MarTech Interview with Sanjay Mehta, Head of Industry, Ecommerce at Lucidworks

Sanjay Mehta, Head of Industry, Ecommerce at Lucidworks chats about the implications of new age technologies like ChatGPT on modern digital commerce and e-tail:

_________

Welcome to this MarTech Series chat, Sanjay, tell us more about yourself and your role at Lucidworks…

I am Head of Industry, eCommerce at Lucidworks. I’ve worked with thousands of ecommerce customers globally in my 20 year history in the space, including tenure leadership roles at major ecommerce technology players such as ATG, Endeca, Hybris, Oracle, Netsuite, and Reflektion. At Lucidworks I primarily focus on digital experience, search, personalization, and improving ecommerce results through applied machine learning and artificial intelligence.

When it comes to dominant ecommerce trends and predictions on the state of this segment for the near-future, what key thoughts do you have?

Thanks to maturation of ecommerce technology, particularly in AI, barriers to entry and operation of a successful online retail business continue to lower at a faster pace. This will lead to even greater saturation of players in an already crowded space, causing online retailers to focus more on technologies to minimize dilution of their brand.

I believe early adopters of these technologies with a niche or domain expertise will rise to the top particularly in the eyes of the consumer.  Trending technologies such as ChatGPT and Bard are setting new expectations among shoppers for how websites interact with them more naturally, particularly when it comes to search and chatbot applications. The implications of these new technologies to digital retailers is less about how well AI bots can perform traditionally human tasks and more about how these technologies are shifting peoples’ habits and expectations when shopping online. In order to respond to these expectations with minimal human oversight, organizations have to invest in the right technologies.

How are you seeing emerging tools and tech (even ChatGPT and similar) change the game for online commerce / online retail today?

Emerging tools, particularly AI-based ones are dramatically improving how customers find, discover, and identify products.  They are shortening the path to purchase while also improving shopper engagement. From the seller’s perspective, they are relieving merchants of tedious tasks such as manual rule building, providing unimaginable productivity boosts for tasks like personalization, and providing deeper insights on their customers and business overall.

Now let’s get into the tools – starting with ChatGPT since it is still warm off the presses. Conversational AI applications such as ChatGPT will undoubtedly disrupt the ubiquitous ecommerce experience we know today, but in a good way. It will cause sellers to rethink their entire shopper journey. The traditional static one-way experience that requires shoppers to think about how to navigate through a site hierarchy or which keywords to use with will give way to more of a fluid one based on natural dialogue. It will reduce effort on the part of both shoppers and merchandisers. It will take out a lot of the guesswork and testing that is done by merchants today as the entire experience will be based on natural interaction—shoppers will be telling the AI what they want versus merchandisers guessing what to show shoppers what they want.

ChatGPT is one of a number of AI-based technologies that online retailers are incorporating into their shopping experiences. Other applications such as computer vision are being used for both product identification, intent detection and product enrichment.

Semantic vector search is being used as a powerful intent inference tool to drive greater relevance. Predictive personalization eliminates the need to manually segment and target customers by automatically sensing their needs. ML-derived insights are being used more to analyze and help retailers better understand their customers, their behaviors, purchasing and product trends, and more to help merchandisers refine their strategies.

What we expect to see is all of these technologies intertwined into a conversation-led experience.

Marketing Technology News: MarTech Interview with Darby Sieben, Chief Product Officer at Unbounce

How in your view can brands drive more impact with the right use of emerging tools and tech: a few best practices they should keep in mind when deploying these technologies?

Employing the right tool at the right time and for the right job is key. To do that, you need to understand your consumer, their journey, and their behavior. Where does the most traffic that results in conversions come from?  Does it start with customers coming directly on your site or app? Social media? A search engine? YouTube/TikTok videos? Email or SMS campaigns? A chat app? Ads? The list goes on.

There are specific categories of AI tools that are purpose-built for each of these channels which can be integrated into your underlying ecommerce infrastructure. Here are a couple of examples:

• For direct traffic, a predictive personalization tool will automatically curate the experience and results aligned to the shoppers predicted intent using their context, behaviors, similar shopper behaviors, and any historical behaviors.
• Social media tools can use the shoppers’ likes, interests, posts, and examine the content within to determine what products are ideal for them.
• Tools like ChatGPT and semantic vector search can be used to both enrich the information value of catalog and landing pages, enrich product data, content, and more to help drive organic search engine traffic.
• Computer vision as well as AI tools for social media can both be used in the context of YouTube/Tiktok/Instagram to extract intent of the shopper and connect them with the appropriate products through images and video frames.
• For campaigns through email or SMS using relevant AI tools can help determine the audience, when to send, generate and personalize content, and uncover response insights.
• Chat apps and chatbots historically have had more presence in the customer service function, but thanks to the awareness created by ChatGPT, we will see even more adoption in the pre-purchase funnel.
• Finally, there are numerous tools to help drive return on ad spend (ROAS). These help determine the optimal keywords to buy, when to buy them, and even suggest or generate the content of your ads and landing pages to drive the highest ROI.

Once the shopper is within your site, applying the correct AI tools for the job still applies. Using search as an example, AI-based technologies such as natural language processing (NLP), relevance algorithms, personalization, semantic vector search, recommender algorithms, speech to text, and computer vision are just a handful of examples.

Providing relevant search results at scale is one of the most important, and most difficult, ecommerce challenges I see online retailers suffer from. A recent survey of search practitioners showed that 100% ranked search relevance as highly important, while 96% said it is difficult to deliver. Complexity is inherent due to gaps in data and signals required, the analytics needed, and the actions and triggers that occur to support large commerce search programs.

Emerging technologies including semantic vector search can compensate for these common data challenges online retailers such as poor or sparse product information, analytic / behavioral data, and customer intelligence.  Semantic vector search can analyze not just the keywords in the query but also the meaning associated with them to provide more accurate and relevant results. They can also automatically detect low-performing queries and map them to high-scoring results that shoppers love.

Machine learning algorithms can automatically index and analyze large volumes of data from various sources, such as structured and unstructured data, product views or catalogs, and customer interactions. This allows the platform to understand the context, intent, and relevance of the data, improving the accuracy and speed of search results while diminishing challenges related to poor data.

AI is also opening more doors on the way shoppers can search. Computer vision based models can be used to support search using photos or image uploads.  Speech to text based models can be used to support voice based search commands. These can be combined with large language models such as ChatGPT to drive a more interactive guided selling or search experience as well as for self service applications in ecommerce.

88% of search practitioners believe that AI is very or extremely important in the future of search. While AI will not solve all search issues, harnessing what it does best—complex processes at scale—is a natural application to the immense data sets and models that help make search and browse more relevant for your most valuable customers.

For brands who are looking at growing their presence in a competitive online market through 2023, what top of mind tips would you share?

In a recent international survey, nearly 100% of shoppers say that the search bar is important to their online shopping experience, and 40% get a negative impression of a retailer if what they’re searching for doesn’t give them the results they want. Your brand’s search performance needs to be locked down. A 2021 Google survey said that 90% of consumers believe an effective search function is very important or essential. In the US alone, retailers lose around $300 billion in revenue to poor online search experiences, with 85% of global online users stating that they count it against the brand after they’ve had a poor search experience. In other words, brands absolutely cannot afford bad search.

Marketing Technology News: Global Digital Ad Spend Trends and Best Practices

A few common misconceptions about ChatGPT you’d like to dispel before we wrap up?

ChatGPT cannot wholly replace the expertise of your ecommerce team. Yes, it has raised the bar of consumer expectations in an incredibly short amount of time. And yes, the thinking is if ChatGPT can spontaneously compose a college-level essay on string theory, why can’t their favorite retailer show the perfect pair of pants. The reality is that for LLMs such as ChatGPT, out of the gate generalized outcomes are indifferent to your niche/domain, brand, individual customer’s preferences or KPI’s. These models still need to be trained, prompted, and tuned by humans with that expertise to truly generate value in ecommerce.