How to calculate video analytics ROI 

In a world built on dollars, cents and occasionally sense, forecasting and attaining a return on resources invested is a must. But, sometimes the maths isn’t straight forward. Further, the return isn’t always strictly financial. How much is public safety worth? A human life saved? 

While video analytics software can programmatically detect objects and actions in video footage, which help streamline processes and generate insights that can underpin new revenue streams, there’s a plethora of security and safety use cases too. 

Editor of Security Solutions Media, John Bigelow, leads a panel discussion on video analytics ROI at Sydney’s 2023 Security Exhibition & Conference, Using Artificial Intelligence to solve real-world customer challenges.

John Bigelow, Editor at Security Solutions Media
As security managers and security departments, the first thing that any board is going to want to ask is [this]: ‘How do we measure Return on Investment, if we’re gonna give you $500,000?’ Or a million dollars or $2 million to go out and deploy this You-Beaut, Whiz-Bang artificially intelligent security system. How do I determine what the benefit of that is to us as an organization? 

John Bigelow
So, Patrick, I’m gonna throw to you and ask: How, when it comes to implementing new video analytics solutions and technologies like this, do we define and understand what the Return on Investment looks like?

Three rules of video analytics ROI: Save money, make money, reduce risk

Patrick Elliott, CEO and Cofounder at VisualCortex
I think this is really important; [this is] where you need to have a stakeholder involved. 

So actually, within the same customer, we’ve had a couple of different use cases that we worked on. There was one that had a stakeholder that actually had a very strong financial understanding and model of what they were trying to achieve. We were able to do the Proof of Concept, we were able to show the initial results, and then work our way through to a production environment that paid off tenfold. 

Patrick Elliott  
We have another use case, in that same organization, that was probably one of the most important metrics to them as an organization, we were actually able to deliver really high accuracy in that case, but they did not have a business stakeholder who could then go forward and say, ‘If we spend money on this, this is going to be the extra incremental reduction in cost or incremental revenue stream’. It was a very strong lesson for us because when you start, you try to do a lot of things, right? Especially when your customer tells you, ‘this is exactly what we need to do’. And it was what they needed to do. But if you can’t financially understand what the outcome is going to be, and you don’t have a stakeholder who’s going to back it, it’s going to be very difficult. 

Patrick Elliott  
So I think there’s probably three [ways to measure ROI]: There’s the can we turn it into a profit center to generate more revenue? Can it reduce costs? And then probably the most important one, and the one that seems to get through the easiest is, is it a compliance thing? Is it something where we’re really helping people: Reducing liability, or we’re helping save lives, or reduce risk. So those seem to be the three key catalysts.

John Bigelow
Michael, you’re sort of nodding your head vigorously there when Patrick mentions compliance. From your experience, how do we measure Return on Investment with these things?

Michael Lang , Solutions Architect Manager at NVIDIA
It can be a little bit tricky. And this comes back to a bit around building and buying. 

So when Patrick’s organization gets involved, you know, VisualCortex can do all the magical stuff and the business can say we’ll deploy it, right? So there’s a much lower overhead for the business to do that. So they’re buying. If you’re building yourself, that gets a lot harder, because you don’t know where the end point is going to be. No software development project goes according to plan or on time, right? But I think, if we’re looking for an outcome driven capability, again, I think the compliance piece is critical. That might be a safety piece. 

Michael Lang
We’ve seen in the safety space — people like recyclers, who have deaths because people get used to being around machinery every day and then somebody loses an arm or something [because they’re rushing], and it all goes bad. How do we have the eye-in-the-sky, that never gets tired, never gets bored? And what’s that worth to our business? Either holistically, or to a dollar value as well. So that one is actually very easy to measure. But you know, many businesses on the other side. If we talk about stop-loss — which is an increased consideration over the past several years, and there’s examples overseas where, you know, entire stores get emptied rapidly. How do we deal with those things? Is [the ROI we apply to video analytics in that scenario] a risk mitigation to our business? 

Michael Lang
So I think those metrics are quite different for every business. But also, [if you take this conversation back to your first question],  if you take that from a board level down, how do they perceive the reality of AI? Some customers we still talk to today, still say: ‘Is it real? Come on, Is it real?’ And other customers are like: “Well we’re already doing it’. And I like to introduce the first one to the latter one. 

Michael Lang
So again, your reality may vary. But we are trying to change the reality of the business. In the same way that, perhaps, going on the internet and getting an online presence may change how you do your business. So there’s a few different views of that. But, at the end of the day, your competitors are doing it too, right? So that’s a big push.

Video analytics ROI: “The Return on Investment is human life”

John Bigelow  
Aaron, in your experience, working with the facial recognition side of things, I imagine you set out to develop a solution that would solve certain problems. And then you found, along the way, that it’s picked up all sorts of other things or, or solved all sorts of other challenges that you hadn’t initially foreseen. 

John Bigelow  
Things like being able to identify people coming into clubs, who have self-excluded, that then save the club from being sued against loss. Or maybe, falls in hospitals, where people are suing the hospital. They’re easily quantifiable metrics. Are there things that you’re seeing that help determine how we can measure a Return on Investment around these things? And, as security managers, do we need to start thinking outside our traditional silo and looking at what we can bring — from a business intelligence [perspective] — to other departments within the organization? 

Aaron Terrey, Director of Vixles Pty Ltd
I’m pretty passionate about this. And the Return on Investment is human life. And, for those who don’t know, South Australia — for venues and gaming — legislated facial recognition to help with bard people. Bard people included those who self-excluded themselves [because they] have a gambling addiction, and who ultimately may lose their family, may lose their homes, and may actually take their own lives. 

Aaron Terrey
And the complexities of trying to understand if a person who’s got, you know, a hundred people coming in every half hour to a venue, [can] actually remember a face of somebody who’s actually put their hand up and said, ‘I want to self-exclude, I’ve got a problem’. The facial recognition will, hands down, beat a human in terms of detecting that person coming in. And then the human will validate that to make sure it’s accurate on top of that. So in terms of helping people’s lives, it’s significant. 

Aaron Terrey
We’re working with a customer in New Zealand, it’s one of the largest supermarkets. Two days ago in the press it came out that the CEO of the supermarket said that criminal activities — so, shoplifting and violence — are up 60% In the last 12 months. Twenty percent of that is in the last quarter. And in terms of loss, that’s significant. But what is actually happening is, one in seven of those people that are caught shoplifting, are turning violent and aggressive. Staff members are being assaulted.  

Aaron Terrey
They [now] have a staffing problem because… they’re actually taking leave, at a cost back to the business, because security guards and shop attendants… have been injured. And, again, just to understand that you’ve got a problem, [by] using this technology, [is significantly advantageous]. To understand that you’ve got a problem, where a repeat offender has re-entered your store, and [to perhaps run that person’s likeness against a known individuals list] and [determine that they’re] a violent person with a trespass notice [is important]. And you might have a flag [as part of your video analytics system, which will appear] in real-time. So let’s use this technology to help save lives.  

Aaron Terrey
And Patrick’s working on a couple of initiatives around understanding this behavioral analysis. So, again, using this technology for good and Return on Investment [in regards to] human life. They are working with customer, a couple of customers, on understanding the behaviour of a person that might commit suicide. And you’ve only got to save one life, by having that early detection, [to make it worthwhile]. And the technology is not saying there is [definitely] a problem. The technology is flagging that there could be a problem. So let’s then get a human involved, and go and [potentially] save a life.

Where to next?

Keen to hear more? Check out the full panel discussion – Using Artificial Intelligence to solve real-world customer challenges – here: https://visualcortex.com/2023/11/29/solving-real-world-challenges-with-video-analytics-software/

About VisualCortex

VisualCortex is making video data actionable in the enterprise. Its Video Intelligence Platform provides the stability and flexibility to productionize computer vision technology at scale. Able to be used for any video analytics use case in any industry, VisualCortex’s production-ready cloud-based environment transforms video assets into analyzable streams of data.

The VisualCortex platform delivers the artificial intelligence smarts, governance and usability, enabling organizations to connect any number of video streams, repositories and use existing commodity hardware. An intuitive user interface, out-of-the-box reporting, range of configurations and integrations empower non-technical people to produce, analyze and act on insights derived from computer vision throughout the enterprise. Organizations can easily combine these AI-generated video insights with other data sources and systems to facilitate both real-time operations and strategic analysis. The VisualCortex Model Store also provides a secure marketplace for customers, partners and independent machine learning experts to share quality controlled computer vision models.

For more information, visit www.visualcortex.com

For regular updates, follow VisualCortex on Twitter (@VisualCortexApp), LinkedIn (VisualCortex), YouTube (VisualCortex) and Facebook (@VisualCortexApp).

For regular industry news and analysis, subscribe to VisualCortex’s mailing list here: visualcortex.com/contact-us