DATA IS the new oil, but it’s valuable if it is refined. Either tech giants like Google, Facebook or local listed firms like Thai commercial banks heavily utilise data to get a better understanding their clients so they can offer them better services in a personalised way.
According to McKinsey, marketing campaigns using ads launched on social media could increase conversion rates by 30 per cent above more traditional approaches.
But what if it’s not a marketing campaign, but rather a political campaign – could we exploit data to run a successful electoral campaign, much as we do in business?
The answer is yes, using a relatively new number-crunching technique called data analytics. The approach is widely used to analyse data to guide business decisions and test scientific theories.
Both marketing campaign and political campaigns have in common the use of data analytics to analyse target clients. From a marketing perspective, the more you attract clients, the more your sales increase. Political campaigns in many countries have used data analytics to identify their potential supporters. In the US, data analytics were a powerhouse enabling both Barack Obama and Donald Trump to win US presidential elections in 2012 and 2016, respectively.
To attract voters by using data analytics, there are two key success factors.
Apparently, the first one is data.
Given the Big Data era, acquiring sufficient observations should no longer be a constraint for campaign marketers.
The big two US parties – Democrats and Republican – have been utilising voter databases. These have collected basic information, such as name, location, phone number as well as some politics-related fields such as voting history. Parties are able to benefit from these data sets if they can extract everyone’s preferences and target their campaign to the right group.
Even though these political data sets provide the actual revealed preference about the voter, this is never enough because most sets are static data such as demographics and geography. These collected variables rarely change over time, whereas they are dynamic in an individual’s real life.
Hence, for an election, it would be more useful if you could get the most recent data, which dynamically evolves over time and reflects voters’ behaviours.
That comes with the harnessing of social media data, which captures the voters’ immediate preferences. Additionally, given the myriad number of Facebook users (in the US, over 200 million or 80 per cent of the voting age population), capturing the data from the platform would enable a party to influence their social media user’s needs and point to how they could convert undecided voters to cast a ballot for them.
After a good data set, the other key factor is technique.
One of the most common techniques to be widely used by a political party is microtargeting. It is a technique of political communication based on the use of analytics to tailor messages to a subgroup via different channels (e.g., mail, phone, billboard or social media advertising) in order to build a relationship with prospective voters and supporters with the aim of winning them over as voters.
Modern microtargeting classifies voters not only based on demographics and issues, but also based on their personality characteristics, which can be predicted with relatively high accuracy by analysing their public shares online for both text and behavioural patterns.
Combining data together with techniques, will enhance the chances of influencing voters to choose the political party. For instance, targeted ads and messaging could be created to play on individual biases, fears and loves – effectively creating a bond between the target and the candidate.
One caveat for using data analytics relates to data quality. “Garbage in, garbage out” is axiomatic for data analytics as the data quality is the foundation of all analyses.
As well, some fancy algorithms should be used with caution. Consider the quote by famous statistician George Box: “All models are wrong, but some models are useful.” This is very important to remember when addressing the assumptions build into algorithms in explaining the model results. The next question is whether political parties can apply data analytics during the upcoming Thailand general election.
Thailand is scheduled to hold its first general election on March 24, 2019, eight years after the last official election in 2011. Much has changed in the economic and digital landscapes of Thailand, such as the growth of social media users to 57 million people or 82 per cent of total population. Analysing social media data would be the most effective way to spend a budget for microtargeting ads.
According to We Are Social and Hootsuite – social media solution providers – Thai people spend three hours on social time across media and across platforms. Hence, the data regarding voters’ preferences would let the party effectively focus on their target voters.
For example, while a voter scrolls through Facebook, Instagram and Twitter accounts, they may see posts or ads related to political campaigns suiting their preferences and be influenced to vote for such parties.
Ultimately, we voters should make sure we are not easily influenced by targeted ads or viral marketing from hyped political campaigns. Instead, we should listen and thoroughly think through the feasibility of a party’s policies, asking whether they make complete sense. In other words, we can do our own simplified data analytics to ensure we get the right candidate before going to vote.
Views expressed in this article are those of the author and not necessarily of TMB Bank or its executives.
ContrIbuted by PANAWAT INNURAK. He can be reached at firstname.lastname@example.org