The time for measuring brand mentions with social listening technologies has come and gone.
To demonstrate business value and get more business buy-in you need to innovate and do something that directly relates to business growth. When we caught up with Preriit Souda, Data Science and Strategic Insights Director at PSA Consultants, that’s exactly what we found.
In this article, learn how Preriit used a combination of data sources to turn around the dropping sales at a restaurant chain. You’ll also get his insider tips on how to analyse the context of social data, how to deal with meta-data and integrating other third-party data to provide a deeper picture of customer behaviour.
Help, We Need to Sell More
So, what do you do when your sales are declining and you’re losing your relevance in the marketplace?
If you ask Preriit Souda, he would say that you turn to the data that the customers you want to target are posting on social media.
And, yes, I agree with him too.
If you’re looking to add a few more zeros to your sales figures, his approach of combining data sources and analysing the context of social data could be the intervention you’re looking for.
Too many people rush into creating a query and pulling data through a social listening tool before really thinking through the question they are trying to answer and the problem they are looking to solve.
This is a big mistake.
To increase the utility of your social intelligence initiatives you need to think about how people have conversations, the context of the conversations, the limitations of the data and ways that you can plug the context holes.
Preriit is one person who know all about this.
He’s created a framework that links derived insights with social and digital data (with all the meta data) to the sector or problem specific data sources.
The framework helps you to identify the right data sources and the types of analysis to run.
Let’s go through an example to show you how it works…
Analysing the Data
Here’s a picture of Preriit eating ice-cream. You could say it is a pretty standard social media post.
So, how would you begin to analyse this post?
For Preriit, he looks at the What, Who, When and Where.
For Preriit’s example, the what looks at three different types of analysis:
Image mining, in this case the customer was eating a pink coloured ice-cream.
Meta data mining, for the time of day – lunchtime.
Text mining, the customer likes the ice-cream colour but is sarcastic about how cold it is.
The who also took two types of analysis:
Meta data mining, for the sex and ethnicity of the customer.
Image mining, to identify the age range of the customer.
To analyse, the “when” Preriit linked the meta data of the what to weather data. This allowed him to understand the temperature outside when the customer was ordering his ice-cream.
While that might not sound that important just now, when you think about how the weather may impact food choices in the restaurant it becomes critical.
The “where” concentrates on meta data linked to geospatial restaurant and locality data. This brings insights into the types of restaurant he is in (Italian) and what the locality of that restaurant looks like. In this case a multi-ethnic area surrounded by Indian and British restaurants.
The where helps to identify what the competition looks like as well as insights into the type of people – the consumers – in the locality.
What the Framework Gives You
In Preriit’s example the framework provided him with insights on:
- Context: the person, their feelings and actions
- Location: where, the competition and the culture
- Weather: the fluctuations in the weather
- Demographics: sex, age range and ethnicity
- Time: the date, day and time of day the actions were being taken
Now imagine this at scale.
But how does this fit into increasing sales at a restaurant?
Saving a Failing Restaurant with Data Insights
For the restaurant sales project, Preriit explored six key data sources from social data, store sales data and the ethnic composition of localities.
What’s interesting about this is that by exploring meta data and alternative data sources, Preriit and his team were able to fill in the context holes with thick data.
The analysis highlighted that the restaurant was struggling to differentiate itself and that customers had questions over the premium pricing at the restaurant.
By understanding the problem, that allowed Preriit and his team to explore the brand image at each of the restaurants’ locations.
The social and digital data, as well as the geofenced social chatter data, helped Preriit and the team to understand the cultural dimensions of the consumer. But the analysis didn’t stop there. Combining these findings with weather data and sales data highlighted what food the customer favoured in different weather conditions.
They could predict what the customer wanted and when to increase sales across the chain locations. This lead the team to create insights about what dishes to offer as specials depending on the weather, and also ideate to offer new menu items based on cultural preferences.
What This Means For You
I really liked the depth of Preriit’s framework for selecting and analysing data sources. It’s much more than entering in keyword searches into a social listening tool and using the data that is gathered.
For you, this means you have a framework to help you work through the best data and data points to analyse in your projects. You can see how he breaks this down and plugs the context holes with additional data sources.
Social data isn’t just data from Twitter, Facebook or Instagram and the additional insights hidden in meta data are also just as valuable. You need to consider what you’re currently missing from your studies.
An action for you today: plot out your next project with Preriit’s framework. And, do let us know how you get on.
Sources: Preriit’s paper was originally published at the Advertising Research Foundation (ARF) AudienceXScience 2018.
Join The SI Lab today to share your social data case studies.Join Now