What Amazon Go teaches us about opportunities in smart buildings

This week, Amazon unveiled its new Go prototype retail store, which will open to the general public in 2017. Some have said the store is more of a marketing exercise than a real business, but as an experienced product manager in smart building technologies, I think this is a viable offering. Additionally, it helps illustrate why smarter buildings and smarter systems can provide us all with significant benefits, and should be of interest to facility and building management professionals.

In case you have not heard of Amazon Go, here is a link to the video and FAQs: https://www.amazon.com/go

And, USA Today has a summary infographic and article available here that also is interesting: http://www.usatoday.com/story/tech/news/2016/12/06/amazon-go-surveillance-cameras-shopping-grocery-supermarket/95055200/

The video and USA Today article about about Amazon Go both highlight a few available and exciting smart building technologies that Amazon will utilize, including:

  • A/I and Machine Learning: Amazon probably will use your purchase history to predict what you will buy and augment any data collected in real time in the store. This will make the virtual checkout easier by eliminating confusion between items that look the same or are located near each other in the store. Combined with location sensing (see below), Amazon might know that when you walk into the yogurt aisle in the morning, you often buy Dannon strawberry fruit-on-the-bottom. When you actually pick up a container near where this is stocked, Amazon will know what you are buying. Additionally, with the sheer volume of data that Amazon will collect, it is likely that they will learn even more about buying patterns as they mine and analyze it all.
  • Pressure and weight sensors: to confirm when items are picked up from shelves or placed back on them. (These probably will work better than the sensors on mini-bottle liquor shelves in hotel rooms…) This might be more accurate than computer vision to detect what each customer is buying, especially when the store is very busy.
  • Cameras and facial identification: a patent application from Amazon indicates that it could use cameras and facial identification to track shoppers and recognize which products are being picked up. (This may be done in combination with pressure/weight sensors and indoor location/beacons.) This enables Amazon to distinguish between each shopper, creating a virtual receipt while shopping continues.
  • Indoor location sensing: This isn’t detailed in the video or USA Today article, but I would suspect that Amazon is using beacons or some other technology to track movement throughout the store. If not, they are tracking each customer’s location during each visit using video, sound, etc.

What is not discussed in the video, but potentially just as interesting, is how Amazon will be able to use this data to operate the store more efficiently than an average grocery store. Examples include:

  • Optimizing the routine product restocks. With more visibility into demand for goods and sales patterns, Amazon can order what it needs, when it needs it, to make sure the stores don’t run out of popular or high-margin products. They also will be able to schedule restocks during lulls in the store or put more of certain items on shelves to avoid restocking as frequently.
  • Bulk purchasing of goods. By tracking what is bought (and when) more closely, combined with data analysis, Amazon will be able to place bigger, longer-term orders based on data-driven sales forecasts. (Walmart has significant market power due to its huge base of stores — could Amazon challenge this by ordering for a longer time period?)
  • Optimal buying conditions. Interior comfort in buildings is hard to quantify, but Amazon likely will make strides to better understand the optimal buying conditions, which include lighting, air flow, and temperature/humidity, among others.
  • Real life “click through rates”. By tracking various steps of the purchase process (picking up a product but putting it back, picking something up and then switching to a competitor’s product), Amazon may be able to develop its own in-house branded goods more effectively, or provide this data and insight to the companies that sell goods in its stores. This data will enable Amazon to calculate real life “click through rates,” which will help them optimize the process to improve sales.
  • Location tracking. Tracking movement around the store will allow Amazon to maximize space and drive sales higher. For example, Amazon will know, in the aggregate, if shoppers actually buy their cold items last. If so, they could place milk and ice cream near the exits to reduce total time in the store (and increase the sales volume rate per minute spent shopping).
  • Selling shelf space. Amazon probably could change the model it uses to charge product companies for shelf space. It will have a data-driven view of where the best shelf space is, which will allow it to charge a premium.

Amazon has bundled these technologies to optimize the retail experience within a smart building. Their goal is to quantify the what, when and why of consumer purchases (which will make Amazon more responsive to its customers) and make it easier to buy (since easy means you do it more). Amazon mostly cares about getting more customers (both in these stores, and as overall Amazon customers) and wants to grow its share of wallet among customers.

Amazon’s goals of the Go store are different than a commercial office building manager that is integrating smart technologies, but there are some similarities. Looking at the list above of technologies in the Amazon Go store, many have value to commercial buildings:

  • By identifying individuals within the building, conditions can be optimized for each occupant. This could go beyond changing heat/humidity, air flow and lighting, though these are top of mind within the smart building space.
  • Using cameras and other technology to track indoor location, which could help utilize shared assets more efficiently (for example, conference rooms).
  • Lighting optimized for productivity: it is likely that the best lighting for a retail store is different than for an office, but the technology that optimizes for shopping could do the same for work environments.
  • On a related note, machine learning could be used to optimize the environment for meetings, getting work done during a deadline, brainstorming, etc. The more data tracked in different settings and circumstances, the easier it will be to start figuring out how to optimize for a given setting.
  • Amazon didn’t highlight loss prevention technologies in the video, but they likely also would help with general building security and fire prevention/response.
  • Technology used to price shelf space also could be applied to office leasing scenarios: which offices are most productive, which storefronts have the highest number of visits and best sales, etc.
  • Optimizing physical space is a good way to increase total rents in a office building. If an owner can get more tenants into a space without making them feel over-crowded, it will have a significant impact on the bottom line. Amazon Go also will aim to increase utilization — if each trip takes less time than an average shopping trip, Amazon will be able to get more shoppers through its doors each day.
  • Finally, there are some parts of commercial office buildings that are retail stores: in particular coffee shops and cafeterias. The technologies displayed by Amazon could be used in these parts of the building for a better checkout experience, better prediction about the specific items that will be in demand, etc. These experiences also could stretch to office supplies, reducing the likelihood of green highlighters ever being out of stock.

So, what is the ROI for such technologies?

For a grocery store, the answer may be very different than for a commercial building. First, grocery stores have notoriously small margins and are focused on optimizing their shelf space to move products. Additionally, they are selling perishable goods, so there is risk that low sales volume may lead to food spoilage (damaging the already thin margins). Amazon is using sensors and data to improve the experience and make it easier to buy, but they likely also will use the sensors to track which products sell quickly and to stock only what has been proven to move (while testing new products more quickly than other stores). They can stock just enough to drastically reduce spoilage or unwanted discounting. They also can reduce friction in the shopping process to drive sales higher.

Second, Amazon is focusing on small format stores. While grocery stores typically have high energy intensity (due to the significant refrigeration, commercial kitchens, lighting, etc) these stores probably will use less energy than a typical grocery store due to overall square footage (and certainly less than the average Class A office building, also due to square footage). The cost reduction opportunity to make the building more energy efficient may be insignificant compared to the revenue growth opportunities Amazon has by improving in the customer experience. (For more discussion on the ROI of energy savings in commercial buildings, see this article. For example, Amazon may find continuous commissioning lower on the priority list than shaving seconds off the average time it takes to shop the store (e.g., increasing the dollars sold per minute per customer is more important than reducing the facility management cost per square foot per day).

Finally, Amazon might see these stores as loss leaders, bringing more customers to the Amazon brand, which will generate more online orders and more prime memberships.

Conclusions — what can smart building practitioners learn?

Perhaps the most important lesson to be learned from Amazon Go is that smart building technologies will be more successful as they become more mission-oriented. Amazon Go’s goal is to sell more products and elevate the consumer experience, which is enabled by smart building technology. The technology is simply an enabler. The mission of most buildings is to provide a place for people to live, be successful, productive, etc… Smart building technology needs to enable these things to occur more frequently, and base the value proposition on that increase. In the past, some smart building vendors focused solely on saving energy — the goal of most buildings is not about energy consumption, so focusing a value proposition on this fact won’t scale. For example, in the case of commercial real estate, many of the Amazon Go convenience technologies likely would lead to happier tenants who renew leases, pay more per square foot, and recommend the space to others.

Also published on LinkedIn: https://www.linkedin.com/pulse/what-amazon-go-teaches-us-opportunities-smart-joseph-aamidor?published=t

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.