Machine Learning Provides Competitive Edge in Retail

A simple concept behind machine learning is proving that computation software can access a dataset and create results from that access. That concept also serves as the most crucial element in providing meaningful personalized service for customers.

In marshaling its resources, Amazon has begun to school retailers and search engines on how crucial an element machine learning is to a competitive environment.

Several Amazon advertising services are starting to rival Google in a significant business model for Google, online advertising. Amazon has been long offered Product Display Ads that feature product images and text that relate to people's searches. It just launched a few advanced advertising services such as a cloud-based header bidding service, according to MarTech.

More to the point of machine learning, Amazon is now beefing up services related to this technology. The company has announced a new program that will allow developers to build and host most Alexa skills using Amazon Web Services for free. It also introduced three new AI services -- Amazon Rekognition, which can perform image recognition, categorization, and facial analysis; Amazon Polly, a deep learning-driven text-to-speech (TTS) service; and Amazon Lex, a natural language and speech recognition program. The initiatives will bolster Amazon Web Services (AWS) against Microsoft and Google.

These product milestones for Amazon occur as the retail industry -- the most frequent users of personalization ads -- confront a complex puzzle of tech and trends. Retailers face a massive distribution transformation. Retailers are shifting away from their traditional locations. For instance, mid-level malls are losing stores such as JCPenney's, Sears, and Macy's. Other retailers are experimenting with smaller stores and kiosks in an effort to adjust their floor space. Even once online-only retailers such as Warby Parker and -- you guessed it, Amazon -- have added small brick-and-mortar stores to establish a cohesive consumer experience.

Changing Consumer Behaviors

Changing consumer digital behaviors are adding to the challenge for retailers. Behaviors such as "webrooming" and "showrooming" have become more popular over the last five holiday shopping seasons, and now have become standard activities. Webrooming and showrooming are when shoppers visit physical stores but use their smart phones to comparison shop and check competitive prices, and even place orders with a store's competitior. The adoption of these behaviors meant retailers had to improve their mobile sites, launch apps, examine beacons, and consider virtual reality to create a customer experience that supports the brand and retains sales.

All of this has raised the bar for correlating data variety for trends -- new sources, new contexts, and new intentions, all at different times. Managers who had just converted to the church of analytics now must listen to a new measurement sermon: where does machine learning fit within their business? And because of Amazon, retail managers are experiencing an urgency to learn machine learning protocols and also plan how to execute strategy in a world becoming dominated by a giant competitor.

Through its operational prowess, scale of services, and inroads into IoT devices and cloud solutions, Amazon has positioned itself to make a myriad of correlations between business metrics and technical metrics. I mentioned in recent posts that nascent search activity emerging from Amazon site visitors is rivaling search engines as a consumer starting point for researching products and services. Amazon can now take significant advantage with machine learning. Much of machine learning relies on data preparation, addressing data quality such as treating missing variables. Amazon has an opportunity to provide better context with the search conducted, and play a central role with partners who want to better understand how their products are received.

Amazon can then leverage its discoveries into meaningful customer and business value. A potential example is implementing tactics influenced by BizDevOps, a blend of front-end software development with business development and operations tactics. Its purpose is to align app development to customer and business value in upfront planning. That alignment has become critical as analytics has shifted from singular inferences from website activity into a central measurement of various activity across digital media and IoT devices. If you do a Google search, you'll find more than a few posts on the topic of BizDevOps mentioning Amazon as a model example.

Retail's Machine Learning Future

Amazon's potential with machine learning is a long way from the early years when Wall Street analysts criticized the once-only-a-book retailer about its quarterly losses. Amazon's machine learning potential also has far reaching implications.

Amazon's interest in personalization ads and growing machine learning prowess is tantalizing to supporters of programmatic advertising, which aims messages and gains access to highly targeted, highly valued audiences. Marketers can better predict how ad creative, products, and services can be combined to better appeal to customers in different cycles of the customer experience or a sale. Amazon can ultimately play a central role with platform partners who want to better understand how their products are received.

If this Amazon news makes your strategic team feel that they are behind the curve, take heart. The good news is that machine learning is in its early stages with retailers seeking ways to integrate data and devices that produce the data. Retailers turn to Google for search and paid ads because it covers a large number of industries, so Amazon will remain a retail niche for now.

But if business managers want to find potential success like Amazon has found, they must look internally with technology teams to see how machine learning techniques can be the operational glue between business resources and personalized experience for customers.