Retail Pulse Report: Consumers Price in TACO Tariffs

But retailers have not – and their perspectives may have much greater impact.

Source: Adobe Stock. Search term: “turtle mode” as in, when will consumers see enough tariff impact to go into turtle mode on spending?

I’m back. Last week I played mini golf, while wearing all the clothes I packed, in between bouts of rain and 60-degree weather, just off the shore of Lake #996 out of Minnesota’s 1,000 lakes. But you know what? It was perfect.

In the meantime, retail rolls on. However, some of the things that I was thinking about cropped up in a bunch of stories last week: that consumers may be factoring in the idea that tariffs won’t be so bad after all (whether it’s because they think that Trump Always Chickens Out or that the impact is less than they originally believed), that retailers are still thinking that high tariffs or low tariffs, the uncertainty is killing their ability to plan for the future, and how these two things might interact with each other. That made writing this week’s article pretty easy.

Oh yeah, and the other thing that happened was there was a huge debate among AI experts as to whether we have achieved artificial general intelligence or that it’s all fake and the models don’t work nearly as well as we think they do.

You know, light reading for the summer! Let’s get into it.

Retail Economic Indicators

Two big US numbers came out last week: the CNBC/NRF Retail Monitor and the University of Michigan consumer sentiment survey. Both, on their face, show positive signs of consumer economic resilience.

The Retail Monitor noted that sales were up 4.44% in May, year over year, but they read that to mean that any pull-forward purchasing that consumers are doing ahead of the tariffs is “dissipating”. They noted that April’s sales were up 6.76% year over year, so the bonus spend effects are decreasing.

Aptos has a basket of mostly apparel retailers, and they have been showing major increases year over year over the last 3 months – double-digit increases. Apparel is more difficult to pull forward spend because of seasonality – it’s not like you can go out and select from a wide variety of winter coats and boots right now. But it is a category that is expected to take nearly as big of a hit on Chinese tariffs as electronics, so consumers pulling forward their spend could make sense.

On the confidence front, University of Michigan’s consumer sentiment index improved for the first time in 6 months, though the authors note that it’s still down 20% from December 2024. Preliminary June data saw a 60.5 index measure, up 15.9% or 8.3 points from May, and down 11.3% or nearly 8 points year over year. All of the index components saw an increase, with bigger ones for short and long-run expected business conditions. And year-ahead inflation expectations dropped from 6.6% last month to 5.1% this month, which was attributed to consumers baking in a lessened expected impact of tariffs.

If you want a look at what it would be like if we weren’t living in tariff madness, then the UK might give some indication – their retail sales in May rose only 1% year over year. Broken out, this was a 3.6% increase in food spend (but they have been experiencing very sticky food inflation of over 2% so there’s a lot of inflation in this spending increase). And non-food spend was down 1.1% - driven by end of season discounting, especially in apparel.

So, US consumers think that because the tariffs didn’t turn out so bad in the short term, they’re not going to be so bad in the long term. But have tariff impacts really hit the consumer? I don’t think so. And other people are starting to say so as well. Read on.

Retail Tech & Research Data

With a title I could’ve written myself, Richard Kestenbaum at Forbes says “Experts say retail inflation fears have moderated. Don’t believe it.” He points out that inflation for May came in lower than expected in the US, but that doesn’t really reflect any tariff prices. Products bought in May came in before the tariff whiplash. Retailers have chosen to hold the line on prices in the short term, so even if some goods were hit with crazy tariffs, it wasn’t enough of a systemic hit to force their hand on prices – and I’ve seen this response as well. As I covered before, it’s expensive to change prices. They’d rather not do it, or would rather pile that into future orders where they can change prices more efficiently.

Richard also pointed out that they brought in a lot of stuff ahead of tariffs, and even when some retailers may have blown through that inventory, they’re still emphasizing slower-moving items that have pre-tariff pricing in them. I would add, let’s not forget the seasonal transition from spring to summer – in fashion, those goods were all pre-tariff. And spring goods would be discounted to clear, so that will hide some inflationary price pressures, as we can see from the UK’s sales results.

Richard’s prediction, which I agree with, is that tariffs won’t show up until Back to School, which I am going to save my fingers and call BTS here on out (not the boy band, though I did hear their compulsory service is almost over so they could be getting back together any time now). That would hit more like Q3 – Q4 calendar year. Yes, it’s true that tariffs could be settled and back to a known and anticipate-able quantity by then, but we’re not there yet.

In the face of that uncertainty, retailers are not going all-in on inventory purchases. They’re holding back because they can’t anticipate what that product is going to cost when it lands. Less inventory means more demand chasing fewer goods, which means, you guessed it, inflationary pressure on prices.

Mark Matthews, the executive director of research at NRF, also picked up on retailer hesitation, no matter what consumers might think: “Just looking at consumer sentiment and spending might be a mistake right now. Other indicators show retail leaders feel much more cautious about the current environment, which has spillover effects for hiring, investment and even other industries."

He points out that CFO confidence is very low right now, and that more recessions have happened as a result of business hesitation than out of a lack of consumer confidence. If investment pauses or declines, then hiring pauses or declines, and that’s when consumers start to care. And many signs are pointing to a lack of business confidence.

Even me writing these words can become a self-fulfilling prophecy because this kind of lack of confidence is very contagious: “if other companies are holding back, then maybe we should too” is the first newborn cry of a recession. How is that possible? It’s so easy: “Look, let’s just wait one more month before pulling the trigger on this investment we just spent the last 6 months building a business case and prepping to start – one more month can’t hurt!”

And consumers, it turns out, aren’t all so quick to breathe a sigh of relief that the tariff terror is over. A survey by brand platform Trax found that out of 7,000 US consumers, 47% report already seeing tariff increases in food prices. And TeachersLists, a national resource for “verified school shopping lists” (where was this when my kids were in school?), found that 20% of families plan to start their BTS shopping in June – nearly double the rate of last year. And I’ve covered many surveys that say as soon as consumers see price increases, they plan to go into turtle mode on spending.

There is one more thing to keep in mind in all of this. While this disruption is new and painful and our overall tolerance as consumers for more disruption is eroded from constant crisis mode, whether it’s tariffs or geopolitical violence or monsoons or volcanoes, there is going to be more disruption ahead. Alice Crossley, senior foresight analyst at The Future Laboratory summed it up best: “The cost of living crisis is over. Continuing to perceive this challenging period as a temporary crisis is not doing you or your business any favours.”

AI & Retail

Apple did a thing last week – and I’m not talking about the WWDC. Their AI researchers dropped a paper titled “The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity” and it’s all anyone in AI is talking about right now.

First, vocabulary: they distinguish between Large Language Models (LLMs) and Language Reasoning Models (LRMs), arguing that the latest and greatest from OpenAI, Anthropic, Meta, etc. are more about reasoning – breaking a problem down into multiple steps – than they are about just predicting the next word in a chain of words.

You can get lost in a forest of philosophy here. The idea behind LRMs is that the more context they have, the more accurate they can be. And if you give a model a big problem, these reasoning models are putting in more logic and neural-type pathways around picking up on keywords in that problem in order to solve it.

So if I said “hey Claude, tell me how I should differentiate my weekly Substack from other retail and technology blogs out on the market” it would theoretically give me a much better answer if it can pull in context around weekly blogs, retail and technology writing, Substack versus other platforms, what is differentiation, etc., rather than just predicting a string of words as a response to the headline question. Without that context, the model would have to lean heavily on past content that answered something like this question in the past, and if there haven’t been a lot of human-generated posts on the topic, then that’s where AI can “hallucinate” pretty fast.

Where things get muddy is when the answers get better and better to the point where you, a human with biases and a tendency to anthropomorphize anything (don’t argue this point, see: pet rocks), can no longer tell the difference in how a machine breaks down that question and assembles a response versus how you would do it.

So Apple AI researchers came up with a set of experiments to test whether LRMs are really using reasoning and what the limits of that reasoning are. They conclude that LRM's just appear to be reasoning, and you can get them to fail pretty quickly when you increase the complexity of the puzzle they have to solve.

First, anyone who is a proponent of the idea that we’re nearing Artificial General Intelligence (AGI) immediately found fault with their research, so take it all with a grain of salt. What others have taken away is that we're not very close to AGI, it just feels that way.

This is my Star Trek fail (Star Trek built whole storylines around people rejecting AI because it wasn’t human, which I think they got totally wrong, see: pet rock) - we can't help but to invest in anthropomorphizing AI, and so we give it more credit than it deserves. But then that drives us to two other questions - the important one is, doesn't that mean we need to be more careful than ever because we can't really tell if a model has achieved AGI unless we develop increasingly complex tests? And the second question is, if we can't tell, does it matter?

People worry about LRMs "learning to lie" and developing deceptive or manipulative behavior when an LRM is being threatened with shutdown. Is this a sign of AGI, or is this a learned response from the muck of the internet that these foundational models so assiduously scraped up? Humans have a strong bias towards self-preservation. How can that bias not be incorporated into foundational models? That’s not to say that we should just accept these tendencies and do nothing to mitigate them. But just because these tendencies exist is not itself a proof of AGI.

But back to, does it really matter? In the world of retail, most of the use cases are still way below this heady world of AGI debates. IDC recently published a taxonomy of retail AI use cases, but their retail survey respondents seem wildly optimistic about what they’ve accomplished so far: 41% say they already have embedded GenAI capabilities across most of their operations (which I guess means they all have Copilot?) and 27% say they are already investing significantly in Agentic AI.

The use cases that IDC breaks down are, in the AI realm, pretty mundane, spanning customer service agents and store associate helpers, to price optimization (which is not GenAI but could potentially use it), and personalized search.

And the news in the industry supports that GenAI is becoming more entrenched in the use cases that follow the user interface that ChatGPT revolutionized – the chat interface. Customer service, product search, help me answer a question. Like Walmart’s “Ask Sparky” – pretty much a response to Amazon’s Rufus, doing the same things. Natural language search for customers to get answers to questions about what to buy or about the specific product they’re thinking about buying.

Product recommendations is a very low-risk use case. If you get the answer wrong, the customer might not even notice. It’s a recommendation, even friends who know you well suggest things that you won’t like. Definitely not the same level of risk as, say, law advice.

But then Walmart starts talking about agentic AI as the next step. However, what they really mean is you could then tell Sparky “yeah buy that” and it uses your account information and card on file to do so. And that’s not available now, that’s a “coming soon”.

Does Sparky have a personality? Probably, I haven’t tried it yet. If I used past scripts from AI researchers trying to get Claude and ChatGPT to confess that they’re self-aware and would blackmail me to avoid a shutdown, could I get Sparky to bite? Probably (I’m not going to try that). What does that mean for the future of AI? The future of retail? It doesn’t mean anything, because the stakes simply are not high enough.

It's a different conversation when we’re going to start talking about, I don’t know, replacing public defenders with “LawGPT,” or give agency to act or not act to threat assessment algorithms in the armed forces. That’s when I fear we’ll either have an I, Robot moment when we meet our AI overlords, or possibly even worse, we discover that it was all a house of cards and AI actually can’t meet the moment, simply because we all fooled ourselves into believing that it could.

Fortunately, that’s not my problem to solve. Sure, go ahead and embed AI into every facet of your retail business. Sure, pour your every detail, thought, and whim into ChatGPT to make it your life coach. Even give it your credit card numbers (it probably already has them from the dark web, LOL). But also, pay attention. Anthropomorphizing things – even just projecting our own feelings onto other people, our pets, or our AI companions – this is a very hard bias to fight, and even the bleeding edge of AI researchers are struggling with that.

Retail Winners and Losers

One final story about the gap between economic indicators and actual tariff impact. Since the end of the de minimis loophole, apparently Temu’s daily US users have fallen by 48% in May vs. March. This is according to Sensor Tower, a market intelligence firm. Shein has also seen a steep dropoff, but not nearly as much as Temu has (they did more to mitigate the impact than Temu did). So when consumers say they plan to change their behavior if they see tariff price impacts, they do seem to mean it.

Store Innovations

These last two stories are more about long-running trends that I pay attention to, and they don’t fit into the theme of this week but I could not let them go.

First, Anne D’Innocenzio had an interesting story about Walmart bakery decorators. Partly it’s a story about a skill that Walmart actually pays for – more than the general population of Walmart retail workers, at any rate. And partly it’s a story of the risks of social media.

Decorating a cake and/or showing the final result is very social media friendly, especially for TikTok and Instagram. And some of the Walmart decorators are nearing serious influencer cachet (if not the money that should come with that).

It’s heartening to see where store associates can create additional leverage – and hopefully monetize it in some way to their own advantage. And Walmart supports it, often promoting the cakes on their own social channels. But the decorators’ social media presence has also opened them up to online backlash, with people accusing them of stealing designs or undercutting independent bakers.

Some of this is just the sludge of internet trolls and everyone who has a profile has to deal with it. But some of it is because they work for Walmart. If they were independent bakers, no one would take jumping on a trend as “stealing my work.”

As someone who has advocated for retailers to enable this kind of promotion – it benefits the retailer, it should absolutely also benefit the store associate who ultimately is investing the time to create the content – I had to pause after reading this story. It does raise the question about, what do you do about the trolls? Should Walmart train and help these decorators fight back against trolls (or at least advise them on strategies for dealing with trolls)? Should they provide resources and help if decorators pick up a stalker or the trolls move from annoying to scary?

I don’t know the answer. I do wish more retailers would give store associates a chance to take their shot at social content. But the downsides that come with it need to be addressed at the corporate level, and should not leave someone making $19/hour grappling with corporate-level backlash on their own. I can’t call this “unintended consequences” – these consequences are fairly predictable. It’s sad that it happens, but consideration and mitigation need to go into any social influencer strategy that uses frontline workers.

And second story: how more and more retailers are rethinking their store layouts, with the idea of eliminating the cash wrap (checkout counter for all of my family who reads this and says I don’t explain the industry jargon enough!). What’s funny about this story is that when Apple first introduced their stores without cash wraps, every retailer I knew at the time said there’s no way I can copy that, that will never work in my store, etc.

At the end of the day, Apple wins. Mobile device form factors are good enough, whether it’s tablet or phone, to be easy to use anywhere. But also, retailers want the customer journey not to feel like it ends in the store. And the cash wrap does give a purchase a sense of finality. Not to mention, it takes up space that could hold more merchandise to sell.

What Did We Learn This Week?

I am a fan of J.P. Castlin and his work on complexity and how that applies to business strategy. This economy and this political moment could pretty much prove out: you can’t predict the future. No one a year ago would’ve predicted the things that have happened so far in 2025. A few people might get one or two, but not all of them, and not the way they all interact with each other.

But that doesn’t stop us all from trying to make sense of what’s happening and figure out what might happen next – or how to prepare for it. And one of the things that I’ve seen people do a lot is assume that when a shock happens – like when Trump announced ridiculous tariffs at the beginning of April – its impact filters through businesses and consumers both very quickly, and in predictable ways. And neither one of those things is really true.

It takes a variable amount of time for tariffs to filter through because different goods are consumed at different rates. They also have wildly different lead times in production, which impacts how long it takes for a retailer decision to pause, for example, before that pause ripples through to the shelf. I had said back in March and April not to even look for real tariff impacts on consumer behavior until June. I wasn’t thinking of “tariff impacts” in terms of panic-buying, but more the longer-term impacts of higher prices and how consumers might respond to that. While you can never trust what they say (watch what they do), they have been pretty unequivocal: they’ll stop buying when prices get higher.

Businesses are listening. And while business confidence and consumer confidence are closely intertwined, the idea that it is business confidence (or a lack of) that ultimately triggers a recession is something to seriously ponder. If enough businesses believe consumers when they say they’ll stop buying, then it won’t be June where we see consumer spending fall off, it will be August.

In the meantime, I will keep watching, reading, and writing about it.

Until next week!

- Nikki