Just thinking today about the differences between Analytics in the video world, and in particular the business case for the additional costs. It seems there are two very broad categories of Video Analytics. First, Critical Infrastructure Security Analytics, might be considered those used to provide Asset Protection, Loss Prevention, and/ or secure an organization's infrastructure and resources. Most often these are marketed to Airports, Seaports, Government Buildings, and Corporate Offices, because this is the most likely area of an organization that can justify the higher price tag. These are the most popular in nature currently and the ones most of the industry is referring to when they speak about "Video Analytics". These might include items like:
- Setting tripwires
- Object (object might be defined as person or vehicle) left behind
- Object entering an Area Of Interest (AOI)
- Directional awareness
- Flow control
Second, there are what could be referred to as Business Intelligence Analytics. As the descriptor implies, this refers to those analytics in which provide additional business intelligence and by their nature are developed to serve that purpose primarily. Most often these are related to Retail-type of applications (Consumer Retail, Retail Banking, etc.). The analytics would enable the video system to provide items like:
- Conversion Rates - comparing and analyzing shoppers to purchases
- Directional Analysis - analyzing which direction customers move within a given Field of View (perhaps based upon in-store signage)
- Employee Performance - analyzing and comparing numbers of customers serviced by coworkers and providing performance metrics between employees
- Queue Analysis - providing automated response when queues are getting too long, or there is queue abandonment
Of course, the two broad types of analytics could overlap in certain ways. For instance, identifying an object (person) within a field of view and then identifying that object crossing a tripwire could allow an analytics to "Count" people entering a department or Area of Interest - giving the appearance of Departmental Conversion Rate.
But perhaps what is more important is how the analytics, and the overall software package included, can display a Return on Investment. Analytics to date have had a hard time justifying the higher price tags. Now, we're starting to see prices drop, but also quality, accuracy, and expectations drop also.
A brief look at typical Critical Infrastructure Analytics show the most popular, and historically the most impressive Return on Investment (ROI). Frankly, the CI analytics are touted to save lives and reduce loss. Setting up an analytics to detect a "suspicious bag left behind" in an airport can protect the patrons and employees of the airport as well as reduce the possibility of costly terminal shutdowns. Detecting a small boat approaching a larger vessel and triggering alerts can have a huge impact in protecting precious cargo and even even soldiers in battle. But the reality is in those markets, it is very difficult to show ROI as the endeavor is to stop dramatic, and potential very public in nature, events from occurring. In this case, a single "Failure" in a mission critical event such as this may not be acceptable (because it could cause loss of life and or expensive resources).
On the other hand, there is Business Intelligence Analytics. By using department/ product level conversion to identify why "X" number of customers enter an area and yet do not buy a specific product, one can make a very real and immediate change to the layout of the store to increase sales of that product (or replace the product altogether).
Also using a combination of Line Queuing Analytics, as well as, Real-Time Dashboard with Alerts and Forensics, a Store Manager could immediately affect Customer Service (by reducing wait times), and review and compare employee effectiveness.
In addition the risk of failure is insignificant in the overall plan. Missing one or two people in an Area of Interest during a 1 hour period will not cause dramatic, negative results.