Why Retail AI's ROI Remains Unproven - AI Takes Over

More than eight in ten retailers claim to have embraced artificial intelligence "to a moderate or large extent," according to a recent report. That sounds impressive, doesn't it? Like the entire industry is on the verge of a tech-driven revolution. But let's dig into what that *actually* means. Are we talking about fundamental shifts in profitability and efficiency, or just slightly smarter product descriptions? The Berkeley Research Group suggests that AI is being applied to marketing (70%), IT (62%), e-commerce (56%), and merchandising (54%). All well and good. But the same report throws a bucket of cold water on the AI party, noting that this adoption doesn't necessarily translate into "tangible business impacts." In other words, retailers are *doing* AI, but are they *benefiting* from it? That's the multi-billion-dollar question.

AI ROI: Show Me the Numbers, Not the Hype

The KPI Conundrum The report suggests measuring AI's benefits through KPIs like average order value, inventory turnover, revenue, customer retention, and labor efficiency. Seems reasonable. But here’s the thing: retailers have been tracking these metrics for decades. Are they *really* seeing a statistically significant jump *specifically attributable* to AI, or are they just riding the wave of general economic trends and marketing pushes? I remain unconvinced. Sam's Club is rolling out AI-powered "Scan & Go" at 600 stores. Levi Strauss is partnering with Microsoft for an "agentic framework." Walmart has unveiled its own AI framework with "super agents." Target is using AI to generate new product ideas. All splashy headlines, but what's the *actual* ROI? What's the delta in sales, reduced costs, or improved customer satisfaction that can be *directly* linked to these initiatives, net of all other factors? These press releases are long on buzzwords and short on hard numbers. The article from McKinsey presents a more bullish view, estimating that automation could affect up to 55% of workers' current activities by 2035. That's a broad claim. They also suggest that CPG and retail companies leading in digital and AI are showing three times greater total shareholder returns compared to their peers. Now, that's a statistic that grabs your attention. But correlation doesn't equal causation. Are these companies simply better managed overall? Did they invest in AI at the *right* time, when costs were lower and the technology was more mature? Or are they just benefiting from some unrelated market tailwind?

Automation's "Potential": A Reality Check?

A Methodological Critique Here's where I start to get skeptical. How are these "automation potential" numbers being calculated? McKinsey says they're using a proprietary automation model that assesses over 2,000 activities across 800 occupations. But what are the underlying assumptions? Are they assuming *perfect* implementation, with no glitches, no employee resistance, and no unforeseen consequences? Because in the real world, those things *always* happen. Also, the timeframe is important. McKinsey's analysis stretches out to 2030 and 2035. That's a long time in the tech world. What seems revolutionary today could be obsolete in five years. It's like trying to predict the dominant smartphone platform in 2010 based on the state of the market in 2005 (remember Blackberry?). Any long-term forecast in this space should be taken with a massive grain of salt. And this is the part of the report that I find genuinely puzzling. McKinsey notes that many executives are "overwhelmed by the volume of gen AI use cases" but haven't laid the groundwork to get started. That's a contradiction. If they're overwhelmed, shouldn't they be *more* motivated to get started? Perhaps the problem isn't the *volume* of use cases, but the *lack of clarity* on which ones are actually worth pursuing. Tariffs are also throwing a wrench into the works. As one article notes, off-price retailers initially thought they were relatively immune to import duties, since much of their inventory comes from other retailers. But it turns out that a significant portion of their *indirect* sourcing is from China – about 40% for TJX, according to one estimate. Now, TJX seems to be navigating the tariff situation reasonably well. Ross Stores, on the other hand, pulled its guidance for the year, citing "supply and demand volatility." The difference? Apparently, TJX began "aggressively moving out of China well in advance." That proactive move seems to be paying off. Burlington's CEO noted that tariffs can cause not just excess supply but also shortages, which complicates things for off-price retailers. He also pointed out that suppliers and retailers will eventually adjust, so the long-term impact may be less dramatic than the short-term disruption. Tariffs squeeze off-price retailers more than expected So, What's the Real Story? AI in retail is real, but the hype is outpacing the hyper-performance. Retailers are investing in AI, but it's not yet clear whether those investments are paying off in a meaningful way. The data is murky, the claims are often exaggerated, and the long-term impact is highly uncertain. Until I see more concrete evidence, I'm sticking with my initial assessment: proceed with caution.