Technology ushered in the ecommerce era, and technological advances and evolutions continue to shape the way we shop and sell online. The importance of digital channels goes well beyond the functional, transactional relationship and deep into branding, communications and customer service. Now, the way in which brands and retailers are set up to deliver ecommerce services to their customers is changing again, thanks to machine learning which transforms their product data.
Machine learning is often discussed as part of the future of retail, but in some use cases, it’s not the future – it’s the present. The key to successful ecommerce deployment of artificial intelligence and machine learning is in identifying the tasks in which it can outperform a human. Product data mapping is one such task.
Why is product data so important?
Creating high quality digital representations of products is essential to success. The quality of product data influences every stage of the consumer’s journey to purchasing online – and online research often underpins offline commerce too, so its importance is wider than the numbers can directly show.
For example, search engines have a common goal to display relevant results and show the consumer what they’re looking for. Product data, formatted correctly and fully enriched, allows the search engines to understand the product better and therefore rank it more highly for the most relevant searches. High quality product data increases visibility and discovery.
This goes for paid search too, whether Google or Amazon – better relevance means lower bids are required for the same rank. That means that higher quality product data translates into lower cost per acquisition in paid search campaigns. Knowing more about the product also helps to convince customers that it’s the right item for them. Descriptions, titles and other copy all should add to this increase in conversion. When the customer knows exactly what they’re getting before they purchase, that’s an easier and better experience for them, and reduces the risk of returns.
In short, the dry term ‘product data’ sums up every aspect of your products’ representations online. Improving those representations has major benefits in visibility, conversion, CPA, and customer satisfaction.
So what benefits does machine learning offer in product data transformation?
The reason for machine learning’s success in the specific field of product data transformation is that it’s a job which is impossible for a human to do. When mapping product data to different channels, listing team members have to keep in mind hundreds of thousands of parameters, such as optimal ways to describe each field, the different field names on different channels, category and sub-category mappings between channels, et cetera.
While no human could maintain a perfect track of the right options for these huge clouds of mappings, machine learning software can be programmed to train itself on the categories and structures. Even better, when a channel makes changes to its data structure, as often happens, the algorithmic intelligence can adapt to accommodate the change.
This means that machine learning algorithms can intelligently map products to relevant channels, in a way that includes all the right attributes, phrased in the right way for the channel to understand and surface to shoppers.
Human listing teams try to achieve this but it’s essentially an impossible task to do manually, as even with only a handful of sales and marketing channels, it takes an incredibly long time to optimise a few thousand products for each channel. Realistically, human listing teams aim to get a minimum viable standard rather than the best possible quality of listing.
A new mindset
Part of the reason for the change that’s coming to ecommerce now is a changing mindset amongst retailers, brands and technology companies. The current and previous generations of ecommerce technologies have been all about moving data around – perhaps sending it to as many channels as possible, or just getting it all into one place for your internal team.
Cutting edge retail technology is no longer a ‘dumb pipe’ for data to be pumped through. Instead, there’s a focus on improving, enriching and adding value to that data at every stage. This mindset can help retailers and brands to deliver better experiences for the customers and results for their stakeholders in a model which is scalable and constantly improving.