Generative AI Streamlines B2B eCommerce by Clearing Back-End Bottlenecks 

Personalization will always win vis-à-vis a one-size-fits-all approach.

And that is increasingly true, even across commercial and B2B engagements. 

But an effective program that leverages contextual clues to provide a bespoke experience is a heavy technical and cost lift — even across consumer-focused digital platforms, let alone commercial-centric ones. 

Enter: the game-changing capabilities of generative artificial intelligence (AI). 

In a world where messy data plagues the products we buy, generative AI can now generate new attributes and correct existing ones.

“Wrongly tagged items were a really difficult problem to solve — at least in the past. But the advent of generative AI has proven to a be a really good use case when applied to this area,” Eli Finkelshteyn, CEO and co-founder of eCommerce search and discovery process solution Constructor, tells PYMNTS. 

By seamlessly integrating with upstream systems, AI solutions offer a minimal technical lift for existing customers, making this an attractive option for eCommerce platforms looking to win share from businesses across industries.

“A lot of the time you’ll find useful hints about attributes that should exist in something like a description or other textual data,” Finkelshteyn explains. “The AI can figure out from the context, mixed with visual data and clickstream latent attributes, that are missing or were tagged incorrectly.”

AI solutions particularly shine in B2B use cases. 

Read moreWhy B2B Tech Will Drive the Next Innovation Cycle

B2B Tomorrow Looks a Lot Like B2C Today

“A lot of B2B catalogs, they tend to be very large, and the data is typically tagged by humans. And especially when you’re talking about third parties doing it, there’s going to be a lot of messiness there,” says Finkelshteyn. “Nobody’s going to go through and manually curate 40 million of these, especially for some of these larger commercial datasets.”

It just makes a whole lot more sense to have AI solve that problem for you and do 90, 95% of the work, he adds.

By analyzing clickstream data and learning from customer behavior, AI can adapt and improve its search filters and results over time. This dynamic approach ensures that what commercial customers are seeing as they search remains relevant and effective in meeting their needs. 

“B2B has needed to put up with bad software for a long time, or one-size-fits-all software, that wasn’t really built for them,” says Finkelshteyn.

PYMNTS has long been tracking how today’s B2B tech innovations, which are inherently meant to organize information and better structure previously unorganized relationships, are now allowing firms involved to capture easy-win efficiencies that have historically been lost among legacy inefficiencies and manually reliant procedural gaps.

“AI shopping assistance, for example. I think that there’s a lot of potential promise there [for B2B]. It’ll be more a question of what’s the right user interface, and will users actually want to use those, and will they find it to be an easier and better experience than traditional search? I think the jury’s out, and we’ll see,” Finkelshteyn explains.

Read moreThe Trickledown Consumerization of B2B Payments Helps Firms Win Business

The Devil Is in the Data

If the world’s ongoing digitization has just one transformative impact, it’s that it is continually evolving the digital B2B marketplaces organizations use to meet their needs. 

But while removing legacy bottlenecks along the B2B journey is emerging as a key competitive differentiator for both buyers and suppliers, there remain no shortage of bottlenecks that still need to be plucked out of the process. 

“Right now, when that data is incorrect, it largely falls on the site merchants to fix. These are incredibly smart people who understand their businesses very well, and they could be doing strategic work. And instead, they’re correcting colors from green to yellow. That’s not a good use of anybody’s time. Being able to fix that problem for them as much as we can I think is a good thing for everybody involved,” says Finkelshteyn.

With its ability to generate new attributes, correct incorrect ones, and refine search filters and results, AI promises to revolutionize the way businesses handle product data, paving the way for enhanced efficiency and improved commercial customer experiences.

But what does the Constructor CEO think of generative AI more broadly? 

“The number one caution that I’d give to everybody is force companies like mine to prove that the value is actually there. Test the AI, make sure that it’s actually driving the KPIs, what is being promised — I think that will be what keeps generative AI from being a hype cycle that eventually ends, to AI being something that leads to actual, very real benefits across the marketplace,” Finkelshteyn says.