Why every brand will capitalize on deep-learning to optimize revenue & margins

By Moataz Rashad

Every brand has become an online retailer, some managing their own e-stores, and some relying on hosted solutions such as DemandWare. Similarly, every brand has invested in social presence, social listening, and social campaigns. In mobile, we’re definitely lagging behind the emerging markets, as many still have 2-star apps where a web experience has been shoe-horned onto a mobile screen, while an e-tailer like Flipkart in India has gone mobile-only.

Brands realize that they increasingly need to own the customer journey from intent to purchase, as they know the user experience (UX) is what inspires and retains users. A recent HBR article emphasizes how crucial it is for every brand to eliminate the guesswork in understanding customer’s emotional motivators. For online retailers, the study showed customers who are “Fully connected and satisfied and able to perceive brand differentiation” were 52% more valuable than the “highly satisfied” baseline users. The retailer study on the bottom line impact of the “Fashion Flourishers” category and its strategy impact on online-to-offline and store locations is well worth a careful look. However, most brands still face several key problems in understanding their customers’ motivators and optimizing their UX to get them to the fully connected stage.

Source: “The new science of customer emotions” by Magids, Zorfas, Leeman, Nov 2015 Harvard Business Review

Problems and decisions

For the vast majority of brands there isn’t much intelligence aggregation across these channels and touch points, nor is there a feedback loop to leverage the relevant analytics to inform the decisions various stake-holders are making every day. It’s decisions like these that need every byte and every pixel of intelligence you can gather:

- Is this the right collection for Thanksgiving week sales?

- Why is this recommendation converting against this SKU for San Francisco users, but not for Boston users?

- Is 20% lift in CTR good, or should we continue to refine to aim for 30% lift?

- Should we discount the handbags by 21% or 29% to meet X revenue goal in Q4?

- Does that Twitter or Tumblr photo featuring our product carry positive sentiment or negative?

Viewing excel sheets–or even business intelligence dashboards–that lack the continuous deep-learning behind the scenes on all these activities almost always puts you at a significant disadvantage. It gives you, the decision-maker, a partial, spotty view of your world. Also, because it’s already stale, you lack the most relevant real-time ammunition to make the smartest decision at any given moment. Compound this over time and across functions and you’re essentially flying almost blind, kind of like Wattny in The Martian during the scene where he punctures his astronaut suit and hopes to make it to the rope by free-flying through space. His endeavor was fun to watch, but it could have cost him his life and billions of dollars!

What is 360 degree AI-merchandising?

We call this deep-learning across all touch points to affect a smarter user experience, AI-merchandising. We’ll cover only 2 aspects of this in this article: product design and catalog design. Other elements will follow in subsequent articles.

Product design

A deep-learning engine that aggregates all clicks, purchases, save-to-wish-list, cart-abandonment, and returns for every product in your catalog gains internal insights (in the neural network) into what’s appealing and what’s not about your products. The information isn’t captured in some metadata attributes in a row or column in a database, it’s embedded in its “mind” and can be recalled. This intelligence can then be queried for suggestions on what to design. The engine can act as an AI-assistant to your designers, suggesting cuts, styles, patterns, colors, and fabrics for next seasons’s collection, which can even be refined by geography and demographic to meet revenue and margin goals.

Catalog design

Cross-category recommendations, often called “Style it with”, or “Complete the look” recommendations, are a unique discovery mechanism used to inspire shoppers to create ensembles from your catalog. It is one way to encourage shoppers to maximize basket size, and to love your brand even more as they dress head-to-toe from your catalog. A deep-learning aiCommerce engine would garner the intelligence to advise you what’s working well with what for your shoppers (for ex: what handbags and shoes with what shirts, etc) so your catalog collections can indeed deliver that head-to-toe ensemble.

No brand can afford to ignore #aiCommerce and deep-learning powered user experience. We’re very excited to collaborate with brands of all sizes on deep-learning and AI-merchandising to inspire shoppers with a futuristic and joyful user experience, while maximizing revenue lift.

Originally published at blog.deepvu.co.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.