Innovation needs to be part of your culture. Consumers are transforming faster than we are, and if we don't catch up, we're in trouble. --Ian Schafer
As the quotation mentioned above by Ian Shafer states, consumer tastes, wants, and needs are changing faster than the retail products and marketing are. It is vital that retailers keep up with the ever-changing customer wants and needs; otherwise, they will go out of business. Fundamentally, the question that continually needs to be asked and answered is what does the consumer want ? The next question that needs to be answered is how do retailers (and digital marketers) keep up with the continuously evolving retail landscape?
Data-Driven Innovation: Keeping up with the consumer
The way to keep ahead of the consumer's changing needs is to use Big Data and its associated methodologies to forecast future trends. The terms "Big Data", "Predictive Analytics", and "Machine Learning" have become watchwords for the retail and digital marketing industry. More specifically, the requirement is for predictive analytics and machine learning to combine their strengths to manipulate raw data to predict the unknown. This forecasting information, in turn, will help buyers, retail management, and digital marketers in their decision-making processes.
Let's look at the role predictive analytics plays in the digital media marketing space and use a case study to highlight how Big Data and machine learning can assist marketing strategists to make the right decision.
Our case study
A high-fashion e-retailer's core business is to sell affordable high-fashion clothing across the globe. They have set up several warehouses, one on each continent which makes it easier and more cost-effective to ship the purchased items to their customers. Business is good, and they are ranking at the top of the search engine results page.
However, they are acutely aware that there is intense competition in this space, and they have contracted a digital marketing agency to analyse their current statistics and come up with a marketing strategy that will allow them to upscale their business.
Therefore, here is the basic process that the marketing agency will follow to assist the high-fashion retailer to drive innovation and change in our complex and fast-changing post-modern world (and stay ahead of the competition):
The first step is to collect as much data as possible about the brand, the retail environment, and competitive brands. For example, generalised consumer behaviour, sales figures, website analytical data, social media data, as well as demographical data.
The takeaway point is that as much data about the brand and its relationship to the consumer should be collected. It should also be noted, however, that the data has to be relevant to the brand and the retail environment. It does not help collecting masses of data in case it might bear relevance to the topic. The more connected the data, the better as it will add depth to the information required to predict future trends.
Transform data into information
Once the source data has been identified, the next step is to extract it from all the different sources, transform or cleanse it so that it fits into the data warehouse tables, and upload it to a data warehouse where it can be manipulated and analysed.
It is best to house the data warehouse on an off-site server or cloud storage to allow for easy access by all the different global decision makers.
The next step is to convert the data into useful statistical analysis that is used by both the digital marketing agency and the company's senior management. There are two approaches that can result in useful information.
The first is to utilise predictive analytics in the form of a recommendation engine (machine learning). This platform takes the data and constructs a typical consumer persona for the many different types of people who have purchased goods from the e-retailer. This engine will build as many different personas as necessary by looking at elements such as buying trends and demographics.
Finally, the recommendation engine will, by using the current data, then predict or forecast what each persona is likely to buy in the future.
The recommendation engine and the model built on a neural network are similar; however, the neural network model utilises several hidden nodes or layers to make predictions to answer the questions asked of it.
Let's assume that the question that the business is asking is how many rain jackets they should purchase from their suppliers based on the number of rain days the world has had over the last two years. They also want the answer broken down by geographic region so that they know how many jackets to send to each warehouse.
The first step is to access the total rainfall data per geographic area for the last 720 days. This raw data will not necessarily be in the correct format. Therefore, once the data has been extracted from its source, it needs to be transformed so that the neural network engine can analyse it and make realistic predictions.
The neural network can only analyse numerical data. It cannot work with categorical data or text. Therefore, days of the week and geographic regions need to be hard-coded in a binary format.
The next step is to extract, transform, and upload global population statistics per region as well as the number of rain jackets bought in the last two years per gender and size. Again, the categorical and textual data like gender and age demographics need to be hard-coded in a binary format.
Once all the data has been transformed, it's time to build and train the model. After the model has been trained and tested, it can be used to predict the number of rain days per region for the coming year as well as the number of rain jackets that are likely to be purchased by the consumer.
Finally, this information can be transformed into charts, graphs, and 3D-models, which in turn will be utilised by the brand marketing strategists and company executives to plan future marketing campaigns and adjust the current marketing campaigns.