How does it work?

CoolSnap is a rewards program based on consumers taking photos of eligible Coca Cola coolers. Coolers are tagged with four markers that have to be part of the photo and either a static unique id or dynamic code (via magnet or attachment to cooler).

The rewards program is part of a mobile smartphone application. Points per cooler are based on cooler demand and timing of the last photo sent. The app will include a map of coolers with point values. Photos will be compressed and sent to cloud server for processing via mobile app. Data extracted from the pictures will be used for long-term analysis by Coca Cola/bottlers or tactical decisions when developing restocking routes (and their frequencies) and deciding which products/quantities to load. Extracted data can be packaged via a secondary app for bottlers to help drive decisions.

What does it measure?

This strongly depends on the level of image processing that Coca Cola has already developed. Significant value would start from being able to identify empty rows and SKUs with non empty rows. Additional value would come from ability to determine number of products behind the first more easily visible product per non-empty row. Depending on other advancements, the images can provide more or less value for different regions/markets via the same mobile app that is used for consumer image capturing and sending to cloud server via data plan.

Does the retailer/bottler/distributor have to do anything different to make your solution work? (if yes, please explain)

  • ~10 minutes per cooler of installing markers and unique id tag
  • Use of secondary app for re-stocking decisions/demand analysis (free or pay to use as Coca Cola will own the data). Bottlers would need to "register" their coolers in the application to receive data.

What materials does your solution require?

  • Mobile app for consumers
  • Database server
  • Image Processing code
  • Mobile app for bottlers
  • Photo assistance markers for coolers
  • Unique identifier tag for cooler as can be deciphered from photo
  • Marketing campaign
  • Reward program

Best guest on cost to implement

  • Development of mobile app for consumers ($500,000-$1,000,000)
  • Maintenance of mobile app for consumers ($100,000 per year)
  • Cloud server setup and maintenance ($100,000 per year)
  • Image processing logic and calibration ($5,000,000)
  • Marketing campaign ($10,000,000 per market)
  • Installation and creation of necessary cooler markers ($5 per cooler)
  • Development of mobile app for bottlers ($500,000-$1,000,000)
  • Maintenance of mobile app for bottlers ($100,000 per year)
  • Reward program pay outs

Challenges you/your team ran into

  • Potential gaming of the reward system
  • Marketing effort to get consumers on board
  • Lack of free source image processing technology that is sophisticated enough

What you/your team learned

  • Computer vision algorithm applications such as SIFT and and object detection
  • Problems with custom shelf designs in coolers and SKU shape varieties on image recognition
  • The influence of Coca Cola’s market demographics on solution viability

Next steps

  • Determine if current image processing logic available to Coca Cola are sufficient
  • Assess chance of success with new reward program
  • Develop the phone applications
  • Test in a pilot market with marketing campaign

Anything else you want to add

The main strength of this approach is from leveraging hundreds of millions of smartphones for image capturing and sending data to a cloud server. If the idea is attractive enough now, it will only get better as smartphone use rises dramatically around the globe and image processing technology improves. In time, the app can remain as a way to send data from cooler to cloud server without need for images if the coolers know their inventory position but it is too expensive to send the data.

Share this project: