Airport Retail Media Has a Secret Weapon: Context Beats Cookies

Picture yourself sprinting through an airport, phone at 12%, boarding in 25 minutes, and wondering whether that sad sandwich counts as lunch. You are not exactly in the mood for a sophisticated, personalized brand journey.

But you might be extremely receptive to an ad for a charger, a bottle of water, or something that makes the next three hours more bearable.

That’s the fascinating thing about airport retail media: the best targeting signal may not be who you are. It may be where you are, what you’re doing, and what you need right now.

In a new Media in Retail conversation, I spoke with Katie Blackburn, Vice President of Commercial Strategy & Operations at WHSmith North America, and Sean Crawford, Managing Director at SMG North America, about what airport retail can teach the rest of the industry. Their answers raise a bigger question: have we become so obsessed with identifying shoppers that we’ve underestimated the value of understanding the moment?

The airport shopper isn’t the everyday shopper

Most retail media networks are built around some combination of purchase history, loyalty data, product categories, and audiences. That’s useful. But airport shopping happens in a strange little universe where routines disappear.

You might be a committed bargain hunter at home and suddenly pay a premium for headphones because you left yours on the kitchen counter. Or you might be the person who never buys bottled water, except when the security checkpoint has just confiscated yours.

Blackburn and Crawford describe a retail environment where travelers are unusually open to trying new brands. The episode cites roughly 65% openness to new brands, but that doesn’t mean every advertiser gets a free pass. Relevance starts with the traveler’s immediate mission. Essentials come first.

Contextual targeting deserves another look

Here’s where the strategy gets especially interesting. WHSmith can use signals like terminal, time of day, and season to make its in-store media useful without requiring detailed personal profiles.

Think about the difference between a morning commuter flight and a late-night international departure. Same airport, very different needs. A screen above a cooler can communicate something genuinely helpful at the exact moment a shopper is considering a drink.

We’ve spent years treating first-party data as retail media’s ultimate competitive advantage. And yes, it can be powerful. But context is data, too. Sometimes it’s more actionable than knowing which cereal someone purchased three Tuesdays ago.

This isn’t an argument against personalization. It’s an argument for not confusing personalization with relevance.

Retail media still has to prove the sale

The most compelling part of the conversation wasn’t theoretical. WHSmith and SMG discussed comparing stores with and without campaigns to evaluate outcomes. In one Gatorade hydration campaign across more than 80 stores, they reported a 22% sales lift.

That’s an encouraging result, although one campaign doesn’t establish a universal benchmark, and every advertiser should still ask how the comparison was designed and what other factors could explain the difference.

Still, I love the direction of travel here. Instead of telling brands that airport foot traffic is a premium audience and hoping that ends the conversation, the network is tying media to what happens at the shelf.

The lesson goes well beyond airports

What if grocery stores thought more about dinner-at-6-p.m. intent? What if convenience stores optimized for the specific missions shoppers bring through the door? What if retailers asked not merely who is this customer? but what problem are they trying to solve in the next five minutes?

Airport retail media is a useful reminder that the physical world has something digital advertising often struggles to replicate: unusually rich context.

The winners won’t necessarily be the retailers with the most elaborate identity graphs. They may be the ones that understand the customer’s immediate need, show up helpfully, and can demonstrate that doing so changed the outcome.

Listen to my full conversation with Katie Blackburn and Sean Crawford: WHSmith’s Retail Media Playbook for the Airport Shopper (November 16, 2026).

AI Agents Are Breaking SaaS Pricing. The Real Question Is What We Should Pay For.

For years, software companies had a wonderfully convenient business model: count the people using the product, charge for each seat, and call it a day.

Then AI agents showed up. And suddenly, one “user” might do the work of an entire department. So what exactly are we charging for now?

That was the question I brought to the stage at Web Summit Lisbon 2026, where I joined Joshua Wöhle, co-founder and CEO of Mindstone, and Edith Harbaugh, co-founder and CEO of LaunchDarkly, for a conversation about SaaS pricing in the age of AI agents. Web Summit published its recap on November 20, and the debate goes well beyond software invoices.

The seat is no longer the unit of value

Traditional per-seat SaaS pricing assumes that more human users roughly means more value. That assumption gets pretty wobbly when an AI agent can execute thousands of tasks without ever logging in like a person.

Harbaugh described how LaunchDarkly moved away from developer-seat pricing long ago, instead tying its model to service connections and the activity of customers’ end users. That matters because it aligns the bill more closely with the product’s role in the business, regardless of whether a human or an agent initiates the interaction.

Wöhle pushed the conversation toward value capture: if the software creates substantially more value than it costs, the model can work even when the customer has fewer traditional users.

Sounds reasonable. But there’s a catch.

Usage is easy to count. Outcomes are harder.

AI vendors can measure tokens, API calls, requests and completed steps. What they can’t always measure is whether the customer actually got what they wanted.

I raised the problem of paying for an AI system that burns resources without finishing the job. It’s a bit like a contractor billing you for every swing of the hammer, including the ones that miss the nail.

That distinction matters enormously. An AI agent that generates ten drafts but never produces an acceptable deliverable has created activity, not necessarily value. And when businesses start paying for agents at scale, the difference between the two becomes a CFO-level issue.

Wöhle made a useful counterpoint: early experimentation needs room to breathe, so usage can be a practical starting metric before businesses have enough evidence to measure ROI. I agree there’s a progression here. But the destination should be clearer business outcomes, not simply more machine activity.

The overlooked product is trust

The panel also explored an increasingly popular idea: if AI makes software easier to build, will software itself become nearly worthless?

I’m skeptical of the extreme version. Writing code and operating dependable software are two very different things. Harbaugh highlighted the ongoing importance of security, reliability and controlling AI agents that drift or fail without announcing it.

And that is where I think the next generation of software pricing gets interesting. Businesses may pay less for access to a tool and more for the assurance that the tool actually works, stays within guardrails and produces an outcome worth paying for.

In other words, the product isn’t just the software. It’s the accountability around the software.

What SaaS leaders should rethink now

Instead of asking how many seats an AI-powered product can sell, vendors should ask three tougher questions: What measurable result does the customer receive? Who absorbs the cost when an agent fails or retries? And what controls give the buyer confidence to let the system act?

The next SaaS pricing model probably won’t be one-size-fits-all. Expect hybrids that combine predictable access fees, usage allowances and outcome-based components. But the guiding principle should be simple: charge for value customers recognize, not activity machines generate.

Because when an AI agent can do the work of ten people, charging for ten imaginary seats is not a business model. It’s nostalgia with an invoice.

Read Web Summit’s recap of our discussion. The organizer notes that its recap was AI-generated from a transcript, so this article reflects my interpretation of the discussion rather than a verbatim transcript.

Did Silicon Valley Break Science? What AI, Publishing, and Trust Have in Common

Silicon Valley has spent decades selling us a pretty irresistible proposition: make everything faster, cheaper, and more accessible. Remove the middlemen. Eliminate friction. Scale like crazy.

And sometimes, that works beautifully. But here’s a less comfortable question: What happens when the friction we’re removing is actually doing something important?

That’s the question at the heart of my latest FUTUREPROOF. conversation with Kent Anderson and Joy Moore, veteran scientific publishing executives and authors of How the Internet Disrupted Science. We explored how the internet transformed scientific publishing, why artificial intelligence complicates an already fragile system, and what business leaders can learn from the unintended consequences.

Listen to the November 17 FUTUREPROOF. episode: Did Silicon Valley Break Science?

Open access changed who pays, not whether someone pays

Making scientific research available beyond expensive journal subscriptions sounds like an obvious win. More people can read research. More ideas can circulate. Who’s against that?

But access and incentives are two different things.

As Kent and Joy explain, some scientific publishing models shifted the bill from readers to researchers. Instead of making money primarily from subscriptions, publishers can collect fees when articles are accepted for publication. That’s not an indictment of every open-access journal. Plenty do serious, valuable work. It does, however, create a potential conflict when publishing more material also means collecting more revenue.

Suddenly, the metric that looks like progress, more published papers, may not tell us much about the metric that actually matters: trustworthy knowledge.

AI makes content abundant. Verification becomes scarce.

Now introduce generative AI into the equation. Producing plausible-looking text is cheaper than ever. That’s useful for legitimate researchers, but it can also make low-quality submissions and fraudulent research easier to manufacture at scale.

And there’s a deliciously frustrating twist: AI may also help identify anomalies, flag suspicious submissions, and support the people responsible for maintaining research integrity.

In other words, the same technology that can flood a system with questionable material might help us defend that system. The real issue isn’t simply whether AI is good or bad for science. It’s whether our processes, incentives, and oversight can keep up.

The lesson for CEOs: Don’t optimize away trust

Here’s where this conversation moves beyond academic journals.

Every executive has some version of the same dashboard: output, speed, cost per unit, throughput. They’re useful measures. They’re also dangerously easy to mistake for outcomes.

If your customer support team closes twice as many tickets but solves fewer problems, did you improve service? If your content operation publishes five times as much but readers trust it less, did you build a stronger brand? If your AI agents complete tasks faster but nobody checks whether those tasks were worth doing, what exactly did you gain?

Scientific publishing offers a particularly stark illustration of the problem because credibility is the product. Peer review and editorial scrutiny aren’t merely annoying steps between creation and distribution. Done well, they’re part of what gives the final work value.

Of course, gatekeepers can be slow, exclusionary, and wrong. The answer isn’t to romanticize every old institution. It’s to distinguish unnecessary friction from necessary safeguards before tearing everything down.

The next competitive advantage might be knowing when to slow down

As AI makes generating information almost effortless, I suspect the premium will increasingly shift toward evaluating, contextualizing, and verifying it. More output won’t automatically mean more insight.

That’s a challenge for science, journalism, marketing, and pretty much any business currently racing to automate its workflows.

Disruption is easy to celebrate when you count what gets produced. The harder, more important question is what gets lost along the way.

And sometimes the most futureproof thing you can do is protect the step everyone else is rushing to eliminate.

AI Shopping Can Win the Recommendation and Still Lose the Sale

We have spent years teaching marketers to reach shoppers at precisely the right moment. Now AI is threatening to change what a “moment” even looks like.

In the November 13 episode of Retail Roundup, I was joined by Sam White of LiveRamp and Diana Leza Sheehan of PDG Insights to explore how AI is changing the way brands reach and understand consumers. The episode also features a separate conversation led by Ricardo Belmar with Bank of America about the cost of a declined sale. Together, these topics raise a question that I think deserves more attention: what happens when the shopping journey gets smarter, but the systems supporting it still have very human-sized blind spots?

AI shopping is changing where influence happens

For a long time, digital marketing operated on a familiar assumption: consumers would search, browse, compare and eventually click. Brands could compete for visibility at each stage. Retail media networks, search ads and measurement platforms were built around that sequence.

But AI shopping assistants can compress those stages. A consumer might ask for the best option within a budget and get a shortlist before ever visiting a retailer’s website. The assistant can influence which products make the cut, what evidence matters and which alternatives never get considered.

That doesn’t make branding obsolete. Quite the opposite. It raises the stakes for product information, credibility, availability and the signals that make a recommendation trustworthy. The marketer’s challenge is no longer simply to win a click. It’s to be considered when an algorithm helps decide what is worth clicking on in the first place.

Better targeting doesn’t automatically mean better understanding

There’s a temptation to assume that more sophisticated AI will magically solve marketing’s oldest problem: understanding what customers actually want. I wouldn’t bet the quarterly budget on that.

Data can be incomplete. Intent changes. A shopper buying a gift looks very different from the same shopper buying for themselves. And the cleanest audience segment in a dashboard can still be a poor substitute for understanding the real person on the other side of the transaction.

For brands, this means AI-driven marketing needs a more demanding scorecard. Are recommendations relevant? Do they build trust? Are we measuring incremental outcomes, or simply giving new technology credit for purchases that would have happened anyway?

The payment failure is part of the customer experience

Then comes the part of the journey that tends to get less conference-stage glamour: actually completing the purchase.

A declined transaction can turn a beautifully personalized experience into a frustrating dead end. That’s why the episode’s conversation about declined sales belongs in the same discussion as AI discovery. Customer experience is not a relay race where marketing hands the baton to payments and declares victory. To the shopper, it’s one experience.

As AI takes on more of the shopping process, that continuity will matter even more. Retailers and brands need to think about discovery, identity, authorization, payment and fulfillment as connected parts of the same promise.

The bigger question for retail leaders

My takeaway is that the next era of commerce won’t be won by the company with the flashiest assistant or the cleverest ad placement. It will favor businesses that connect better recommendations to reliable execution.

AI may change how shoppers find you. It won’t make them forgive a broken checkout.

Find the November 13, 2026 episode of Retail Roundup, “Alexa Ads, AI Shoppers, and the Cost of a Declined Sale,” here.

AI Shopping Has a Front Door Problem. Retail Stores Have a Follow-Through Problem.

Everyone wants to own the beginning of the shopping journey. Amazon wants it. Google wants it. AI assistants want it. Payment networks want a piece of it, too.

But here’s the question I keep coming back to: What good is winning the customer’s first interaction if you can’t deliver on the last one?

That’s the tension at the heart of our November 6 Retail Roundup conversation, “Who Owns the Front Door, and Can Your Stores Deliver?” I was joined by Martin Bailie of MWB Advisory and Ricardo Belmar of Retail Razor to unpack the race to control product discovery, AI shopping and the infrastructure behind the transaction. Later, Salem Collins of DecisionPoint Technologies brought us back to the decidedly less glamorous reality of store execution.

AI shopping is becoming a battle for the first decision

For years, retailers competed for traffic: search rankings, app downloads, email opens and the occasional customer who actually remembered a password. Now an AI assistant may become the place where shoppers describe a need, compare options and decide what to buy.

That changes the economics of attention. If a shopper asks an assistant what to purchase, the retailer might not get to frame the choices. Its website could become a fulfillment destination rather than the starting point of discovery.

Amazon’s Alexa shopping ambitions and the expansion of agent-friendly payment infrastructure illustrate how many companies see opportunity in that shift. The strategic prize isn’t simply a better chatbot. It’s influence over which products are considered, which merchants are selected and which payment rails carry the transaction.

And let’s not pretend every participant wants the same thing. A retailer wants the customer relationship. A marketplace wants transactions and loyalty. A payment network wants to be indispensable regardless of which interface wins.

The store still has to keep the promise

Here’s where the conversation gets interesting. Imagine an AI assistant finds exactly the right product, at the right price, at a store five minutes away. Fantastic. Now imagine the inventory count is wrong, the associate can’t locate the item or the pickup workflow falls apart.

Congratulations. We’ve used sophisticated AI to create a remarkably efficient disappointment.

That’s why the second half of our discussion, with Salem Collins, matters just as much as the race for the digital front door. Store technology pilots can look spectacular in controlled settings. Scaling them across locations with different staffing, systems and operational constraints is another story.

Retailers don’t necessarily need another dazzling proof of concept. They need reliable inventory data, workable integrations, clear employee processes and technology that survives an ordinary Tuesday afternoon.

Winning retail AI means connecting discovery to delivery

My takeaway is that the industry is at risk of treating two connected challenges as separate conversations. The AI crowd talks about agents, interfaces and intent. The operations crowd talks about inventory accuracy, store systems and fulfillment. Customers, inconveniently, experience all of it as one purchase.

The next competitive advantage won’t belong automatically to whoever builds the cleverest shopping assistant. It may belong to the retailer that makes the assistant’s promise consistently true.

So yes, fight for the front door. But before you spend a fortune acquiring the next AI-mediated customer, make sure the product is actually on the shelf.

Listen: Retail Roundup: Who Owns the Front Door, and Can Your Stores Deliver? (November 6, 2026).

Retail AI Is Raising the Bar for Frontline Workers. Will Their Pay Follow?

Retailers have spent the past two years asking how artificial intelligence will change the store. Here’s a question I don’t think we’re asking nearly enough: if AI makes frontline employees more capable, who gets the financial upside?

I kept coming back to that question while writing my October 29 recap of Rethink Retail’s AiR Virtual Summit, The Agentic Store. The event was packed with practical examples of AI helping associates do their jobs. But one exchange about wages and skills exposed a tension that deserves more attention than the latest shopping-agent demo.

AI is making retail associates more capable

Consider Lowe’s. Its Mylow Companion helps store associates answer questions outside their own areas of expertise. Someone who knows plumbing can get useful information about a plant. A shopper doesn’t have to wait for the one employee who knows a particular aisle.

At Boot Barn, an associate-facing assistant called Cassidy gave staff easier access to product information and inventory. Tractor Supply described an approach that keeps the greeting and human relationship in associates’ hands while using technology to support recommendations and repetitive tasks.

These are meaningful improvements. They can reduce the time spent hunting for answers and help less-experienced employees serve customers with more confidence.

But notice what they also do: they change what employers expect from an employee.

The new baseline may be a much higher bar

During the summit, Richard Kestenbaum of Triangle Capital raised the labor question directly. If AI closes product-knowledge gaps and makes employees more productive, does that change hiring, compensation, or staffing levels?

Matt Ezyk, now at Maxwood Furniture, argued that repetitive tasks become less valuable while human judgment becomes more important. Associates increasingly need to recognize when an AI-generated answer is wrong, handle unusual customer problems, and make calls that a system cannot.

Then came the uncomfortable part. The discussion suggested that stronger judgment could become a baseline expectation rather than a reason for higher pay.

That’s possible in the short term. But it’s hardly an iron law of economics. If every retailer suddenly needs employees who can navigate complex situations, verify AI output, and deliver excellent service, those abilities may become more valuable in the labor market, not less.

Productivity isn’t automatically shared

We talk about AI productivity as though everyone benefits by default. They don’t. A retailer might reinvest savings in training, improve customer service, raise compensation, reduce hours, or simply protect margins. Those are management choices, not technological inevitabilities.

And this isn’t just a fairness debate. It is a competitive one. A company that asks associates to exercise more judgment without giving them time, training, or incentives may discover that its shiny AI deployment creates new kinds of frustration.

On the other hand, retailers that make their best employees even better, and give them a reason to stay, could build an advantage that competitors can’t download.

The metric I’d like to see next

Retailers understandably track AI adoption, time saved, and sales productivity. I’d add a few harder questions: Are associates more confident? Are customers getting their issues resolved the first time? Are top performers staying longer? And are the people doing more valuable work actually sharing in the value they create?

We should celebrate the useful, unglamorous AI applications reaching store floors. But the real story won’t end with which retailer deployed the best copilot.

It will be about which retailers figured out how to make AI, employees, and customers better off at the same time.

Read my full reporting: Retail’s First Real AI Customer Is the Associate, published October 29, 2026, at Rethink Retail.

Silicon Valley Promised to Disrupt Everything. Democracy Wasn’t Supposed to Be the Product.

For years, Silicon Valley’s favorite instruction was simple: move fast and break things.

It sounded exciting when the things being broken were clunky interfaces, expensive middlemen, and the assumption that you needed a taxi company to get a ride. But what happens when the same attitude gets applied to institutions that millions of people depend on?

That’s the question behind my latest FUTUREPROOF. conversation with investigative journalist Jacob Silverman, author of Gilded Rage: Elon Musk and the Radicalization of Silicon Valley. The episode, “Move Fast and Break Democracy?”, was released October 27, just after the book’s paperback edition arrived in September.

And while Elon Musk is an obvious focus of the book, the conversation is really about a much bigger question: how did the people building our digital infrastructure become such consequential political actors?

The founder mythology has escaped the startup

Tech has always loved a charismatic founder. Someone who sees what the rest of us supposedly can’t, ignores conventional wisdom, and emerges victorious while everyone else is still writing a committee memo.

Sometimes that story is deserved. Building a transformative company is genuinely hard. But success in one domain doesn’t automatically confer expertise in every other one. Being brilliant at engineering rockets or scaling a platform doesn’t make someone uniquely qualified to redesign public institutions.

Yet the founder mythology encourages us to make exactly that leap. We start treating wealth as evidence of wisdom, and business success as a mandate to govern. Those are very different things.

Platforms don’t just distribute opinions. They shape the arena.

There is nothing inherently wrong with technology executives having political views. They are citizens, too. They can advocate, donate, disagree, and participate in public life.

The more difficult issue arises when an individual can simultaneously influence political debate and control a major platform where that debate takes place.

Social networks, AI systems, and digital marketplaces are not identical to governments, but they increasingly influence which ideas people encounter, which businesses get discovered, and how public arguments unfold. Their owners have legitimate business interests. The public has legitimate questions about transparency, accountability, and concentrated power.

That tension deserves more than the usual binary of “visionary entrepreneur” versus “villainous billionaire.” Both narratives are convenient. Neither tells us enough about how the underlying systems work.

The real question isn’t whether tech should be political

Silverman’s book traces the political evolution of prominent Silicon Valley figures, including Musk, Peter Thiel, and David Sacks. Our conversation explores the industry’s ideological shifts, the incentives of social platforms, and the relationship between private technological power and public authority.

One useful distinction: criticizing concentrated power is not the same as arguing that technology companies should have no voice in policy. They absolutely should. Their expertise matters when lawmakers are considering AI, privacy, competition, and digital infrastructure.

But expertise should inform democratic decisions, not replace democratic accountability. We should be skeptical whenever someone suggests that being good at building a company is a sufficient qualification to rewrite the rules for everyone else.

Who gets to decide what the future looks like?

I’ve spent much of my career studying how technology changes business and society. I’m generally excited by innovation. I also think the industry does itself no favors when it treats every demand for accountability as an attack on progress.

Progress for whom? On whose terms? And with what checks on the people making the decisions?

Those aren’t anti-technology questions. They’re exactly the questions a society should ask when a relatively small number of companies wield extraordinary influence over its information systems.

Silicon Valley promised to change the world. Mission accomplished, in many ways. The next challenge is making sure the rest of us still have a meaningful say in what comes next.

Listen to my conversation with Jacob Silverman on FUTUREPROOF.

The Agentic Store Has a Scaling Problem, Not an Imagination Problem

Retail has no shortage of impressive AI demos. Ask a chatbot to recommend a jacket. Show a camera spotting an empty shelf. Watch an agent generate a replenishment task. Everybody applauds, somebody says “game changer,” and then Monday morning arrives.

Now try making that experience work across 800 stores, with different layouts, inconsistent inventory data, stretched store teams, and customers who have absolutely no interest in participating in your technology experiment.

That’s where the conversation gets interesting.

On October 28, Rethink Retail’s AiR Virtual Summit: The Agentic Store brings together retail and technology leaders to examine what AI can actually do inside physical stores. I’m listed as a participant in a breakout discussion titled “What Would It Take to Scale AI Across Your Stores?” alongside Matt Redwood of Diebold Nixdorf.

That question is more consequential than the latest AI product announcement. Because the hard part isn’t imagining the store of the future. It’s operating one.

A clever pilot isn’t a retail operating model

In a pilot, you can give an AI project extra attention, handpick a location, clean up the data, and put your most enthusiastic associates on the floor. That’s useful for learning. It is not proof that the same system will work in every location.

Scale introduces the things a demo can conveniently ignore: uneven connectivity, legacy systems, regional assortment differences, training needs, and the wonderfully unpredictable behavior of actual humans.

My rule of thumb? Before asking whether an AI tool can perform a task, ask who owns the result when it gets the task wrong. If nobody can answer that, you haven’t designed a scalable operation. You’ve designed an exciting meeting.

The associate is not an implementation detail

One of the summit’s central themes is the AI-powered frontline. That’s the right emphasis. Retail associates don’t need another dashboard that makes them spend more time looking at a screen while a customer waits for help.

They need tools that eliminate tedious work, surface reliable information, and make the next decision easier. A useful associate copilot might help locate an item, answer a product question, or flag an inventory discrepancy. But the value isn’t the AI response itself. It’s the better customer experience or the time returned to the employee.

That also means measuring adoption honestly. A tool installed in 500 stores isn’t necessarily a tool being used effectively in 500 stores.

Personalization needs a trust budget

Another summit discussion, “The Store That Knows You,” explores the line between helpful and intrusive. It’s a line retailers should take seriously.

Just because a store can infer something about a shopper doesn’t mean it should immediately act on that inference. Personalization can be valuable when it reduces friction. It becomes creepy when it surprises customers with information they didn’t expect the retailer to have.

Good agentic retail will need restraint, clear customer benefits, and an understandable way to correct mistakes. Trust is much harder to rebuild than a recommendation engine.

Measure the work, not the magic

I’d rather see a retailer show a modest reduction in out-of-stocks, fewer minutes spent on repetitive tasks, or more accurate product availability than announce a sweeping AI transformation with no operational evidence.

The agentic store won’t be won by whoever can produce the flashiest demo. It’ll be won by whoever can make the technology reliable, useful, and boring enough to work on a busy Saturday.

That’s a less glamorous headline. It’s also where the money is.

Explore the October 28 AiR Virtual Summit agenda and speakers.

The Next Retail Battleground: Who Owns the Shopping Cart?

For years, retailers have fought over where customers shop. Increasingly, the more interesting question is who controls the shopping decision before the customer ever reaches a store or website?

That question came up repeatedly in the October 23 episode of Retail Roundup, where I was joined by Suzy Davidkhanian, VP of Content at EMARKETER, and Adam Moyer, Chief Customer Officer at Planalytics. We covered Walmart’s new membership bundle, Google’s Universal Cart, the holiday sales outlook, and the role weather plays in retail demand. On the surface, that’s a pretty eclectic agenda. Underneath, it’s a conversation about who gets to understand the shopper—and who gets to act on that understanding.

Walmart isn’t just bundling benefits. It’s connecting shopping habits.

Walmart recently announced a combined Walmart+ and Sam’s Club Plus membership for $120 a year. For consumers, the immediate appeal is straightforward: more benefits for less than the two memberships previously cost together. For Walmart, though, the strategic possibilities go well beyond a more attractive subscription.

Think about the difference between seeing a customer’s weekly grocery run and understanding that same customer’s warehouse-club purchases, delivery preferences, and other shopping patterns. A more complete picture can potentially inform merchandising, personalization, and retail media. As Suzy and I discussed, the interesting question isn’t just whether the bundle attracts subscribers. It’s whether Walmart can translate those relationships into better experiences without making shoppers feel like walking data sets.

Google wants to be there before the retailer is

Then there’s Google’s Universal Cart, discussed in the episode as a way for shoppers to collect products across retailers, watch prices, and evaluate rewards across Search and Gemini. The implications are potentially enormous—even if the experience isn’t seamless yet.

Retailers have spent decades trying to become the destination. But if an AI-powered intermediary increasingly helps shoppers compare options and assemble a purchase, the destination may matter less than the product’s availability, value, and distinctiveness.

Here’s the tension: convenience for consumers can mean less control for merchants. If a platform handles more of the comparison and decision-making, retailers may have to work harder to defend margins and explain why their products deserve to win. Being easy for an AI agent to find is useful. Being meaningfully different is still better.

Holiday growth isn’t the same as healthier demand

The National Retail Federation forecasts 2026 November–December retail sales will grow 4.2% to $1.013 trillion. That’s a substantial number. But the episode also explored a critical caveat: roughly half the projected growth reflects higher prices rather than more goods being purchased.

That distinction matters. A retailer can report higher sales while moving fewer units, and a strong-looking headline can conceal pressure on households. Executives should be asking what is happening to volume, margin, and repeat purchases—not just celebrating a bigger revenue figure.

And yes, the weather belongs in your AI strategy

Adam Moyer brought a useful dose of physical-world reality to the discussion. Weather can change when people shop, what they buy, and whether a promotion succeeds. A forecast that looks smart in a spreadsheet can fall apart when temperatures or storms shift actual demand.

For retailers investing in AI-assisted forecasting, that is an important reminder: better algorithms won’t rescue incomplete inputs. If an AI system is making inventory or marketing recommendations without accounting for conditions that change shopper behavior, it may be confidently wrong.

My takeaway: The next era of retail isn’t simply about owning more channels. It’s about understanding customers across channels, protecting a differentiated proposition when intermediaries gain power, and making decisions grounded in what shoppers are actually doing. The shopping cart is becoming smarter. Retailers need to be smarter about what that means.

Listen: Google’s Universal Cart and the Weather Behind Demand — Retail Roundup (October 23, 2026).

Additional sources: Walmart’s October 20 membership announcement and NRF’s 2026 holiday sales forecast.

When Should Grocers Automate Fulfillment? The Answer Isn’t ‘As Soon As Possible’

Grocery delivery is one of those businesses that looks simple until you try to make the economics work. A customer places an order. Someone picks the items. A driver shows up. What could possibly go wrong?

Quite a lot, as it turns out. Substitutions, inventory errors, labor costs, unpredictable demand, and the occasional missing avocado all have a way of turning a promising digital channel into an operational headache.

That’s why I was interested in a recent conversation I hosted at Groceryshop with Petr Lizner and Dalibor Patek of Rohlik Group, published by Rethink Retail on October 20. We explored a question facing more U.S. grocers: When does micro-fulfillment make more sense than picking online orders directly from store shelves?

Micro-fulfillment isn’t a yes-or-no question

The temptation is to frame grocery automation as a binary choice: embrace robots or fall behind. But that’s a lousy way to make an investment decision.

In our conversation, Lizner and Patek offered useful volume markers. Around 300 online orders per day, a dedicated fulfillment area becomes worth exploring. Around 600 daily orders, automation may deserve a serious look. Those aren’t universal rules; they’re practical starting points from operators who have spent years wrestling with the problem.

And the right model depends on geography. Dense delivery markets can support different economics than sprawling regions where pickup is more attractive. A clever warehouse solution doesn’t repeal the laws of distance.

The robots are the visible part. The hard part is everything around them.

It’s easy to get excited about the machinery: automated storage, efficient picking systems, software that orchestrates the flow. It’s harder to get excited about the mundane work that determines whether the investment pays off.

Inventory accuracy. Staffing. Order sequencing. Delivery timing. The handoff between the dedicated fulfillment operation and the physical store.

Those pieces have to work together, day after day. Rohlik’s experience illustrates how much institutional learning sits behind a seemingly seamless customer experience. You can’t install operational discipline the way you install equipment.

Reliability is a growth strategy

One of the more striking contrasts in the episode involves order errors. The discussion describes some newer online grocery operations struggling with double-digit error rates, while experienced operators such as Rohlik report rates below 1%.

Think about that from the customer’s perspective. The best grocery delivery experience isn’t necessarily the one with the flashiest app or the most futuristic fulfillment center. It’s the one where the ingredients you ordered actually arrive, when you need them.

There’s a broader lesson here for retail leaders. We spend enormous energy on acquisition, personalization, and digital experiences. But the operational basics are part of the brand promise. A missing item is a customer-experience problem, not just a supply-chain metric.

My takeaway: automate the constraint, not the headline

Micro-fulfillment could be a major advantage for the right grocer at the right scale. But buying automation before fixing the underlying process can simply make an expensive problem move faster.

The question I wish more companies would ask isn’t, “What technology should we deploy?” It’s, “Where does our current operating model stop working, and what evidence would justify changing it?”

That’s less glamorous than a robot demonstration. It also sounds a lot more like a business strategy.

Listen to the conversation: Is Micro-Fulfillment the Winning Strategy in the US? (Rethink Retail, October 20, 2026).