By Ralf Gladis, CEO, Computop
Size is not everything when it comes to using data for good effect in any business, and this goes for retail too. The insight that a retailer, online or off, can derive from good quality data is not determined by how much there is of it, but rather by how it is collated, analysed and used to meet their particular customer’s requirements.
The questions can be myriad. Where will demand be particularly high next weekend? How much influence will the weather, or a major sporting event, have on online sales? Under what circumstances is the probability of fraud or returns particularly high? Why does the customer behave like this and not differently? The answer to all of these questions is in the data.
Data is the oil that is lubricating the sales machines at huge online retailers like Amazon, and is exploring user behaviour for tech giants like Google and Facebook. According to Amazon Web Services (AWS), its’ Payments Data Engineering Team alone is responsible for data ingestion, transformation and storage of a growing dataset of more than 750 TB. That enormous volume will dwarf that of most other, smaller organisations, but this doesn’t mean that their data is any less valuable or that there is no room left to compete.
Computop, for example, manages 160 billion data points for its customers, all of whom are using its payment processing platform. Its customers, however, range widely from small to large and managing their data is simply a question of scale. Every single one of them is looking to Computop to unearth the treasure that is contained in that data and evaluate it to make it truly valuable.
But as data volumes grow, the next consideration is how best to manage it. Can we still rely on good old-fashioned statistics, or should we be harnessing artificial intelligence and Big Data? Many questions can be answered by a combination of Big Data and statistics, particularly in companies that are very familiar with their data and the information it is providing. The challenges start if there is no capacity in-house to use a specialist statistician or a suitable Big Data tool. Of course, the larger the database, the more difficult it is to understand the data and the relationships between the data. Statisticians are powerless without knowledge of the connections.
At this point artificial intelligence (AI) really needs to be considered because it is the best way to help retailers, and other organisations, to evaluate data and relationships to gain a better understanding of buyer preferences and predict future behaviour. The danger of not doing this in today’s fast-moving commerce environment is that customer expectations are not met, and competitors quickly move into the available space.
Retailers should not worry about the amount of data they have. Size is not an issue when it comes to AI. Every question and every AI project is based on specific knowledge with specific data regardless of size, so the right questions coupled with the right data mean that a medium-sized retailer whether online or off-line can achieve just as successful results as even the largest players in the market.
Where to start? AI today is at the point that IT was back in the 1960’s, in other words, still very much in its infancy. In truth only a few companies have the expertise, the data competence and the technical staff available to manage its implementation. For this reason, many retailers are outsourcing the management of their data to a service provider.
The advantage of this is that the expertise, particularly in AI, has already been built, which means that the data is in good hands and with those that they trust. Often the company is already providing a service to the retailer, in the same way that we provide payment services, so the data is familiar to the provider. This allows insights to be drawn more quickly and more accurately resulting in faster results that can be implemented.
Before going rushing in however, retailers who are new to the use of AI should ensure that any third party they use is able to provide GDPR-compliant anonymisation and AI data preparation and that the prepared data can then be transferred to AI tools such as Python and TensorFlow to test suitable mathematical models. If the provider is able to automate it on their platform, then this just reduces the workload for the retailer.
AI is transforming the way we do business, helping retailers to realise the treasure they have at their disposal hidden deep inside their databases. To compete successfully, they just need to find the right partner, or expert, to tease it out and turn it into meaningful insight.