Enter your keyword

Let’s drop ‘e’ and enable commerce via all digital avenues. Bring about a paradigm shift in the definition of channels;engage and enthral - to augment stickiness - by maintaining visibility across all possible customer touchpoints.

Thursday, 22 December 2016

Amazon Go and the evolution of Smart Store


Amazon Go is definitely one of the most exciting retail innovations in recent times. At its Beta stage, it tries to give a No Checkout – Just Walk Out shopping experience (https://www.amazon.com/b?node=16008589011).

The Smart Store concept however is not new. A German Cash and Carry had implemented a ‘Future Store’ concept in early to mid 2000s.

The evolution: RFID -> IOT
The story behind the evolution of the Smart store concept is pretty interesting - it all started with the Walmart RFID mandate in 2003. It required its suppliers to tag the case and pallets with a RFID.
The objective was to achieve a Smart Supply Chain with:
·         Near Real Time inventory visibility across the entire chain and
·         More optimized warehousing processes leading to significant cost saving for the big box retailer.



RFIDThese are Radio-Frequency Identification tags, which can carry some small amount of electronic data. This helps in uniquely identifying an asset and also store some additional information.

These RFID tags can be tracked using strategically positioned RFID Readers.

These are of two varieties – Active and Passive; the active ones having a power source of its own; while the passive ones get activated in close proximity of a RFID Reader.

Given the clout of Walmart, this mandate was supposed to bring down the price of an RFID tag closer to a few cents from a few dollars; thus making the technology affordable enough, to tag not only the cases and pallets; but also every item – thus enabling tracking of every individual asset - uniquely.

This spurred a lot of activity in definition of technologies involving RFID. The RFID based technologies involved the following:
  • ·         Reading of an RFID tag using a RFID Reader
  • ·         Real time cleansing and filtration of the relevant data to help uniquely identify a relevant asset / resource
  • ·         Usage of the data to trigger real time action as mandated by a specific business use case

·         An aggregation of these use cases made a bridge between the devices in a physical world and the virtual world. This is where the Internet Of Things (IOT) journey started.

The Internet of Things (IoT) is a system of interrelated computing devices, mechanical and digital machines, objects, animals or people that are provided with unique identifiers and the ability to transfer data over a network without requiring human-to-human or human-to-computer interaction.

Credit: http://internetofthingsagenda.techtarget.com/definition/Internet-of-Things-IoT

Apart from its applicability in making smarter supply chains; some of the earlier experiments with the technology were the following:

  • ·         Asset Tracking – Ensuring that only authorized asset movement happens across a particular checkpoint, car parking automation, pilferage prevention of costly items in a store scenario etc.


  • ·         High End Perishable Items – Some high cost perishable items like stents used in heart surgeries, runs the risk of a huge revenue drain if it perishes before usage. The supply chain and the distribution processes, for such items, were made smarter to address the issue.


  • ·         Smart Shelf – Typically the shelf had their own identity as well as the individual items stacked on them. These shelves, tracked by RFID readers, could sense any changes to its existing state. These state change information could be used

o   To define various customer friendly shopping experiences
o   For store operation automations
If the customer getting into a store can be uniquely identified as well; then a mixture of these technologies can help:
o   Track customer behaviour with respect to his purchase preferences
o   Track the movement patterns of the customers within a store; these movement patterns when aggregated with the individual / segment level customer insight, can devise a predictive approach to store designs

·         Virtual Store Assistants – The culmination of the above technologies can help create a virtual personal store assistant for the customer. Well !....here we are getting into the domain of general purpose Artificial Intelligence (AI)

·         Self Checkout – Long queues at the checkout counters has long been a concern and primary source of customer frustation. Retailers have tried different mechanisms to reduce the customer pain, like:
o   Queue Busters
§  by automatically opening a new POS as soon as the queue length goes beyond a particular threshold
§  mobile POS
o   Automated Checkouts using RFID Readers – As you push your cart into a virtual checkout zone, the RFID readers detect the items, read the price information from the tags or look it up from the database, calculates the taxes and promotions and generates the bill. The customers need to pay on his / her own and move out.
In the current Amazon Go age, if the customer saves money in a Digital wallet, the amount can be directly deducted from there.




The glitches – that needed fine tuning

While the evolution started in early 2000s, most the applications built at that time, did not see the light of the day beyond pilots. This was because of some of the below practicalities and technology glitches, that needed further fine tuning:

·         RFID Tag Collision – An RFID reader takes a small fraction of a second to read a tag. But if multiple tags come within the vicinity of the reader, within that specified fraction; then there are possibilities, that some of those tags go untracked. In certain kind of applications, like that of a shopping checkout, it can cause a drain in the overall billed amount.

·         RFID Reader Collision – Depending upon the positioning of the RFID readers, there are possibilities that the readers try to sense the same tag at the same instance of time. Over a period of time, different anti-collision protocols has come into vogue.

It might as well happen that the same tag might be sensed by multiple readers at different points of time, close to each other. While in certain cases it might be an intended functionality; at other times, a logic needs to be implement to filter out duplicate tracking signals.

·         Standardization – In the initial days, every RFID reader manufacturers had their own proprietary protocols; so every different make, needed modification to the software - at times, written at the level of serial port communication.

·         Quasi solid substances – One of the main impediment to a Smart Store solution was the low accuracy of RFID detection, when it involved semi solid materials. A shopping cart typically will contain different items containing materials of different consistencies; hence any inaccuracy will negatively impact the billed amount of the cart.

Some of these factors impeded mass adoption of RFID; hence the price of the tags did not come down to a level, where it can be used to tag each individual item. Thus the Smart Store concept needed some time to realize till the Amazon Go happened.

Amazon Go:

Some of the technologies discussed above have been fine-tuned over a period of time. Some technologies which were much ahead of time, has found applicability now; and there are some new technological innovations that have come about in recent time.

On one hand we have technologies like IOT, QR Codes, Bluetooth, NFC, motion sensing (eg., Kinect), AI etc. This clubbed with the growth of internet enabled mobile devices, creates a defining moment for the Smart Store concept.

The Amazon Go Beta program will might face some of the practical situations as below:
·         A large number of people walking into the store, mainly during promotion or Holidays – accuracy in uniquely identifying each customer
·         Heavier than normal activity at one shelf, with multiple people picking stuff from it and then putting it back; everything happening in a close proximity of time
·         100% accuracy in tracking the items in a customer cart; every decimal fall in accuracy will have its negative impact on the billed amount
·         The accuracy in tracking small items as well as the ones containing quasi solid material
·         Multiple MRPs for different SKUs belonging to the same product (this might be a very India specific scenario)


Once they figure out, either technology or operational solutions, to some of the above potential issue, Amazon Go would be ready for a mass rollout.

Pranab Das

Let’s drop ‘e’ and enable commerce by all digitally enabled means.

No comments:

Post a Comment

Popular