Editor's note: The following is a guest post by Amir Konigsberg, co-founder and CEO of search firm Twiggle. Views are his own.
Walmart made waves with its launch this spring of Jetblack, the retailer’s concierge shopping service targeting "time-strapped urban parents." Currently available by invitation only in Manhattan and Brooklyn, the service is generating buzz, not because it’s seen as a game-changing new force in e-commerce – fear not, Amazon Prime – but because of the technological changes it may help spur.
Here’s how it works: Members text requests for products, from diaper bags to designer shoes. Jetblack then attempts to source the products from Walmart and partner stores, delivering them directly to consumers’ doorsteps – within 24 hours for household items, 48 hours for all others.
How does the service ensure that it gets what members really want, considering that all it has to go on is a truncated text message? Jetblack approached this task by utilizing human intelligence alongside AI techniques, which reflects the inherent difficulty in such a remarkable undertaking.
"Jetblack makes sure that if you order a 'red dress shirt' for instance, you get a red shirt and not a red dress delivered to your doorstep."
The rise of chatbots and voice-activated shopping has created a fervor for natural language processing (NLP) and natural language understanding (NLU) techniques that can facilitate satisfactory human-computer dialogue. Utilizing machine learning, NLP and personalization algorithms, Jetblack makes sure that if you order a "red dress shirt" for instance, you get a red shirt and not a red dress delivered to your doorstep.
But until now, the widespread use of NLP technology hasn’t always translated into a palpably improved consumer experience. To effectively respond to the needs and expectations of today’s consumers, however nuanced their request may be, or however niche the product they desire, the e-commerce industry must also leverage an intimate understanding of user behavior to deliver the same level of service consumers would expect from an experienced salesperson. It’s not enough to understand just words, these services need to understand context and intention, and that’s very tricky — even humans don’t get it right every time. That’s one key reason why the Jetblack initiative has sparked widespread interest in the e-commerce sector. Can this texting service actually be made to feel personal?
Hurdles to success
Benefiting from the endless resources of a retail giant, Jetblack may yet succeed where others’ efforts - e.g. Facebook M, GetMagic - seem to have come up short. But the challenges confronting Walmart are still considerable.
There is no more essential ingredient to Jetblack’s mission than understanding. The busy urban moms the service is targeting have no time to deal with chatbots that don’t quite get them, or which require them to answer five different questions before producing a relevant result. Reproducing an in-store shopping experience requires a high level of conversational fluency so that shoppers can effectively navigate among choices in a way that feels natural, conversational, intuitive, and is ultimately helpful.
With a text message user interface, Jetblack has almost no margin for error. Its understanding of shoppers’ requests needs to be 'on message,' lest its members quickly lose patience. It is relatively easy to train a machine to understand and follow structures and rules, but capturing shopper intent also requires understanding flexible communication processes involving multiple levels of linguistic analysis. Machines must take into account syntax (the grammatical structure of the text), semantics (a text’s meaning), and pragmatics (the underlying purpose or goal of the text). Our brains do this with relative ease. Can a computer do the same?
For Jetblack, the question remains unanswered – but whether the service thrives or comes up short, it may well pave the way for further advances in the new, more personalized era of e-commerce.
A personal touch
Creating a personal shopping experience, even if it’s powered by sophisticated technology, is simply not possible without a personal touch. Technology is only as good as the brains behind it.
Recognizing this, Jetblack pairs human intelligence with AI and machine learning. While the rapid ascent of e-commerce attests to consumers’ growing demand for convenience and near-instant gratification, the industry still has a long way to go in delivering both convenience and highly relevant, personally tailored shopping experiences.
Combining the best of online retail with the brick-and-mortar experience need not be a pipe dream. Jetblack may not be the only vehicle ushering in the next era of e-commerce, but the results of this experiment will be closely watched for what they portend for a nimbler, smarter e-commerce industry which may just utilize clicks and scrolls less and less, leaning much more on elements of the in-store shopping experience with which we are still familiar.