Yellowfin BI & meldCX: Computer Vision and AI Increases Customer Conversion for Retail

Published on
October 7, 2021
Stephen Borg
CEO & Co-founder
Joy Chua
EVP of Strategy & Development

This is a recap of a webinar that took place on the 6th of October 2021. Brought to you by meldCX and Yellowfin BI.

Brick and mortar retail is not dead. Instead, it is ripe with the opportunity to offer customers a more curated and connected experience! Imagine if you could measure your physical space like you would a website? Now you can, with Artificial Intelligence.

Watch this webinar to see how Yellowfin & meldCX have delivered off-the-shelf AI solutions for major Australian retailers, gathering real-time insights around:

  • Customer sentiment and engagement
  • Customer journey in-store
  • Planogram compliance
  • Back-of-house efficiencies to reduce loss

We will explore real-life use cases ranging from the traditional retailers, to retail banking and healthcare.



Natalie Mendes  0:02  
Hello, everyone, and thank you for joining today's webinar as part of our retail reimagined series. Today's topic showcases meldCX use case how computer vision and AI increases customer conversion. I'm Natalie members, and also your moderator for today's session.

But before we begin with our presentation, I'd like to cover a few housekeeping items on our platform. At the bottom of your screen, and multiple application widgets, you can use all the widgets, widgets are resizable and movable, so feel free to move them around to get the most out of your desktop space. If you have any questions during the webinar, you can submit them through the Q&A widget. We will look to answer these during the webinar. But if a more detailed answer is needed, or we run out of time, we will endeavour to connect and respond. An on demand version of the webinar will be available approximately one day after and can be accessed using the same audience link that was sent to you earlier. So feel free to share this to your network.

Right that covers all our housekeeping items. Now let's take a moment to welcome our speakers from meldCX. First started Steven board CEO and co founder at meldCX. Stephen is a solutions driven leader specialising in identifying and creating opportunities in Greenfield markets. Stephens key skills and strategy and design led thinking has helped multinational companies expand markets, explore and pioneer new niches to introduce game changing technology into competitive spaces. Welcome, Stephen. And joining Stephen from Eltechs is joy Chewa EVP of strategy and development at meldCX joint is a digital solution strategist, with a demonstrated history in information technology and the services industry. Joy is tasked as an advisor to key stakeholders across all industries. And she's passionate about digital transformation, especially merging the online offline customer experience, by utilising the best of connected technology in the physical space. Welcome to you both, and we look forward to your very informative session. So with that, I'll hand the mic over to Steven. Enjoy. Thank you.

Stephen Borg  2:14  
Hi, guys, thanks for having us.

Joy Chua  2:18  
Awesome, thanks, as well, Nat, for that introduction. I think, you know, hopefully after this session, you know, all of you will learn a little bit more about some of the key insights we have about retail and what we've really picked up I guess servicing some of our customers, as well. So we're really keen to take you guys through, as Matt said, this is really interactive. So feel free to drop us any questions, you also see through our session and the slides, there'll be little polls that pop up as widgets, it really helps us tailor you know, our content to some of your needs as well. So if you could, you know, feel free to log your responses there. That would help Stephen I greatly. Cool. So a little bit about us. You know, if you want to connect, please do reach out to us any questions that you thought we didn't cover? If you want to go deep dive into some of your problem statements for your clients, or your partner's open your organisation? Please, you know, our contact details on the screen. So yeah, you know, reach out to us, we always love to hear from you. So connect with us on LinkedIn, or our email addresses are there as well.

Stephen Borg  3:26  
Yeah, we're very passionate about this space. So this is merely a snapshot. So if you want to get deeper or understand more, by all means reach out.

Joy Chua  3:38  
So you see on screen now it should pop up a little bit. Um, you know, we want to understand, how would you rate your own organisation on its data driven journey? So that should pop up? You could log your responses, that would be awesome as well, for us. Um, see a little bit about mouth. Yeah.

Stephen Borg  3:56  
So you want to make up a little bit about mouth. So meldCX was actually born out of a large project that we were working with Google on the site at the time. And we're trying to work out how to best enable edge based analytics and data to filter back out backup to the cloud in a in a very usable way that can drive drive action based decisions, right? What's happening right now? Does that trigger an automatic decision? Or do I use that as data to make strategic decisions? And we worked closely with Google, we found a gap in the market and starting started to work with some really large cloud customers. And we really started from that. Okay, so what we've done is essentially provided an end to end solution that makes this simple. You don't need a team of data scientists to get up and running and allows you to use it for your business out of the box.

Joy Chua  5:03  
That's right. So that's exactly what Bob was talking about. So that's the product that we're hoping to take you guys through today. So that's called Fianna. So it's a suite of products within our ecosystem. That's all around here, like Bob said, really about driving, you know, and finding best of breed technology across the AI and ML space. And bringing that and combining that into actionable insights that, you know, teams like yourselves can then take and use that to inform business decisions. And later during the the session, as well, we'll touch on a couple of our customers who are using that as well. And you know, how you're using that and some of the off the shelf modules that we have that allow, you know, everyone to just take quickly spin up and run a use case and trial that.

Stephen Borg  5:49  
The biggest gap in the market we found was that multiple customers had providers that will build a model, but not how to execute it, get it to the ground, orchestrated, and pull data from that model that's usable. So that's really what we tried to pull together.

Joy Chua  6:08  
And, you know, one thing that we're passionate about as well is a huge shout out to the yellowfin team as well. One thing we're really passionate about is working with partners to build turnkey solutions. So, you know, Steven mentioned earlier in the, in the session, we work a lot with Google, you can see a list of our partners here who we work with. And, you know, they span you know, a broad range. Obviously, with yellowfin we've got a great partnership with, you know, we use yellowfin to do some of our custom and, you know, standard dashboards as well, that we take to market. And we work with the likes of other tech giants like Google, Microsoft, Intel, and Meraki as well for some of our edge instances.

Stephen Borg  6:50  
And we also have some similarities in some of our customers or our providers, white label us as well, to their solutions. So you'll you'll see, we've, we've got one really big one that you don't know we're in the background, but we deliver Spotify music in stores and do all that analytics. But no one knows it's us. Yeah, we just provide the platform for that to happen.

Joy Chua  7:15  
Exactly. So a little bit of other retail landscape as we see it. Now. I think there's, you know, a lot to be said about retail. And I think there's some really interesting studies, which we'll talk about a little bit more, that even ourselves is not we are looking to participate in, I think COVID has made the whole experience of retail, what I guess, prevalent, I guess, like everybody loves retail. And one thing that we found as well as, as you can see on the screen, we've got some sources that say that, you know, still, the physical store is still actually preferred to make purchases, which was one really key stat that I think when I was discussing with Steven, we will really I guess, in all about, personally, myself as well, I do like going to the retail store, to shopping, even if I might, you know, pick something up online later, after trying it out. That's how I wish for myself and for my age bracket.

Stephen Borg  8:14  
And I'm always seeing some overseas clients, although they had a surgeon online, that once opening up occurs, but they the inside presence is much more in demand. Right. So and there's a higher expectation that that presence will be more shop attainment, yeah, than anything else, right.

Joy Chua  8:37  
Exactly. So these are some key steps that you can see here. You know, we found some key steps. That's it, and the 64% of shoppers say that they still feel that retailers don't truly know them, which is really interesting given, you know, all the tools that we have at our fingertips now, you know, personalised emails, you know, you know, custom offers to ourselves, you know, when we shop at a brand for a while, like, you know, loyalty programmes as well. Still, people, you know, the majority of us feel that, you know, the retailers, which we shop that don't really know us.

Stephen Borg  9:12  
And I think this statistic is becoming more and more important, because when we talk to retailers, they have compressed supply chains. So they really need to target what they do have against the audience that will buy them, they no longer can have a massive in store products to choose from has to be very targeted. I'm finding that they're asking us to help them target that.

Joy Chua  9:35  
That's right. And as you can see, we've got our next stat as well that says, you know, we found an average 40% of the average order value increases when shoppers acted on AI powered product recommendations, and we'll talk about that a little bit more and how, you know, we've got a product that helps track those and measure those so that retailers like yourselves or your clients who are in the retail industry can take and make use of To make those insights and actions.

Stephen Borg  10:01  
Yeah, and our mission with that is to deliver all that cool tech news online around recommendations and product affinity, and bring it into store into a location.

Joy Chua  10:13  
Exactly. And I think shortly as well, I mean, I started off the session by saying, for us, we're going to participate in a couple of studies as well, I think, you know, pre COVID, during COVID, and post COVID, we're going to see, you know, a lot of people have been talking about that, we're going to see a fundamental shift in how customers or that customer experience is, especially in retail, and even for us, because we've got multiple products in our ecosystem. We've been talking a lot internally about what we call that new board and ratio. And that new golden ratio is, you know, what is good for the physical store? How do people interact with kiosk, this is how much of it goes online. And I think for that, that's the new measure, I guess retailers have to unpack in this new, you know, post COVID, as you talk about opening up, and we'll share a little bit about how we're hearing our customers navigate this new kind of golden ratio for retail.

Stephen Borg  11:09  
And one of the thingsyou're seeing in, in Europe, in particular is a saying that customers will only enter a store, if they feel that that store that location is has responsible practices around COVID. So we've been asked to help with that. And you'll see a use case for that as well.

Joy Chua  11:28  
Yeah. And I think we've got a little a question for Matthias as well. So thanks for your questions. And last question is, do you think there's a risk of alienating people and retailers seem to know them too much? Yes, that is a very, that's a question we get asked a lot.

Stephen Borg  11:44  
And sometimes we even pull customers back. Right? Yeah, I think being able to target and not feeling not making customer feel alienated unless there's, they've opted in and want that full experience, I think is really important. So sometimes it's, it can be a subtle communication. That seems like it's too all. But it's, it's, it's quite targeted. And on the other hand, we do things like hotel check in, which has retail spaces where they want that single experience after check in and want to know, everywhere they go, it continues that experience. So really depends on the environment, and the use case. But 100%, you need to be conscious of that. Yeah. And use the technology in a responsible way.

Joy Chua  12:31  
And I think all cover that as far as how it's how we're doing that. And I think the issue with alienation and knowing too much, it's all about how much do you personalise really, like how detailed you get to know that individual. And I think, for us as an organisation, there are a couple of things that we take to prevent this identification of the individual. And I think if people are grouped together in a stat, so for example, like myself, I'll be grouped into a certain age bracket, you know, female category, if I'm part of a persona group, and, you know, targeted as part of what my persona group would like. I think that, that, I guess, drawback with that pushback is actually much less than if it was very personal.  

Stephen Borg  13:14  
And we'll take you through shortly that all information is also anonymous. Right? Yeah. We're never going to target someone for the wrong reasons.

Joy Chua  13:26  
So yeah, so in summary, with everything that we've learned as well, you know, I think a lot of things that we've been hearing pain points from our retailers has always been about how do I use my data to tell a story even better, a story about brand, a story to my customers about what I stand for all of that.

Stephen Borg  13:42  
A story about how my visual communication is working and in store.

Joy Chua  13:47  
Wxactly, both internally and externally. And so that's what we've done. Basically, it's an a simple end to end, you know, platform for organisations to visualise the data, or actionable insights.

Stephen Borg  13:58  
Yeah, and it still has the flexibility to output that data into data sources that you might use as a retailer. But what we've done is we've taken all that learning and all that data to create reports and create visualisations that we feel is the most effective representation and correlation of data that helps helps you instantly.

Joy Chua  14:22  
Exactly. And so it will come to you know, how do we actually start talking about all of that guess, how do we help organisations deploy AI or ml. And so if you can see on the screen here, we've actually got three ways. Or rather, these are the three building blocks, I won't say these are the three ways or say these are the three building blocks. You know, customers have worked with us on in order to run your AI and ML use cases. And we'll talk about how you can interact with us and then we'll talk about what's available kind of off the shelf that people can deploy in that vertical

Stephen Borg  14:57  
and this is what we say is a typical order, but we do Have some customers that start right at the end. Because the sophisticated, they've gone through the process. And now they want someone to effectively pull together what they were working on, we think

Joy Chua  15:11  
customers like that have a very key problem statement they want to solve, and you're very clear on the outcome. And that's where, you know, we interact with them, like Steve was saying straight into that personalised or custom bucket, where we work with them to really train specific AI data sets that can help support their use cases. But with this particular building block, we generally recommend, if we were to do that, which we can, and we'll show you a little bit, you need to have a really clear problem statement and know what you want to solve, I think that's how you get the best outcome from like,

Stephen Borg  15:45  
ironically, they might start with that, and then sometimes they end up going back to start using, they've got their specific use case solved. And then they use all of our out of the box, use cases and evolve that way.

Joy Chua  15:58  
So essentially, the three building blocks for that we use to help organisations in their data driven journey. So the first is ready to go. And basically, you know, we generally recommend that for, you know, organisations that want to quickly start a POC, proof of value to the business, you know, they don't want to get your IT team in touch with you know, them too much as a lot about driving the business and a use case, we've got, you know, ready to go off the shelf modules, which we'll show later, that can be you know, easily taken, deployed, and just run. And so that's one of our building blocks that we do. The other building block that we've got is to add your own module. So with that, basically what it is, is, you know, you, as an organisation are already, you know, experimenting, or you have started your AI and ML journey. And, you know, you're looking for other models or algorithms that can help supplement that, or complement that a perfect

Stephen Borg  16:58  
example, we had a retailer that did a lot of work around people slipping and falling. And they wanted to detect that. But they had no other models that complement that. So in isolation, it didn't do well. So they combined it with some of our existing models to know if someone slipped, and then they got up and left, or they went for assistance, or they couldn't get up again. And once they combined it with our models, they they were able to roll out their use case for. So we've got to use what they already had invested in and paired it with some of our technology

Joy Chua  17:33  
definitely. And more often than not, we do see hybrid model, where, you know, we'll talk about some of the use cases where an electricity Australia posts, we work with them on a personalised model dataset. And they also use some of our ready to go modules to service their entire AI or ml needs and that data capture tool. So we're just going to show you a couple of examples, as you can see on the screen here. And these are what we call a ready to go module. So these are off the shelf, they're ready, they can be deployed, all you need is really a camera. And depending on what use case, you've bought, maybe an edge device as well, it could be a small, a small Media Player, like an Intel nook, for example. And these are what you have off the shelf. So you've got entry monitoring, which you can see I don't know if you can see my mouse and monitoring at the bottom, we've got audience measurement, which is here. So entry monitoring, it's all about measuring foot traffic installed.

Stephen Borg  18:30  
And some people use this for the amount of counting. So right now they might use it to monitor capacity. So the doors won't may not let anyone else in if it's at a COVID capacity, right? That that type of thing. Or you can just monitor the amount of traffic. We have some retailers that are monitoring their near store traffic, because they don't believe in the shop, the shopping centre data is his actual so that they're trying to understand the real data that's in front of their stores.

Joy Chua  19:00  
Yep. So that's an example of an off the shelf module. So this entry monitoring, we've also got another off the shelf module over here, which is called audience measurement. And audience measurement is basically you know, we're looking at digital content or tracking how people are interacting and looking at digital content and giving those stats back so that you know digital marketers or product marketers will be able to tailor their messaging to the target persona that's most interested in that content. So that's another off the shelf module. We have another off the shelf module. Do you want to talk about Sammy, which is right here?

Stephen Borg  19:34  
Yeah, so Sam is a module that monitors or created digital manifests for cleaning. So you can point to that table and you can say, Okay, once X amount of people have interacted with that table or that booking, I will make that table unavailable for booking until someone cleans it. You'll want us to the hand movements of someone cleaning it and make it available again for booking It can be used on any surface, you just highlight that surface. And we're finding it being, you know, used in hotels, the likes of Australia Post. And it really makes sure that the cleaners that you've hired to go do that, that role, specifically claiming that and you have a full digital manifest for automatically created. So we're finding that becoming a very popular module right now. And that can be done for anything we've got customers that are using outside of the COVID environment for food preparation areas are anything that needs to be kept clean.

Joy Chua  20:38  
Yes. And the last module that you know, is available, you know, off the shelf as well is right up top here is what we call our zone engagement model, or module. And the zone engagement module is really a module where we're tracking tokenized individuals, so anonymous tokens as they pass through the different zones of a building or store, for example. So that what that means is we're aggregating an individual as they cross through the different camera feeds to understand their dwell times how long they're spending in a certain zone. And you'll see in the next slide as well, we can even go as granular as to what products they're interacting with. So just bear in mind these four core modules. And so that's if you're monitoring zone engagement, semi, which is our surface cleaning, and audience measurement, and we'll show you so these are all our ready to go modules, but we'll show you in a context as to our second building block, which was, um, bring your own model, how we are combining ready to go with bring your own models. So this is our next slide, did you want to chat a little bit about prior engagements.

Stephen Borg  21:49  
So product engagement is a is a great model and module we've just released. So it allows us to determine engagement with product, right. So for example, we work with a client to determine on their, in their retail spaces or their benches, how many people engaged with a Samsung Galaxy versus an iPhone 12. And we provided that detail, we provided sentiment analysis when they were holding the product, how many times they switch between the product, I can force that the product placements effect that product engagement. So effectively, it's interesting is that customer got to the point where they wanted to go back to vendors and charge per engagement like charging per clicks in their retail stores. And this would allow that to happen. So it's an exciting module, we can also compare regions of shelves. So if you're a brand, and you're you've, you know, invested in the whole day, but for some reason the sales in the next day of another brand is is getting better results, then it could be offered based on those shelving structure or could be how you've presented that shelving. Yeah. So all of that detail. trickles up comes to light, you can compare across multiple stores. And you can also cross match it with all the other data versus how many traffic how much traffic there is and how much engagement you're getting at a product and brand level.

Joy Chua  23:20  
So if you notice on the screen, we've got those same four core modules that I talked about that are available off the shelf. So as you can see on screen, we've got entry monitoring, we've got zone engagement, we've got audience measurement. And we also have Sammy, which is our surface cleaning module. The only difference in this example is some customers have come to us and they've already started playing with the idea of product engagement and customising some models specifically for product. So what one of the retailers came to us with is they had some model data sets for say, lipsticks, for example, or mascaras. And they put, you know, they brought those own data sets that they have, and they capture. And we ingested those models into our platform to add kind of a more holistic, you know, end to end start to finish, customer journey, kind of mapping an AI and an alpha. So yeah. So the last one as well, hoping this place, please let us know if it's not playing by if you can see on the screen here, this is the last one that we wanted to talk about in terms of building blocks. And this is what we do when it comes to personalised training or how we build custom models. And as you can see, what we do is were creating a 3d environment. And our approach is all about using synthetic data. So that we can simulate kind of really stick environments without having to, you know, me that much video feeds and allows us to manipulate elements in it

Stephen Borg  24:57  
and create multiple scenarios without needing To train. The other benefit of this, and this is why we're endorsed by Intel, Microsoft, Google and Cisco is we don't use human data in this training. So we make sure that the training can never be biassed, racially biassed or any of those things that we've seen in facial recognition of the past, right? So it's pure synthetic data, we've created avatars. And we don't need to use hours of data of people stealing, or doing the wrong thing to to create a dataset.

Joy Chua  25:35  
That's right. So I'm going to see that the videos or the videos now playing I guess, so apologies, guys. So this is an example of a 3d environment as to Yeah, you know what matching, as you can see, for this particular example, we're actually looking at loss prevention. So you can see the gentleman in grey is actually going to start picking off the shots. And with that, you know, we can manipulate elements, like what Steve was saying, we can change. You know, if the he's picking lipstick, so this is.

Stephen Borg  26:09  
So this was a scenario where people put on uniforms or tops that look like employee tops, they went into multiple stores and stole a whole lot of high value items. So no one bothered them. And they, they, it was pretty good, pretty good heist. So what we did is took those actions and monitored any hand movements that were removing products, successively, and then applied those back to the cameras. So now all the stores will look for that action, and notify.

Joy Chua  26:44  
So we're going to talk a little bit about the customer journey. And I think this was also to Matthias previous questions. So thank you, again, Matthias for your question about, you know, alienating people, because we seem to know too much. And so we're just put up on screen here a little bit, talking a little bit about what we mentioned before our anonymized approach to capturing data, and how we're taking the steps as an organisation to do that. So if you can see on the screen, we've got six kind of blocks, okay, six key statements are things that we do to make sure that, you know, we're keeping everything anonymized. And so we split those up into, you know, anything that's non face, and things that are, you know, to protect people's privacy, if you can see over here, so some key things that we really follow, when we're looking at any images on our platform, faces are blurred in all streams and process at the edge. So every person has turned into a token, you can see, on the left, this is an example of how people would stuff is on our back end platform. So we actually don't capture any face. And to add to that, we, as an organisation don't do any facial recognition. So one thing that we've learned in the data journey very early on, is that a person is more than a face, really. And so there are other points of reference to a token that are important. And these includes, you know, objects, as well as non face behaviour. So object recognition could be things like, you know, what someone's wearing, recognising tag or a lanyard, and non face behaviour could be, you know, acts of aggression, or falling in aged care, or even, you're not that action of picking something from the shelves.

Stephen Borg  28:26  
So by not identifying the individual, where it gives us more scope to look for behaviours, and every token creates a manifest. So then we start getting a manifest of that token, and can build out, you know, the frequency of shopping, where they travel in location, what they actually bought, when over line. And even right down to we have a customer that wants to look at the spending capacity of a particular store or location. So we're looking at Brands, you know, are females holding a Gucci brand? are they holding, you know, a coach, right? That indicates spending capacity of the region. So the other type of things that we start to look at, because we're not identified

Joy Chua  29:17  
yet. And so, if you can see on the left this, this is an idea of what kind of metrics that you can get from a tokenized individual, and including mood. And as Steve mentioned, you know, apparel, what you're wearing your cultural Gucci bag, you know, and attention span. So these are just elements of what we can capture of a unique tokenize individual. Bless them. And so this is, you know, what we do as well in a process called Anonymous re identification, and that's where, you know, as I mentioned before, we can aggregate basically data across the different camera feeds to track individuals as they move across the different zones. So that means we're able to get more accurate data because we're capturing that persona, we're not recounting a person twice, if that makes sense.

Stephen Borg  30:04  
And these can be very powerful. We even have some customers that monitor if, if they're right wearing the right safety equipment in their warehouses or in their picking or truck areas. So it's the same rules apply, we just make it look for certain things such as a vest and a hardhat. Or, you know, it's an authorised person that has a badge. Right?

Joy Chua  30:27  
Understood. And so this is a really high level snapshot of how we can capture customers journey as they pass through the different zones, and how you're interacting with the space and what we can do with that, as well. So that's just a high level overview. So you can see we're capturing the different zones that they're matching with, what their actions are, and how long they're spending in each of those zones, as well. So this is just a high level view. So before we went to the next section, as well, and how we're actually using data fee, yellow yellowfin as part of our data journey and how are surfacing up to dashboards, we've got a little poll as well, to help understand, you know, last sell off the top outcomes and metrics that are important to your organisation, and what's the first thing you want to do if you could do it as well. So awesome, I think, if you can see on screen, this is a high level overview of the customer journey, we've broken it up into six key points on, you know, from when a customer enters the store to when they make a purchase. And so these are six key actions that we see from Discover, to connect, explore, Search and Compare, experience, and then purchase. And we'll show you some of the more modules that we have within the Viana product that match up to these actions, and how we're using yellow fins. toolset to help us surface that data to our customers.

So this is soul level activation, or rather, our discover phase. And as you can see here, you know, we've matched is to basically our people counting or entry monitoring module. And you can see a couple of examples as well as to how we're using yellowfin to surface that data. And we're using that to build a really clean looking dashboard for our customers. And, you know, if they need a little bit of customization, we can do that as well, we can work with one of their partners or the agency partners to help bring that across. So this is an example of how we're matching basically the entry monitoring module together with a particular dashboard.

Stephen Borg  32:52  
And this one's an interesting one, because one of the most common problems we find for retailers, and even sort of banks, we find that the content they're producing does not match the dwell time of the region. Right. So my most often the content is longer than the amount of time people will pay attention in that particular area. So what we do is we we monitor that, and we help you tailor your content. And we had a recent use case where the content effectiveness or the amount of watching went from 30% to 84%, just by making the content a little bit shorter, and having the right colour backgrounds because it wasn't the same as the backgrounds that was on the wall. So that made a massive difference and increased conversion dramatically. So if you're going to spend all that money on content, and a digital signage network, and you're delivering content that's too long for that location, they don't get to see the call to action. If it doesn't make sense.

Joy Chua  33:55  
I will talk about that case study in a little bit. It's I don't have slides, I might skim through some of these so that Stephen and I can focus more on some of the use cases that our customers are using. So this is another example of our off the shelf module. As I mentioned before your engagement is one that we use to track a personas interaction with Zoom, and how long is spending with the zone. So this is some of our dashboards that we've got. And this is what we mentioned as well in terms of shelf engagement or product engagement. We can overlay different forms of technology like kidnapping, or even with interactions to see how different personas are interacting with products on the shelves. And this was to Bill's point before we can track if someone's interacting with Samsung phone versus an iPhone 12. So as you can see, as well how granular we can get you can see a little dashboard here with all the different you know, data that's captured here, but a different point the heels up here that's just an example. or what we've done as a POC for one of our retailers, where we're tracking, you know, the different brands of phones and how customers are interacting with it. And we're able to help them get to an outcome where, you know, you were able to either request for more MDF from a certain brand,

Stephen Borg  35:15  
or position it the right way or relocate it in this case,

Joy Chua  35:19  
yep. And so this is another interesting one, I might skim through it. But we can also do what we call post correlation. And that really is, you know, tracking, you know, whether a particular item, I'm just going to use a phone, for example, whether a particular phone has transacted and so we can assign amounts to it, there's two ways you could do it, we could integrate to pause to get that or we could assign, you know, we don't want to integrate to pause, because there's a lot to do from a technical perspective, we could assign certain data values to it and just get an aggregate of what has passed through to

Stephen Borg  35:53  
one customer that countries can attribute every single ad to the sales uplift of a product. So they're correlating the ad went up played versus didn't go over line versus positive data.

Joy Chua  36:12  
And so these are some of the, you know, retail stats that we can get. And, you know, we also provide this to the, you know, staff from to head office, for example, staff working at head office to understand you know, how people interact their content, are they happy or sad, but these are all available as well. So it's not just available to, you know, frontline staff to understand but also, back in off coordination, multiple departments want data in different ways.

Yeah, exactly. And so wanted to talk a little bit about some of the use cases. And so this was something that fog mentioned, as well. So we're working on one of the big four banks, as you know, Westpac, we actually do have an Intel white paper on it as well. So if you're interested in, you know, the Intel white paper that we did together in collaboration, please visit our website or reach out to Nat and Stephen, we can send that over to you the link,

Stephen Borg  37:02  
ther e is a bit of content online, we want a Paris Design Award for this work. So there is someone that submission online search as well.

Joy Chua  37:10  
And so with this particular example, this was exactly what Steve was talking about where, you know, we weren't helping the team at Westpac understand, you know, Westpac has so many product lines, so many different first home loans, so many different credit cards, we're helping them understand which product was most appealing to which persona or target demographic and validating that. And so we deploy, basically, our audience measurement and content effectiveness modules, deployed together with their digital signage network in branches to understand you know, which audience segment was more interested in what product and, you know, bring more value to the branch that way,

Stephen Borg  37:52  
and we even got viewers, we even got sort of how many views through when they experimented with multilingual content, so they can hone in the appropriate languages.

Joy Chua  38:05  
So that's right. So that's a use case that we've got for retail banking. Another one that we have is up that's just popped up. And I find that we have is we're working together for Australia Post this one as well. There's a lot of press on this. We've actually displayed this together with Intel, at the NRF show in 2020. Before the pandemic happened, we're actually in New York, demonstrating this happening, taking up, but we worked with them to do two things on our floor. So this is an example of a very personalised use case or a custom use case, when we were working with them to understand possible lodgement in two ways, one, how a kiosk would work and how, you know, the user experience would be when you're doing self service on self service parcel lodgement, on the kiosk. And then the second one that we're also working with them is proof of lodgement for small businesses, when you're also doing self service, but not to at the kiosk, but actually had that parcel drop off shoot that most of us may have seen it in Australia.

Stephen Borg  39:17  
And the net result for this is that he didn't have to scan your own item and wait at the counter, it means you can go and drop it off at the at the at the shoot. And we would acknowledge that and let you know that we've received it so you didn't have to wait in line. So they did that for multiple reasons. COVID less contact and also to have less people in store. So you can just get in and out.

Joy Chua  39:42  
Exactly. So these are, you know, given two examples, I guess one is a standard dashboard or off the shelf modules that are available now. And this is another one that's custom and so this is another one that We're working on. So this is a little bit of a hybrid one, where we're working on zone engagement for an Australian electronics retailer. And we're measuring, you know, how customers are interacting as the pass through to different zones. What's the target persona for the different zones, you know, dwell time, especially, it's all around getting more information on the customer.

Stephen Borg  40:23
And this also enables more vendor funding and confidence around their vendors that are providing the store in store experience, or concession areas.

Joy Chua  40:34
So for this particular retailer, as well, we've got some other stuff in the pipeline, this will be a hybrid approach, where, you know, we'll be working with them, this is an off the shelf module, of course, but we'll be working with them on some other personalised use cases like loss prevention, but that's kind of phase two. So watch this space, guys. So yeah, so I guess, you know, to sum up with the remaining time, we have a yes, you know, what's next, retail really. And so I think we spoke about this a lot, in the beginning of the session, I think, in the post COVID, well, there'll be so much that we won't need to re understand, obviously, you know, this 18 months for everyone has really shaped customer behaviour. And we'll share customer experience as we get out of lockdown. And we spoke earlier about that golden ratio, you know, kiosk, self service, or kiosk versus, you know, visible stores, versus that online experience, and really working out kind of in a degree here between how customers want to interact with each of these. But one thing that we're finding is, you know, read a really good article that see, as you know, post COVID is going to be so much that's going to be needed in a customer experience basis. Yeah. And, you know, a lot of customer experience with CIO is especially are really looking to redefine the CX experience over the next five years.

Stephen Borg  42:01
And I think the expectation is greater. So we're finding, as we said, where they opened, where we see open up, opening up, again, the expectation is greater from, not only from experience point of view, but from a perceived compliance point of view. So those two combined is a big onus on any retailer. So, you know, we've tried to make it as easy as possible.

Joy Chua  42:24
And I think that's why, you know, I think organisations need to start that journey. Now, if they haven't already, on how do we start capturing and understanding customer experience, using, you know, some of the AI building blocks so that they can quickly pivot, they can quickly understand, and they can be nimble and agile enough to, you know, come out of this kind of post COVID world and understand consumers again. So, yeah, I think that's where, you know, us and yellowfin come in. And we're really excited to partner with the team as well. So I think in terms of next steps, you know, thank you, again, everyone, for your time today. If you have questions, please feel free to reach out to us. And, you know, please drop myself or Steve, a note, we'd love to be connected. We love to explore new things, new use cases. So we're always together and let us know.

Stephen Borg  43:22
We appreciate we've covered a lot of content in a short period of time. So if there's any questions, by all means, feel free to.

Natalie Mendes  43:30
Yeah, thank you. Thank you, Steven, and Joy. And down, we have got the Q&A chat box open. So for the audience out there, you've probably got a couple of minutes to submit a question and get time from seven enjoy. But really the way you guys are using computer vision and data is so sophisticated and bringing edge analytics to surface is probably an area where you don't just have to use it in retail. I see a lot of other applications in you know, even train stations and you know, police stations and other areas with this. Computer Vision, Edge analytics and sophisticated data can be used. So congratulations. I'm going to be very wary about if I walk into a shop and I hear 80s Music playing. It might just get my experience happening if it comes up on Spotify. One quick question. One quick question is from having the data in a ready state for the use of your AI models. How long does it take to to implement a programme like this?

Stephen Borg  44:39
It depends so if you if you're looking at one of the out of box ones, it's really as simple as connecting it to your cameras. to your IP cameras or your USB camera, we have a module that you can go in and select the areas of interest yourself, and you're up and running. So we have customers get up and running in a day, and it starts flying down to their, their cloud, Western models, it really depends on the complexity of it. But we are doing one right now, which is very complex. And it's taking us phase one, it's, it's taking us about three months.

Natalie Mendes  45:24
That's quite a short timeframe, which is great. If people have no drowning in their own data, they can get up and running with some quick start.

Stephen Borg  45:33
I'll give you perspective, our first project, we didn't use synthetic data, it took two years to get the first data model up. Because the the amount of data. And this is a published one, right. And then we did Australia Post. And we had to do handwriting recognition. We did it synthetically. And our synthetic engine was creating 11 billion references a day, rather than us having to scan 11 billion parcels a day, which is impossible. So three years work in 12 weeks.

Natalie Mendes  46:06
Wow. Well, I'm sure some of our audience out there would be excited to hear that. Because you can get value quite immediately. Well, we have run out of time, I wanted to just congratulate you, you did win the Yellowfin innovation award that was presented to you yesterday. And it's no surprise why you won that because of the sophisticated stuff that you guys are doing, and the great partnership that we have. So thank you to you both for your informative session today. And thank you to the audience out there. If you have any questions, please reach out. There's a resource section where you can download some information in the platform about meldCX and yellowfin. And there'll be a short survey once you want to sign out that if you can complete it helps us get some feedback on our webinars. But again, Stephen and Joy, thank you so much. That was a great session.

Stephen Borg  47:04
Thank you.

Joy Chua  47:06
Thank you, everyone.

Natalie Mendes  47:08
Thanks, Bye

Latest from meldCX

meldCX's Vision Analytics Solution Showcased at TD Synnex Showroom in Munich, Germany

Mar 19, 2024
2 minutes
Featuring our Content at the Right Opportunity (COATRO) solution for digital signage. In partnership with Intel and Signagelive.

Get the latest meldCX news and insights right to your inbox!