In the last few years, enterprise data has exploded, both in terms of how much of it we are creating and the many uses we are finding for it. In technology publications we mostly read about data from social media, digital technologies or wearable devices. But security is one of the most data-intensive industries and it’s ripe for advancements in analytics: you’re dealing with pressing issues in security and fraud while collecting an incredible volume and variety of data, most of which hasn’t been tapped.
For the past 5 years, Solink has focused on turning unstructured surveillance video into a relevant, meaningful and accessible source of data. In November of 2014, we were invited to present our Contextual Analytics technology at the renowned World Continuous Auditing and Reporting Symposium (WCARS) at Rutgers Business School.
Michael Cangemi is a business advisor, author, speaker and Senior Fellow of WCARS. He is the Founder & President of Cangemi Company. Between 1991 and 2004, he was President, Chief Executive Officer and Director of Aigner Group, Inc., a wholesaler of women’s footwear, accessories and licensed products.
What you will hear about this morning is a new concept called “contextual analytics” where you combine the use of data and video analytics. I will introduce it from 2 personal stories:
Mike Lewis mentioned yesterday that we are both on a tech committee. Two years ago, SOCOM, the United States Special Operations Command, came to us with a problem: how to use analytics to view their video. The problem was the sheer volume of video data they had, it blew us away. At the time we had no answer to that problem, but today Contextual Analytics has the capacity to address it.
The second story is about personal use. You may know that I used to be the CEO of a retail company. Before that I was CFO and I had the record for about 10 years in a row: I refused to fund loss prevention. We had a 74-store chain and 0.5% inventory shrink. That number was smaller than the cost of running a loss prevention department so I would just say no to it every year.
But I was really intrigued by the problem and since I learned so much from Rutgers and have a focus on analytics, I kept talking to our POS vendors about it. They came up with a solution where we could monitor the POS data for things like an unauthorized return. So we looked at simple analytics like no sales during four hours but in the middle of that time frame there was two returns. So perhaps our staff was taking a product and returning it to pocket the money.
This type of internal fraud detection was a breakthrough and we thought we were really smart, but what you’re going to hear about today makes that look archaic.
Video is the biggest big data
It started as a single problem: how can we solve ATM skimming fraud? At Solink we like to say that 1s and 0s can only tell you so much. At some point you need real-world insights. So our mission is to make IP security video more relevant, meaningful and accessible.
Just to give you some perspective and data points, everyday Google processes 24 Petabytes of data. At the other end of the spectrum, the surveillance cameras of the world produce 413 Petabytes. You really can’t have a conversation about big data without including video. (Check out our popular Big Data Infographic)
You’ve probably heard the saying that data is the new oil. If that’s true, than video is high-octane fuel. Here’s why.
The transactional data shows that nothing is currently happening. But here’s the full picture:
Unless you are physically there or looking at the surveillance video, you don’t see that there’s a long queue at the ATM or that a customer is frustrated and needs assistance. That’s the missing context of the picture and why so many systems and enterprise apps are limited by just “1s and 0s”.
The Problem With Video
As valuable as it is, there are several problems with video that make it difficult to work with. Businesses invest tens of millions of dollars putting in place video infrastructures across all locations and it’s just sitting there collecting dust. Here’s why video gets unused:
- Since it’s unstructured, it can’t easily be searched or correlated to other types of data.
- It’s an unconsumed asset. The infrastructures in place are not made to be easily accessible. There are a lot of privacy considerations around who should have access to video data and what they can do with it.
- There’s so much video that you can’t sit all day and look at it so how do you filter the important footage from the rest?
That’s where Contextual Analytics comes in. It’s defined by three elements: technologies that do real time advanced analytics, a large variety of data sources (from simple files to video) and the ability to tackle large datasets.
The technology that we use can be defined as event-driven. From all that video we extract metadata to create dashboards where people can search, filter, and interact with it. This allows users to quickly find specific events or exceptions that are significant. With just a few clicks you can filter thousands of hours of security footage down to just a dozen events that are relevant to your search.
The real-world use cases of Contextual Analytics:
ATM data feed registers 10 transactions using different cards- but the video shows the same person at the ATM
POS data feed shows a product return but the video doesn’t detect a customer in front of the register
In-store/In-branch Analytics (beyond security)
Brick and mortar stores and banks don’t know their “bounce rate”: how many customers walk in vs how many purchase, how long they wait in line and so on. For ecommerce companies, these are basic insights about customers and are easy to obtain. Video analytics allows brick and mortar businesses to get these statistics.
We are seeing shifts in the analytics and security industries that support our beliefs of where the market needs to be. These are three trends which have helped Solink grow and gain traction.
- People are moving away from proprietary DVRs and Vendors. More and more organizations want IT friendly equipment.
- There’s an explosion of data in the enterprise beyond comprehension.
- We’ve been monitoring the terms “contextual data” and “contextual analytics” for quite some time and at first nobody used to talk about it. But over the course of the last year more and more companies and publications are talking about how preserving context is a must have for business intelligence apps. In a big data world, having a full picture is what saves you from overwhelming response of false positives.
A famous statistician said “you can make data say whatever you want it to”. But the way to prevent that and maintain objectivity is with context.
Imaging you’re a retail store manager and you’ve found a case of internal fraud when auditing the transactions. You confront the employee in question who says “it wasn’t me, someone stole my employee card and used it for this transaction.” It’s easy to argue against limited data but if you had the video and transactional evidence side by side, there would be no question.
We are on the verge of a new era in security where video is becoming a continuous fraud monitoring tool as well as a valuable source of business intelligence. When you think about it, video captures so much detail about a business that it just makes sense to exploit and treat that asset in the same way as other enterprise data. At Solink, we are focused on making video analytics as intuitive to use as any other data system so that anyone in the organization can access the information they need at the right time.
Have questions about Contextual Analytics and how it might help your business? Schedule a call with our Loss Prevention specialists.