Cash Converters was established in Spain in 1995, specialising in the purchase and sale of second-hand goods, as well as other secondary services, such as microcredits or recoverable sales. Since then, its physical establishments have been its main strength: with more than 80 stores distributed throughout Spain, customers identify the brand mainly with the traditional store.
As a result of the economic crisis, competition in the sector has intensified in recent years, with companies clearly positioned following a very defined strategy. This situation led Cash Converters to rethink its global strategy, determined not to give up its leadership in the second-hand market.
To achieve this objective, action had to be taken to strengthen the digital sales channel, which had been discreetly pushed into the background in the consumer’s mind. One of the pillars of this strategy would be the complete redesign of the digital store, completed in 2018; the other, the knowledge obtained from the analysis of the connections between the online and offline world.
Extracting knowledge from physical and digital data
We know that we have become omni-channel consumers: we contact the brand through different channels, we access the digital store from different devices, we research online and buy offline, or vice versa, etc. How can we take advantage of these overlaps in the case of a brand that is essentially identified with the physical, but wishes to reinforce the role and importance of its e-commerce? Simple: by using real-world information to strengthen their digital presence.
“We investigated the connections and mutual influence between the online and offline universes”.
To strengthen the growth of Cash Converters’ digital business we developed a data-driven project, based on data and insights, which would investigate the connections and mutual influence between the online and offline universes, and then transfer (through digital marketing campaigns) the knowledge obtained to the new brand positioning.
We started with three data sets to work with:
- Physical stores: from each of them we knew the number of visits, the location and the turnover. Through the ratings and comments that customers left on Google, we could also extract the sentiment that each store generates.
- Digital store: for each operation, we knew the location of the customer, the price paid in the transaction, the object bought or sold and the origin of the product.
- Socio-demographic variables: from almost 12,000 postal codes throughout Spain, we could obtain information such as the level of income and expenditure per household, population density or the number of people living within a radius of 5 kilometres.
Connecting the ‘on’ with the ‘off’
Does the proximity of a physical store influence brand awareness? Does consumer behavior vary depending on where they live? Does a visit to the physical store influence digital shopping? To answer these and similar questions, we started working with our data.
First, we took the postal codes, which cover more than 45 million people throughout Spain. We located the physical Cash Converters stores and used GIS (Geographic Information System) to calculate the distance from each point in the territory to the nearest store; then, in order to delimit the area of influence of each store, we divided the zip codes between those who have a store nearby (a radius of less than 20 kilometers) and those who do not. We got our first insight: about 25 million people have a store less than 20 kilometers away.
We could already start comparing results: is there a relationship between distance to the physical store and consumption in the online store? The results show us that there is, which allowed us to identify a first awareness problem: consumers know the store by its physical presence rather than by its digital one. But was there a relationship between distance and behaviour in the physical store? Yes; mainly regarding the purchase of objects by Cash Converters (more purchases in the areas near the stores).
Dissecting the consumer as much as possible
Using Machine Learning technology through BigML and programming developed in Python, and data visualisation with Tableau, we proceeded to cluster the results. We obtained a total of twelve groups: six clusters for consumers with a store nearby, and another six for those far from the stores. This classification outlined each type of consumer (detailing variables such as population density, income level and spending), which allowed us to segment geomarketing campaigns as much as possible.
In the case of consumers with a nearby store, we incorporated one more variable: the quantitative (from one to five stars) and qualitative (opinions and comments) reviews of the physical Cash Converters stores that consumers left on Google. This allows us to distinguish the strengths and weaknesses of each store (product offer, prices, treatment, after-sales service, etc.) and to assess how they affected the consumers in the area.
Finally, clustering provided us with twelve consumer profiles. This will allow Cash Converters to segment its digital media campaigns according to variables such as distance to the store, type of store or digital and physical consumption patterns, seeking objectives such as increasing digital and physical cross-selling, correcting the weaknesses of physical stores and taking advantage of their strengths, or increasing awareness among consumers who are far from a store.
A complete cocktail of marketing, data and technology, with the added difficulty of working with data from very diverse sources (socioeconomic variables, Google reviews and internal company data), to integrate them into a single output of value for the customer.