The Gist
- Fast vs. fun shopping. The spectrum of online shopping could evolve drastically with Amazon’s AI integration, shaping a new social commerce landscape.
- Generative AI impact. Harnessing AI for product storytelling could dramatically alter how online products are promoted and perceived.
- Content revolution ahead. As generative AI lowers content costs, rich, engaging digital experiences could dominate ecommerce, influencing culture and consumer behavior.
If Amazon were to acquire a top social media company, what would ecommerce look like in 10 years? Buying goods through social media is nothing new — nearly half of U.S. consumers have done it.
However, generative AI ecommerce could change a store into more of a story. That, in turn, would reshape the work of ecommerce teams, marketers and administrators of digital asset management (DAM) and product information management (PIM) systems.
Whether or not Amazon buys a social platform, the value here is in the thought experiment. The marginal cost of content creation is plunging toward zero, and AI ecommerce has the analytical power and creative means to remake barebones web experiences rich with media. How that affects the experience of shopping is a multitrillion-dollar issue.
Fast and Boring or Slow and Fun?
Online shopping occurs on a spectrum between two opposites: fast and boring or slow and fun. Amazon, Walmart and eBay are generally fast and boring. They are made for planned purchases. In fact, 50% of US consumers start product searches on Amazon compared to 31.5% on Google and only 2% on social media. The experience is about speed — get in, find what you need, check out. No one “hangs out” on Amazon.com (though they might watch plenty of Prime Video content).
Social commerce on TikTok, Instagram and Facebook is designed for serendipitous discovery and unplanned purchases. Between influencers hawking products and ads in the algorithmic feed, users never know what they’ll find. But users will only tolerate so many ads or influencer plugs before disengaging. That limits the AI ecommerce potential.
Others fall midway on the fast-boring vs. slow-fun spectrum. For example, countless media outlets earn affiliate marketing revenue when a reader clicks a link to a product reviewed or mentioned in their pages and then buys it. Generally, reading about a backpack on a gear review site is more fun than reading about packs on Amazon — and it’s faster than waiting for social media to serve up shoppable content. But maybe that changes when social media and ecommerce are under one owner and augmented by generative AI ecommerce.
Related Article: Generative AI in Marketing: Unlocking the Next Generation of Use Cases
AI Ecommerce: Fast and Fun
Let’s say Amazon, owner of a hypothetical social media company, creates a database of all the social content that drives purchases. It uses generative AI ecommerce to analyze this dataset and find patterns — about the characters, storylines, jokes, dances, emotions, types of goods, video lengths and styles, etc. For a wide range of product types, Amazon finds replicable formulas for videos that drive purchases.
Amazon could use a generative AI like Moovly or OpenAI’s Sora to turn any product listing into a product story — for a fee, perhaps. Alternatively, Amazon might sell subscriptions to that database for brands, which then use an AI ecommerce service of their choice to create their own stories for product listings.
With this new slew of social AI ecommerce content, Amazon could add a feed view to the usual retail media (i.e., sponsored product ads). That would become another place to run ads. Maybe Amazon even mixes in organic social content so that the experience isn’t too salesy or pushy. Of course, extensive user purchase histories and social platform user data would shape what the feed contains. Potentially with generative AI ecommerce, the Amazon homepage becomes addictively entertaining like TikTok but still drives purchases like Amazon. It becomes an optimal blend of fast and fun.
DAM and PIM to Match
In 10 years and considering the impact of AI ecommerce, maybe we will shop stories instead of stores. Even ecommerce customer reviews could take the form of social content instead of the dry text we usually see today. Arguably, DAM is already progressing toward a future like this.
Kara Van Malssen, partner and managing director at AVP, a DAM consultancy and partner of Acquia (my employer), shed some light on this recently. She argues that in DAM, “There will be a feed-like experience that showcases the latest, most timely, and most relevant content, specific to that user.” That feed, informed by AI, could base recommendations on content analytics drawn for all the places where the content appears. Why not fill the feed with the content likely to perform best?
Van Malssen also predicts that DAM search will resemble an AI prompt box. When users describe what they’re working on, DAM will either find the best images or generate them on the fly — even a product video for ecommerce listings. If the generative AI is trained to execute and enforce brand guidelines, maybe it can bypass typical creative processes with briefs, edits, approvals and so on. DAM, suggests Van Malssen, “may be more of a training hub for the internal AI than a repository of content to reuse.”
The DAM as a training hub, constantly processing and learning from content analytics, would be a formidable tool for generating richer, more dynamic AI ecommerce listings. That said, generative AI ecommerce systems will need training in how to interpret product images and product information. For example, if a shirt comes in navy blue and black, it’s not helpful for AI ecommerce to depict that shirt in dark green. Similarly, it’s problematic if the story for a trailer hitch shows it towing something that hitch isn’t rated to tow. The text input in PIM systems will need human oversight because generative AI can turn one instance of a mistake into thousands, especially if the AI automatically syndicates content to ecommerce sites.
Related Article: Navigating the Impact of AI on Digital Asset Management Jobs
Shopping, if Still We Want It
Plenty of barriers could prevent AI ecommerce from turning a vanilla ecommerce site into a more immersive place. For one, AI is still notorious for hallucinating. On a small scale, people can oversee and approve the content AI churns out before it’s published. If automated AI ecommerce content creation were scaled to thousands of listings — or to customizable products made using 3D printers and weavers — human oversight would become difficult and eventually impossible.
The greater barrier might be generative AI itself. As personal assistant AIs get to know their users, they may choose products and make purchases semi-autonomously. Does a compelling product story matter if no human being sees it? Would it influence the decisions of personal assistant AIs?
The point is that cheap, mass-produced AI ecommerce content means there will be more content in every channel. That content, in turn, could influence daily life, culture, and commerce on a colossal scale.
Learn how you can join our contributor community.
