For a moment, Amazon Go was supposed to be the future of retail. Walk in, grab what you need, and walk out – no lines, no registers, no friction. When it debuted in 2018, some predicted it would redefine the industry.

A few short years later, the narrative flipped completely. Expansion slowed. Stores closed. And the postmortems piled up: the economics didn’t work, the tech stack was too expensive, margins were too thin, age-restricted sales were complex, surveillance and privacy concerns raised questions about customer comfort and trust, and the “automation” wasn’t as automated as it seemed.

Cue the “I told you sos” from the concept’s many naysayers.

But that conclusion misses the bigger, more important point.

 

Amazon Go didn’t fail categorically. Instead, it catalyzed one of the most important shifts retail has seen in decades.

Like AI today, Amazon Go represented a disruptive force that turned assumptions on their head, even if its ultimate impact looks different than originally imagined. In our view, the issue was never the idea. It was the premise that technology could stand on its own as a pillar of retail success. Especially in convenience retail, technology should enable the concept, not become it. Brands still need to win on one or two foundational pillars:

• Convenience
• Merchandise Authority
• Price/Promo
• Customer Service
• Experience
• Social Consciousness

Nevertheless, where Amazon Go did succeed was in forcing the rest of the industry to rethink checkout, labor, and the store itself. And that influence is still playing out across the industry today.

 

Amazon Go moved frictionless retail from “sci-fi concept” into “board-level agenda item”.

Here are five ways Amazon Go’s “success” remains relevant despite the chain’s failure to scale.

 

1. “Checkout friction” became an expensive line-item retailers finally attacked

Historically, retailers have treated checkout as a necessary inconvenience. They’ve optimized around it – adding lanes, improving POS speed, fine-tuning labor – but rarely questioned it. Friction at the register was simply part of the model.

Until Amazon showed us it didn’t have to be.

By eliminating checkout entirely, Amazon Go reframed the register as a design flaw instead of a necessary operational step. And once that idea took hold, it forced the industry to ask: What is checkout actually costing us – in time, labor, and lost sales?

The answer unlocked a wave of economically viable innovations and solutions that actually could work in different retail environments:

• Self-checkout at scale, with improved interfaces, better cash management, and stronger attendant and loss-prevention tools
• Computer vision–assisted checkout, where items can be recognized without scanning
• Scan-and-go and smart carts, giving customers control over the transaction journey and “receipt-in-hand” transparency

While only the most sophisticated convenience retailers have implemented vision-assisted checkout, and smart carts are primarily in grocery (even Amazon itself has shifted here because it learned customers want real-time spend visibility) self-checkout adoption has been dramatic.

Pre-Amazon Go, few if any of our convenience retail customers had made meaningful investments in the technology. Today, more than 90% of operators have implemented it in some form. It’s common to see 40% to 60% of eligible transactions flow through self-checkout, even for late-to-the-game retailers. And that’s the real legacy. Even though Amazon didn’t eliminate checkout, it changed it fundamentally.

 

 2. What’s possible has morphed into what’s reasonable for most retailers

Amazon Go demonstrated that ceiling-mounted cameras, shelf sensors, and a tightly integrated tech stack could deliver a seamless process, but at a cost and complexity that made broad deployment difficult, especially considering typical C-store margins. Yet by showing what could be done, Amazon Go paved the way for the rest of the market to tap into the benefits without all the heavy architecture.

Cheaper, modular, retrofit-friendly systems now exist and can deliver 70-90% of the advantages of full-store sensor fusion at a fraction of the cost. This includes computer vision-enabled checkout platforms from players like Mashgin, NCR Voyix, and Diebold Nixdorf, all of which improve throughput without requiring a full store redesign. These solutions can also accommodate realities like age-restricted items through attendant prompts. Collectively, they represent the fastest path to ROI in convenience and foodservice.

Perhaps most notably the new platforms solve for the operational realities that Amazon Go struggled to overcome. For example, Mashgin’s computer vision checkout technology supports virtually all major convenience transaction types, including age-restricted products, lottery, prepaid fuel, car washes, EBT, loyalty programs, and cash payments, while preserving a familiar checkout experience. According to the company, a single cashier supported by multiple Mashgin units can process as many transactions as three traditional cashiers.

“The goal isn’t to change how customers shop,” says Steven O’Toole, Head of Convenience and Fuel Retail Business Development at Mashgin. “It’s to remove friction while maintaining the parts of the experience that already work. Customers still walk up to a counter, interact with an associate if needed, and complete transactions in seconds.” According to Mashgin, surveys show that cashiers, store managers, and shoppers overwhelmingly prefer Mashgin to traditional checkout; the company’s highest-performing customers now process more than 90% of in-store transactions through its Power Counter product, a one-for-one replacement for a traditional register.

A second wave of companies – including AiFi, Zippin, and Trigo – opted for more flexible, retrofit-oriented approaches to push the concept of autonomous retail forward. Rather than requiring retailers to build entirely new store formats, these platforms were designed to layer frictionless experiences into existing environments, reducing both upfront investment and operational disruption. While not all of these solutions have scaled broadly or proven long-term viability, they represent an ongoing effort to make “walkout” economics work in the real world. And with continued advances in AI, that journey is far from over.

 

3. Rethinking labor allocation and throughput became the real innovation.

Amazon Go is often framed as a labor story – but not in the right way. The early narrative focused on “labor elimination.” The more durable lesson is labor redeployment.

In convenience retail, the cashier is both a bottleneck and a shrink control point. Removing that role doesn’t eliminate labor needs – it shifts them to higher-value activities like restocking, foodservice, customer support, compliance, and loss prevention.

That’s why “blended models” are winning right now. Retailers are combining self-checkout with selective staffing, AI-assisted item recognition, and analytics to manage shrink. The results include faster transactions and higher throughput, better visibility into baskets and totals, and management exceptions – not perfect automation.  That’s the more mature phase of the curve, the one Amazon Go forced the industry to reach faster.

 

4. The “just walk out” concept can and does make economic sense – in the right environments

If the original model struggled to scale across everyday retail, it’s because the economics were working against it. Instead of abandoning the concept, Amazon found where those economics do work – and leaned in.

In travel and venue environments, including airports, stadiums, and arenas, customers are on a clock – boarding, halftime, intermission. Speed is everything. And, queues cap revenue: retailers can only sell what they can transact within their customers’ windows. While space is constrained and labor is both expensive and variable, basket sizes tend to be higher and more impulse-driven. Add it all up, and frictionless retail becomes the key to unlocking revenues in breakout verticals where the economics and customer intent make for the perfect fit.

“Hudson Nonstop” airport stores powered by Amazon’s Just Walk Out are just one success story. Amazon has publicized that Just Walk Out has found other homes in stadiums, arenas, and pop-up stores and events across multiple countries, as well as in controlled micro markets such as workplaces, campuses, and hospitals.  In these environments, the retailers solved for throughput without requiring the full complexity of Amazon’s original model. Even many of the operational challenges – like age verification – have been addressed through solutions such as ID scanning and attendant-assisted workflows.

Just because Amazon Go didn’t become thousands of street-corner stores, the frictionless concept absolutely became a viable tool in the places where it monetizes best.

 

5. The Amazon Go “failure” gave everyone permission to iterate pragmatically.

Last but not least, the cultural impact of all this can’t be overstated. Amazon is synonymous with scale. So, when it couldn’t brute-force Amazon Go into thousands of stores, it sent a clear signal to the market: cashierless isn’t binary – it’s contextual.

It’s not about whether it “works” or “doesn’t work.”  It’s about what works, where, and why, based on store format, mission, economics, and operating realities.

That shift changed behavior. Retailers stopped chasing the vanity metric (“we have walkout tech”) and started focusing on operational outcomes: improving throughput, reducing abandoned baskets, and managing shrink within acceptable thresholds. That’s not failure. That’s progress.

 

Keep your eyes open because the next catalyst is already here.

Amazon Go didn’t fail because the idea was wrong. It struggled to scale because the cost, complexity, customer behavior, and real estate economics didn’t align for broad deployment.

But as a catalyst, it succeeded. It accelerated the shift toward faster, more flexible checkout and frictionless formats where speed drives revenue. It forced retailers to rethink throughput, labor, and the role of technology in the store.

Like most disruptions, it didn’t play out exactly as expected. But that’s a great reminder to stay engaged and experiment with whatever breakthrough is next, whether that’s AI or something else. The question isn’t whether innovation fully delivers. It’s how it reshapes the landscape. And, more importantly, what value will you take away?

 

Let’s Get Started