Tech

Optimizing for conversational AI product discovery starts now

It’s not too soon for retail brands to prepare for a future in which search looks less like Google and more like Bing.
article cover

Francis Scialabba

5 min read

Between inaccurate statements about a NASA telescope and the (presumably) very annoyed wife of a New York Times tech reporter, the chatbot search hype has gotten off to a strange start.

The conversation and concerns around the rapid development and mainstreamization of large language models (like those that power ChatGPT, Google’s Bard, and the new Bing) aren’t new.

But the possibility that consumers may find products in a chat-style interface brings a new challenge for retail brands that have built web presences optimized for search as we know it, and begs the question: What does the future of product discovery look like?

And while there are clearly some kinks to work out when it comes to the accuracy and readiness of conversational AI tools, retail industry and AI experts say the time to start thinking about the future of search is now.

Conversational starters

Now, the retail industry isn’t new to AI—far from it.

Conversational search itself is still new, but could soon be a viable option for consumers, said Sameer Maskey, an AI professor at Columbia University and the founder and CEO of Fusemachines.

“The releases [of conversational search engines] will get wider and wider, and soon a lot of people—especially if they start to be able to confidently say the information they’re getting back is quite accurate—they’ll start relying on it more,” Maskey told Retail Brew.

With that in mind, the retail industry is preparing for a shift.

One seemingly obvious function impacted by conversational AI is search engine optimization. But Jason Hartley, head of search, social, and shopping at digital marketing agency PMG, says optimizing for chat search is more about cranking up content quality, and less about throwing years of strategy out the window.

“SEO has evolved so much over the years and I think that at its core, it’s not going to change,” Hartley told Retail Brew.

But the content takes on a new role in something like the Bing search because chatbots use it to generate responses not written or reviewed by a brand, he added.

That might mean product descriptions need to be longer and more detailed, and websites should be written in a more conversational way.

In a sense, it’s changing the world order, said Maskey: “The world has been run with keywords…but I think conversational interface…is able to take account of context information and dependency of words across very long phrases and paragraphs much better,” he explained.

Retail news that keeps industry pros in the know

Retail Brew delivers the latest retail industry news and insights surrounding marketing, DTC, and e-commerce to keep leaders and decision-makers up to date.

Broadening ad horizons: Nancy Mao, director of paid search at ad agency RPA, is part of the Bing Chat beta, and likened the current fixation with chatbots to voice search, when Alexa and Siri had their moments in the spotlight.

“It’s definitely something to keep an eye on, but I personally don’t know if it’s going to become anything huge until advertising at a large scale will be available on it,” Mao told Retail Brew.

At present, retail brands can advertise in Bing chat through shopping ads, in the form of links below Bing’s answer, said Mao. (Google has yet to disclose how ads will work inside Bard.)

“Our way of preparing is to invest more in broad match keywords,” said Mao. That allows brands to target phrases related to their products that don’t contain the keyword itself, she explained.

  • For example, instead of targeting just the term “workout leggings,” a broad match strategy might also target “running leggings,” “running outfits,” and even “What are the best leggings for running?”

A (slightly) new world: So, what exactly will conversational AI change in the retail world?

To start, AI-generated responses means more uncertainty about the message that reaches a chat-using consumer.

“I’m sure brands will have some concerns around this,” said Maskey, pointing to the so-called “hallucination effect” in natural language generators, which refers to confident responses not backed up by training data.

Stephen Curial, CTO at ecommerce sales platform JungleScout, said conversational search is likely to pave the way for an entirely new industry.

“In the same way that traditional search created spin-off industries, such as search engine optimization (SEO) and digital advertising, we will see something similar with AI and language learning models,” Curial told Retail Brew via email. “Once they are available, early adoption of these new functions will be crucial for brands as they succeed in using these systems.”

Internally, team structure may change for retailers who decide to focus on AI, said Maskey. For example, there’s likely to be an increasingly broad overlap between data teams and marketing and search teams.

Hartley and Maskey agree it would be a mistake to assume that generative AI is taking away the work of retail media and marketing experts. Instead, jobs will evolve.

“New kinds of skills will be required,” Maskey said. “All retailers should be thinking about how to upskill their employees.”

Retail news that keeps industry pros in the know

Retail Brew delivers the latest retail industry news and insights surrounding marketing, DTC, and e-commerce to keep leaders and decision-makers up to date.

R
B