There was a lot of good news in retail economic indicators this week, particularly for the U.S. However, there are still plenty of sharks under the water that can bite consumers, and some of those sharks will lead to a very uneven pace of improvements for consumers around the world. Retailers continue to notice that consumers are trading down – and they really mean it this time! – and are taking more action to finally respond to those pressures, while the world continues to be a more dangerous (and scammy place) for retailers and consumers alike. The question about human-generated data for GenAI foundational models got a real mathematical estimate applied to it and came up with a pretty near-term answer, and quick commerce and Amazon Marketplace prove that the more things change, the more they stay the same. Let’s dive in!
Retail Economic Indicators
Unless you were living under a rock, you might have heard that U.S. inflation came in flat month over month and held steady at 3.3% year over year. This was better than expected, and good news after a few months of unexpected price rises. Shelter is still the sticky part – increasing 0.4% for the fourth consecutive month. But energy fell 2.0% over the month, and gasoline fell 3.6% – which is remarkable, going into summer driving season. Combined with the weekly jobs report (initial jobless claims) coming in higher than expected (indicating a possibly slowing labor market), the markets perceived this as good news. While the Fed is not anticipated to cut rates this summer, a rate cut later in the year appears to still be on the table, and the words “soft landing” are showing up more often again in the press.
The good news on inflation came on top of great (for retailers) spending news – the CNBC/Retail Monitor published by NRF found that after a decrease in year over year sales of 0.6% in April, May’s retail sales were up 3.03% year over year. Cores sales, excluding automobiles, gasoline, and restaurants, was up 2.48% during the first five months of the year. I have to point out that none of these numbers beat inflation, but they do come really close and are definitely a boost after anemic April.
What was most interesting In the numbers came from category-specific results – some of the most discretionary spending, like personal care and clothing and accessories all came in well over 5% growth year over year – beating inflation and then some. Big ticket items like electronics and home improvement, where financing the purchase is required more often, still drag in part due to high interest rates.
And in more good news in the U.S., wholesale prices unexpectedly fell 0.2% in May as well. Normally I don’t pay much attention to things like the Producer Price Index because it can get complicated in interpreting how that really is going to impact consumers – and there is the Consumer Price Index that directly tells us how it impacts consumers anyway. But this one caught my eye this week because we have seen retailers making a big show of cutting prices, and in the end that may be built on a more solid foundation of falling wholesale prices, which would be more lasting than just taking a promotional hit.
In other markets, the news is not quite so good. While the UK consumers have been rallying despite inflation, the weather, and now an election, it seems there is a shark under the water – a loan shark, if you will. In the UK, most mortgages reset their interest rates after 5 years, not 30. Mortgages that reset in 2023 are now a year old, and consumers may have tried to shoulder the increase so far, but it looks like they’re running out of buffer. The Bank of England reports that mortgages in arrears are up 44.5% in the first 3 months of the year vs. a year ago, and is also up 4.2% compared to the last three months of 2023. This is the highest level of recorded arrears since 2014. The really unfortunate part of this is, those resets are going to last for another five years, possibly lengthening the amount of time it takes for UK consumers to get through inflation and interest rates.
And finally, China. It’s a big consumer market – by far the largest in Asia (though India is nipping at its heels). But there are definitely headwinds in the economy. GDP growth is slowing – it averaged 9% since 1978, but in 2023 hit 5.2% and is expected to be 4.5% in 2024. Economists focused on China say consumers are losing confidence in the economy, driven by the troubles in the real estate market. While high home values in the U.S. have kept consumers spending, in China the opposite is happening – consumers see their homes losing value or are underwater vs. what they owe, and combined with slowing income growth, that has undermined their confidence in their future. Chinese consumers are saving more and spending less as a result, in what economics call “demand deflation”.
Retail Tech & Research Data
Placer.ai data reveals one reason behind retailers suddenly finding space to take price cuts: shoppers are not just trading down within stores, they’re actually trading out the stores they visit. Consumers are distinctly shifting their traffic to more discount stores, and retailers are noticing – cutting prices to recapture some of that lost traffic.
Two more research reports this week, both looking at payments. EMarketer found that the P2P payments space (think Venmo, PayPal, CashApp) will grow 14.5% by 2028, to where it will cover the exchange of $1.224 trillion and touch at least three-quarters of smartphone users by that time.
At the same time, another favored consumer form of stored value, gift cards, is getting scammier by the day. The FTC estimates that Americans lost $217 million to gift card scams in 2023, and also points out that is likely an under-reported crime. It’s hard to talk about types of gift card scams without also inspiring new scammers, but at the same time consumers need to be more aware of how gift card scams are executed. It’s getting more sophisticated than “Hi, your CEO asked me ask you to buy $500 in gift cards and send me the numbers” (though that’s still a popular play, now aided and abetted by GenAI). If anyone asks you for the barcode or the PIN on a gift card, you’re getting scammed. It’s gotten bad enough that Walmart has disabled the purchase or redemption of gift cards outside the U.S. The retailer has also instituted tracking on where a gift card was bought, where it was spent, and how much was at stake. If a gift card magically moves quickly across the U.S., then they turn it off. The FTC is asking retailers to do more to educate consumers about gift card scams at the point of purchase as a way to help stop the bleeding.
AI & Retail
Michale Kors is buying into Dynamic Yield’s Shopping Muse. Dynamic Yield is a Mastercard company and presumably benefits from Mastercard’s vast consumer purchase data in some way. Shopping Muse makes recommendations based on “contextual and behavioral insights, the retailer’s keywords, visual cues, and the consumer’s own personal intent and affinities.” The pilot showed a 15-20% higher conversion rate than traditional search queries.
I’m not surprised by this at all. As long as you’re monitoring the results closely, I expect to see a lot more of these stories – retailers getting benefits from chatbots. It’s low risk. It’s not right 100% of the time, but it’s good enough (15-20%) that it’s better than nothing. And it’s not likely to come out and recommend 1/8 cup of glue on your pizza or anything like that. But it’s also not going to be good at much else, at least not without introducing a more significant risk that it will start spouting off about glue on pizza. Chatbot. Cool. Shrug.
In the meantime, Epoch AI took a stab at truly estimating when LLM’s will run out of human-generated data. I’ve seen speculation that we’ve already run out, or at least that the data that these models hoover up from the internet is no longer purely human generated (given the number of bots out there spewing things on X, I doubt that this was true beyond maybe the first year of the internet anyway). Epoch’s estimate is more a look at the rate that humans generate data on the internet vs. the consumption these LLM’s require for the next level of improvement. Based on that run rate, they estimate it will happen somewhere between 2026 and 2032, with a closer guess of 2028. We’ve heard repeatedly from computer scientists that synthetic data is no good, so I guess we’ll know the dimensions of the hype part of the AI hype cycle in the next 2-4 years.
Retail Winners and Losers
If you think that retailers and restaurants are cutting prices out of the goodness of their hearts in their bid to help bring inflation down, this article will give you pause. Inflation has been high, but it sure has not been as high as McDonald’s prices have climbed. It takes a look at 9 pretty popular items and found that prices have gone up an average of 100% over the last decade, and in some cases substantially more than that.
If the headline was expected to shock, the outcome of this next story was actually more of a ho-hum. The British Independent Retailers Association is expected to file a GBP 1 billion ($1.3 billion) lawsuit against Amazon, claiming that the company used non-public retailer data in order to sell cheaper (competitive) products to consumers. The non-public retailer data came from its retailers’ sales on Amazon Marketplace. I haven’t been able to tell if they have actually filed the lawsuit – that would make this more important news. But Amazon has been accused of this multiple times, to very little effect. The only difference is this is coming out of the UK, so a potentially different set of laws apply. I can see how Amazon would be sorely tempted to use its data in this way. I can also see how it makes a lot more money off of Marketplace than its own retail margins, so even if it did use this data in this way in its early days, I just don’t know why it would bother now.
Finally, as I am fresh back from a trip to Bengaluru, India, I am particularly sensitive to stories about quick commercecoming out of the country. I saw my own share of Swiggy and Blinkit (among others) branded moped and motorcycle drivers out on the streets with the ubiquitous box-shaped cooler on the back. As one of my colleagues said, you can order your veggies for dinner and get them before you’ve fully heated the pan (10 minutes or less). Goldman Sachs estimates that quick commerce delivery accounted for $5 billion or 45% of India’s $11 billion online grocery market, and it’s forecast to reach $60 billion by 2030, and grow to 70% of the market.
One thing my colleagues wondered (me too) was how they really pull it off. The article reports that Swiggy had 500 warehouses across 25 cities in India in 2023, with plans to reach 750 by April 2025. However, the company seems to be mostly coasting on investment dollars, as it still does not turn a profit, even as revenue has grown to $1 billion. Contrast this experience to that of Getir, which has retrenched only to its home market of Turkey, after expanding to five countries.
At the end of the day, you can’t get past the physical logistics of getting a product from a location to an end consumer. Like China, India can easily have 25 tier-one cities for population density, and it takes us all back to the most basic of quick commerce questions: does this only work in very dense populations? Or, perhaps more pointedly, has anyone truly made a profit on the logistics side of quick commerce, or have they all just ended up burning a lot of investor money trying?
The Bottom Line
So what did we learn this week? The economic icebergs continue to jostle together, hinting that they may be breaking up more and more, to the benefit of consumers and businesses alike. There also seems to be room for retailers to cut prices without giving away the farm. Right now, they’re taking cuts across limited assortment and potentially for a limited amount of time, but as more and more egregious price rises go viral (see McDonald’s above), then there will be more pressure to give more of that back, and soon.
In the meantime, retailers will find benefit from AI. I have no doubt. It will become an arms race, especially around recommendation bots. I do think once everyone can get a 15% increase in conversion, we’ll get to the point where it’s a base expectation rather than a differentiator or even a benefit. These are low-risk things, and should not be used as a basis for calls on how “GenAI will change the world!” – whether that will happen, remains to be seen and I’m still very much a skeptic. Some things never change, and both the British consortium’s lawsuit and the quick commerce story serve as a reminder of exactly that.
