In the wake of OpenAI’s explosion in popularity, every other ecommerce and marketing technology company appears to be hitting the market claiming to have an artificial intelligence solution. But the Federal Trade Commission has put opportunists on notice, writing in a blog post last month that “false or unsubstantiated claims about a product’s efficacy are our bread and butter.”
Ecommerce companies looking to adopt an AI and machine learning-powered solution should take note. It’s because those are the very technologies that so many companies find appealing that the FTC is pursuing false and exaggerated AI and machine learning claims. For companies looking to scale fast and navigate highly complex technical problems like optimizing distribution networks and analyzing reams of data, the promise of AI and ML looms large.
But how do ecommerce companies, following the FTC’s example, separate fact from fiction? They can start by learning about the kinds of claims the FTC plans to investigate, the established use cases for AI and ML in ecommerce, and how to safely evaluate these technologies to determine which ones are a boon to their business and which are bogus.
What Kinds of Claims will the FTC Possibly Penalize?
The FTC’s blog post lays out four criteria the agency will use to examine AI and ML claims:
- Are you exaggerating what your AI product can do?
- Are you promising that your AI product does something better than a non-AI product?
- Are you aware of the risks?
- Does the product actually use AI at all?
The costs for advertisers that make false claims in this area are huge. Examples suggest violations can bring fines of six or seven figures. But the costs for ecommerce companies shouldn’t be taken lightly either. In a downturn and during a time of high paid ad costs, ecommerce brands can’t spare dollars on digital snake oil.
What are some of the Established Use Cases for AI and ML in Ecommerce?
Machine learning has been playing a role in enhancing the experiences of ecommerce brands and customers for some time now. It’s helped consumers through personalized recommendations and automated customer service, and ecommerce shop owners have seen their workload streamlined through supply and demand management, fraud detection and churn prediction.
Before AI and machine learning, now-common practices like dynamic pricing were largely manual. Dynamic pricing used to solely involve historical data and intuition, which could be time-consuming and unreliable. But AI and ML allow brands to analyze large amounts of data in real time, such as competitor prices, supply chain costs and customer demand patterns. This quick, scalable data analysis, which was previously impossible, is fueling innovation in a number of ecommerce-critical functions, not just pricing.
One of the next frontiers in ecommerce companies’ use of AI and ML is marketing attribution. Machine learning allows marketers to better analyze and understand customer journeys, conversion and retention. Given the amount of data collected by attribution platforms and the amount of time required to effectively analyze and understand it, AI and ML are essential in unlocking marketing attribution’s full potential.
In all likelihood, if a marketing attribution tool isn’t using AI and ML, by the time marketers have taken the time and steps to analyze what is a huge amount of data collected each day, multiple days have passed by — which makes staying up to date nearly impossible. Machine learning helps marketers handle that undertaking by building an attribution model reflecting user behavior on their ecommerce sites.
Another application of ML in ecommerce is pattern recognition. Again, more data means more resources to properly understand it. But ML does the heavy lifting by efficiently crunching the relevant numbers, which means ecommerce site owners are using the most up-to-date metrics as they optimize marketing and customer experience strategies and leaving behind the approaches that just aren’t cutting it.
How can Ecommerce Brands Safely Evaluate These Technologies?
For ecommerce brands, protecting customers and their reputation means protecting themselves from anyone hawking technologies with supposed capabilities that don’t hold up to closer inspection by customers, investors or government agencies.
When it comes to assessing attribution technology that promises marketing miracles, there’s a good rule of thumb: if it sounds too good to be true, it probably is.
Ecommerce companies should approach these claims carefully by taking the following steps and asking the corresponding questions:
1. Understand the technology: How does the technology work? What data does it use? What algorithms does it rely on? What are its limitations or downsides?
2. Consider the vendor’s reputation: What is their reputation in the relevant markets? Do they have previous violations, lawsuits or credible complaints?
3. Ask for case studies: Can the vendor provide customer success stories or testimonials? What technology was used in these case studies, and how closely does it resemble the technology being sold? What results were achieved?
4. Test the technology: Can the technology be tested against the seller’s own data in a real-world scenario? Is the evaluation period going to be long enough? What kind of support will be offered during the evaluation period?
Ecommerce brands should share the concerns of the FTC, but AI and ML, properly implemented, offer marketers very substantial benefits that shouldn’t be ignored. Guidelines emerge as problems emerge, and these types of problems are to be expected in the early stages of an industry or technology’s boom. But protecting yourself against false or exaggerated claims may be as easy as doing what the FTC is doing: taking a closer look.
Phil Dubois is the CEO and Co-founder of AdAmplify, a provider of the next generation of marketing attribution software. Used by online stores, AdAmplify’s software shows which channels and campaigns are working, which aren’t, and where there’s opportunity for growth. Driven by machine learning (ML), Dimensions highlights trends, interprets results, and projects the future revenue potential of a store’s marketing channels.