around robo calling mobile phones. Unfortunately, just supplementing the existing landline data with live calling to mobile phones is not a lasting, or even appropriate, solution.

The way that consumers, and not just millennials, communicate is rapidly changing. The biggest indicator: smartphone penetration, which is now well above 70 percent within the U.S. Additionally, 50 billion text messages are sent daily.

The value of mobile is in the freedom to communicate in the medium best suited to an individual and the ability to do so anywhere, anytime.

Solely relying on voice forcibly selects out a huge swath of the population whose opinion could have a substantial impact on important outcomes.

Lack of validation
The second main challenge facing the industry is the socially desired responses phenomenon.

Here, respondents give answers that they think the pollster wants to hear.

Add to this the fact that there is no definite way to validate their response – is it truly the person’s opinion? – and it is highly likely the conclusions drawn from such responses will be wrong.

Self-selection bias
Surveys tend be tedious and often boring. This effectively narrows down the number of people who are willing to participate in research to a smaller self-selected segment of the population to the ones who either really care about the issue under discussion or have a certain amount of time on their hands.

This obviously means that there is a high chance that the silent majority is not represented in the results.

Picture emerges
How can we solve this?

One approach is for pollsters to embrace the communication media that their target demographics are using.

Unless you are looking for respondents over 50 in the U.S., that medium is likely not voice, let alone a landline. It is mobile.

If you are running a poll, and it is not on mobile and you are trying to reach a broad demographic, than your results will be way off. Bottom line.

But, there are many other benefits that mobile can bring besides just reaching a specific demographic.

Our world is becoming more mobile and visual in tandem, as evidenced by the rise of visual platforms such Instagram, Pinterest, Snapchat and now Periscope and Meerkat.

In fact, visuals are emerging as the predominant posting type on Facebook and Twitter as well. And why is this the case? It is simple. Visuals are often a more concise and easy way to communicate, especially on mobile.

But visuals are not just a preferred language. They also hold an abundance of data, can scale at speed and offer validation.

For example, when a respondent replies to a poll with a picture of her election yard sign, the visual shows us she prefers a particular candidate – not just because she says so – but because the picture she sent proves it.

For further validation, you can even geotag the response to validate that it is, in fact, a residential location.

The power of visuals really shines when you consider complex issue-based polling. It provides researchers with an opportunity to quickly gather meaningful, in-the-moment-of-truth insights, while ensuring a stellar respondent experience – one that is non-intrusive, almost effortless, fun and very natural to people.

These factors profoundly increase both the validity of responses and reduce self-selection bias.

As an example, we recently asked 300 people: “The political issue I'm most interested in seeing addressed in the 2016 presidential race is _____? Take a picture to fill in the blank and explain why in 140 characters or less.”

After an exhaustive analysis, one clear theme emerged. The most important issue in people’s minds was healthcare reform.

NOW IS THE time for pollsters to reinvent themselves – to wake up and rethink their approach if they want to stay relevant.

There are so many new ways for collecting and analyzing data, and an equal number of changes in consumer behavior. What worked in the past does not work today.

From a high-level, if you work for any organization that conducts research, if you are not reaching consumers on mobile, you are missing huge demographics. And if you are not tapping the new data sets available to you, than someone else is.

Parry Bedi is cofounder/CEO of GlimpzIt, a San Francisco-based visual dialogue platform. Reach him at