How an Agency Used Topic Data to Drive Unique Insight for an Automaker

Kester Ford
10th December 2015 0 Comments

Every time we buy something we have to make a choice of which product or brand to choose. Marketing and advertising attempt to sway our choices and the strategies used are, to some extent, determined by how we make our decisions. Sometimes the process is instantaneous at the point of sale – we buy the brand we always buy, or the cheapest in front of us – in which case brand building and pricing strategy may be the most important tools in the marketing kit. Sometimes we take a lot longer considering our options. When we do take longer there is more research involved and often we are proactive in asking for advice and recommendations. This provides more signal for marketers to work with to meet our specific needs at the time of purchase.

Buying a car is probably the biggest purchase most of us will make in any given year and once we have made the purchase we have to live with our decision for years. It’s an important decision for us – a mechanical mini-marriage – and one we give a lot of consideration. It’s also important for the automakers as it is their last chance to capture our hearts, and our custom, for the next few years.

But how do they capitalise on our consideration? How do they know which features we are interested in and how they stack up against the competition? How do they even know who they are really competing against for our business?

One agency was struggling to answer these questions for its client, the US arm of a global automotive brand. They also wanted to know whether these answers were different for English speaking and Spanish speaking audiences.

The agency turned to Facebook topic data for the answers. They queried data from Facebook posts, likes, comments and shares about a range of automotive companies (including their client), the models produced by these companies and various car features. They categorized their data by company, model and feature in each language so they could easily draw comparisons for each of their client’s models against its competitor set. They also built classifiers to determine the intent of posts: were they “considering” buying a particular make of car, did they reflect that the person posting already “owned” that type of car or were they “rejecting” the idea of ever owning a car from that brand.

During the analysis of this data, the agency determined how often the client brand was mentioned together with each competitor brand as an indication of “cross-shopping”. This gave them insight into which brands they were actually being considered against. They were also able to get insight into which features were being compared and what was motivating the comparison.

By analyzing data from Spanish language and English language posts separately, the agency was able to determine which features were most important to each group. Whilst style, practicality and performance (estilo, practicidad and rendimiento) were most important for English speakers, Spanish speakers were more focused on especificación mecánica, ambiente and coste de funcionamiento (mechanical specifications, the environment and running cost).

When looking at the intent of the posts about each manufacturer, the agency found that their client’s brand had one of the lowest levels of “customer consideration” amongst the brands they were looking at (it featured less often in posts about deciding which car to buy). They also discovered that they had relatively high levels of “rejection” (people saying that they would not purchase the brand). This implies that a lot of times the battle was being lost before the consideration phase had even begun in earnest.

These findings meant that the agency could recommend two changes of approach to their client. Firstly, they could change the content of their campaigns to best position themselves against their competitors on the key features that are actually being compared. Secondly, to produce entirely separate content for Spanish and English speaking audiences rather than simply translating from one language to the other. These two changes are intended to help the automaker increase the times it is considered, and selected, to be our automotive spouse.

Kester Ford

Written by Kester Ford

Kester Ford is DataSift’s Director of Product Marketing

Share This