Lessons in data from the #GE2015 campaign: Four key insights

Tableau's Andy Cotgreave is a data fanatic. During the 2015 UK General Election campaign, Andy was busy mining the rich seams of social media data. Here, he summarises the four key lessons in data that businesses can take from the campaign.

After an official 30-day campaign (and an unofficial one lasting well over six months), the UK has chosen their next government, with the Conservative Party winning with a majority almost nobody predicted - least of all the election polls.

The 2015 election has been dubbed the UK’s first ever ‘data-driven’ election. With data analytics currently powering so much of what we understand about both business and the world at large, here are four key conclusions businesses can take from the campaign season into their own organisations:

1. Don’t build dead end dashboards

In the run to May’s election, the press and media dedicated significant time and resource developing elaborate visualisations of the opinion polls in an effort to predict how the results would pan out.

Polling groups from YouGov to Ipso Mori all forecast that the Tories would be neck and neck with Labour, in what they predicted would result in a hung parliament. However, the story of the exit polls had a different ending, with the Conservatives winning a surprise 51 per cent of seats.

One could argue the problem with the opinion polls was not the failure of either data or data analytics, but in the fact that the data didn’t change in real-time. The polls remained static throughout the campaign. The dashboards became dead ends and as a result, the media polls were rendered boring – why go back to a chart when you know it’ll look the same as it did last week?

Companies in general focus much of their efforts on building one-off dashboards that show a representation of their business. Businesses expect employees to return to these dashboards often, in order to track their progress. However, the dashboards frequently become dead ends, rarely visited by anyone. Why? Because the data doesn’t change.

The real value in data analytics lies in being able to seek answers to fresh questions. If the data isn’t changing, stop tracking that metric and look for one that is. Dashboards should be treated as evolving objects that need to adapt as you seek the most vital answers.

Continuous new insights was the inspiration behind Tableau’s election project, Impartiality UK. By monitoring social media data, Tableau produced new data visualisations each day throughout the campaign to analyse the social election from a multitude of different angles.

2. Sentiment tracking is not always accurate

Social media has become a valuable source of data, with many pointing to President Obama’s re-election campaign in the US as the classic example of how social media and data science can play a fundamental role in a political campaign.

In an effort to gauge public opinion throughout the UK General Election campaign, many organisations turned to social media data. Tracking this data can not only be an excellent indicator of consumer trends, but audiences love insight from social media as it apparently reflects the true emotional responses of the general public.

However, as the exit poll demonstrated, insight gained from social media data was arguably misleading in terms of predicting public opinion, as the majority of social media conversation fuelled the perception that the race to Downing Street was set to be neck-and-neck.

As the election campaign demonstrated, it is important to consider that sentiment tracking can be inaccurate. Take for instance, The Daily Telegraph’s debate sentiment tracker. A disclaimer was added during the UK campaign acknowledging that sentiment is only accurate between 60 and 70 per cent of the time.

That said, the lesson for businesses is not to avoid sentiment tracking in their data analysis, but to use it with caution.

Telegraph

3. If data doesn’t support your argument, don’t use it.

As always, the party literature has produced some devious visualisations. In this year’s election one of the major problems lay with organisations truncating axes and even making up data. This was at best patronising, and at worst, deception.

Businesses should trust their audience to make the correct conclusions and show data truthfully. The biggest mistake a business can make is to lie with data. Put simply, if the data doesn’t support your argument, don’t use it.

4. Visualisation has become mainstream

In 2014, the FIFA World Cup, perhaps for the first time ever, normalised data visualisation and brought it into the mainstream. This is true of the General Election too, where data visualisation became part and parcel of the online conversation.

For instance, a look at the photo stream for the official election hashtag (#ge2015) reveals that highly visual charts and data make up a significant proportion of the tweets.

People are increasingly data literate and crave accurate information which represents the world they live and work in. Businesses should harness this curiosity and implement a culture whereby data forms a central part of everyday business practice.

GE2015

Conclusion

The fact that data was a lynchpin in the UK General Election campaign reflects the growing role data and analytics occupies in the wider landscape. Data is increasingly relevant in all aspects of day-to-day life, from politics, to sports and fanalytics, to the wider realm of business.

For businesses navigating the booming data landscape it can be challenging to know where to begin managing the multitudes of data generated each second. By keeping these key lessons in data at the forefront of their agenda, firms should be able to harness the insight gathered from the General Election campaign to improve their bottom line.

When people can see and understand data, it truly fulfils its purpose.