The following is a transcript from:
PreEmptive E-commerce Podcast Episode #5: -Recommendations & the Power of Suggestion with Ecommerce
Introduction:
Jon Carmine 0:03
Welcome to the Emptive podcast, where we chat about cloud based solutions that makes selling your E commerce products easier, more profitable and effective. T
oday, we’ll be digging into the power of suggestion with E commerce recommendation engines. What are they? What kinds of recommendation systems are being used today by big companies like Amazon and Netflix? But also, how can the rest of us get into the machine learning and AI world right now?
Reviews:
David Waterman 0:30
I would say reviews would be the number one that when people think of recommendations, right that come top of mind, and that can be like you said, positive or negative, right?
Generally, those are sort of from strangers, right? People you don’t immediately know. Sometimes they’re right, you know, just Mary or John, and you would never know him.
Right. So I think the other ones would be like, frequently bought together again, other things like that are on the website that sort of on an E commerce platform, push you towards something, if you bought this, you might want to add this to the shopping basket.
Jon Carmine 1:06
Amazon esque stuff .
David Waterman 1:08
Yeah, that will comes top of mind. I think another good one would be like, they call them action feeds, you know, Hey, someone else in Seattle just bought this, you know, when you’d like to buy it, too. So
Jon Carmine 1:22
That’s what that’s called an action feed.
David Waterman 1:25
It’s the term I’ve sort of, you know, found.
Jon Carmine 1:28
Yeah, I know, I see them pop up.
David Waterman 1:30
Pop ups usually right, or jet in? And, yeah, overlay. Those can be they can go to the part of the mind that says, ooh, I want to be part of the pack. Â You know, or somebody else cooler than me is doing this. I need to do it. I imagine it works pretty well. Psychologically.
Jon Carmine 1:50
Right. Then the did you say suggested products? Did you lump that with?
David Waterman 1:54
That was another one. Yep. suggested. I would think a good one. We’ve referrals. You know, people don’t think yeah, we’re recommending now that’s not a stranger. Now, that’s the opposite.
That’s an influencer? Maybe people you know, could be, you know, really well, or just know, in your same circle. I think influencers really fall into that line.
Right. What is their you compare products together? Isn’t that like, you know, unbundled? Wow. Right? Well, with that, right? So if you’re, you know, wine with this cheese, or that kind of thing, you know, yeah, um, so pairings would be another recommendation it’s may have, it could be a bundle, too. You know, you might combine things that do and then say, well, what don’t buy the single product, buy the bundle? Right? I’ve seen that.
Jon Carmine 2:45
All of it sort of stems from that human longing of like, wanting to be a part of something you kind of mentioned that a little bit ago. Yeah, sort of all flows into that, that area of the mind where you want to be in the pack?
David Waterman 3:00
Sort of, yeah, it tugs on those areas of the mind where you want to jump be, we’re all family creatures, right? You know, what I mean? are, you know, the human species wants to be part of the family. So I think it tugs on that.
I did a little research before we, you know, for today’s show is that 91% of people claim they either actively or, you know, occasionally look at reviews, which is pretty high. You know, that means there’s only a small sliver people who, for whatever reason, don’t or didn’t in the survey, say that they do. And 84%, I think was the number of people who, who trust, there was a trust factor. So there’s a high trust factor. Now, I would be not one of that group.
Jon Carmine 3:37
Right. It’s just so interesting to hear some of these reviews, like we know, that they’re not, they’re not always ethical, like they’re not, they’re more. Okay, well,
David Waterman 3:48
Totally, you know, you’ve heard all the stories on Amazon, because they get all the politics, but think of it just the average website where nobody’s looking. I mean, you know, so something like 60% 70, another 70%, you know, after reading one, or to form an opinion, and it doesn’t, they don’t have to read up to like six reviews, and then they kind of have it.
You know, my case I’m looking for now, I’m like a lot of other people, I’m looking for the negative view, a lot of the positive but because I’m looking for some information, you know, and I think a lot of people do that, too. And I weigh two or three negative with two or three positive, so I guess it’s six. And I don’t read more than that.
Jon Carmine 4:26
I mean, see, I don’t always give the positives as much weight. No, neither as a negative, like, if I see a negative one, and it’s negative to the feature that like I’m most want that product for boom, I’m out like I’m out the door. I’m like, Yeah, so I think one of the reasons why reviews are so important is because they sort of tell a story, that individual customer story, and they kind of allow you is this is yourself in that world, and it’s easy to digest. because it’s a another human to another human.
It’s very relatable usually, and it’s written in plain language. So they can envision themselves with that product and the work, you know, the work is done, the human connection is made. And we know all the studies and research into human connection is much more persuasive than traditional marketing endeavors.
David Waterman 5:24
That’s correct. I mean, do you think about it in normal terms? Most of what you do you want to get your house fixed, you ask people for recommendations?
Right. You know, I mean, those are reviews, right? What was your review of getting your windows done? And that’s, there’s plenty of I didn’t even bother to look at statistics for that. But I think we’re, we’re heavily you know, rarely do we pick up the phone book and randomly search for strangers, you know, unless you’re desperate, right, you know, there you go. First of all, you wouldn’t have that anymore.
But you know, trying to say, you know, in some strange directory out and try to start finding somebody. That’s why like Angie’s List probably is fairly popular that they’re, they’re vetted. Right? A little bit. Right. But yeah, so reviews, I think, are a whole, you know, there’s positive and negative. And I think we, you know, like I said, we touched on them, I don’t think that without the negative they’d work if they were all positive all the time, you know, I’ve been a little variation, because nothing’s perfect. And even when I, you know, talk to customers and clients of mine, I tell them, you got to put a few negatives in there. You can’t just, you know, whitewash
Jon Carmine 6:26
You can’t hide all your reviews. That’s correct. What about the word of mouth piece? Because I think you were you’re mentioning that to me before about how this? The reviews are also that word of mouth thing?
David Waterman 6:39
Yeah, I mean, oh, like, You mean, like, in the sense of communicating with the person, you know, the words?
Jon Carmine 6:47
Well, like the word of mouth isn’t like, it’s like hearing it from a friend, even though it’s not hearing it from a friend.
David Waterman 6:54
Right? Well, I think we’re back to that whole. Yeah. That’s why I think reviews are very strong nowadays. But they’re, you know, I have a feeling they’re going to wane over time, because of the misuse and abuse and you know, things of that nature,
Jon Carmine 7:09
They find a great way to validate them. And there’s a couple of companies out there now that like, are trying to make a repository of reviews, for different types of platforms, specifically in the software world, I think to help make that a little bit easier, and obviously make a ton of advertising money, but
David Waterman 7:28
They have tightened it right, you have to at least have bought the product, etc, etc. But I
Jon Carmine 7:32
Actually using AI now and machine learning to not only filter that content for fake versus real, but also the psychological intent of the reviewer. So analyzing reviews, hundreds of them, tons of them, right, and then reporting back buckets, and almost categorizing a classification system to give a tone or nuance of a product based off of its reviews can be very helpful. marketing departments and for product development, all of that,
David Waterman 8:10
Tone has a lot to do with it. Right? I mean, are you mad as hell even or happy? Or why are you happy? Or right? I guess? I mean, you
Jon Carmine 8:19
If you get it reports back that firm is one product, you keep having a high percentage of the the feeling of confusion. Okay, well, then, maybe you’re not, you’re not explaining that product well, to the customer ahead of time. And so there’s, I mean, there’s so much here,
David Waterman 8:35
That would show up in the total, you know, the conversation, it would show up in the words they used on that review, I get it.
Jon Carmine 8:42
Yeah. And it’s cool because the the AI is analyzing, like, the way the arrangement of the word, it’s like passive voice not like all of that and kind of putting it together and you’ve had some experience you’ve been testing out different kinds of soggy
David Waterman 8:57
Dabbling in that area. Yeah. That’s why it’s kind of fascinating. Yeah. The way you use the words and how you use them, and what words you choose, are, the AI can really analyze that and then I’m using it to help recommend better words or review my you know. My posts or whenever I have headlines, headlines, you can even email you know, you can, every time you communicate. It kind of analyzes you know, did you really mean to be negative there.
You were like, You really didn’t, but you pick the wrong set of words, that it could come across negative or the personal read it in a positive way, but in the back of their mind, it’s negative, right? You know what I mean? It’s maybe you weren’t meaning what you’re, it’s, it’s mostly I use it to feed out negativity, but to be honest with you, but then I don’t I accidentally add in there. Because, you know, I’m an angry person.
Jon Carmine 9:49
I mean, everybody, it’s the nuance, but then that’s like, what makes communication so interesting, and even over hundreds and hundreds of years. I mean, language is ever evolving and always changing. So, you know, you you’re writing a headline for somebody who’s in a different generation or even 10 years behind you that that so much can change even by region and location. And as we become global, it’s even more important to have sort of a baseline.
David Waterman 10:16
That’s so true. I write this. Now, in this sense, these AI algorithms that I’m using don’t take in language, you know what, I would think that would be the next step, right? You know, different, you know, global language regions, or like, tell me don’t use that word. That means, you know, it translates poorly, right? That’s what I’m trying to say. Oh, yeah. Yeah. Um,
Jon Carmine 10:40
Let’s go into some of these kind of been talking about like reviews. We mentioned some recommendations and, and the best recommendations and how reviews are recommendations? What are some of the different types of recommendation engines being used today? And I know one of the ones I can start off with the one that we kind of started .
David Waterman 11:06
yeah, go ahead. You mean, like, like, are you thinking of Amazon? Cuz that’s the number one.
Jon Carmine 11:11
Yeah, well, so I was thinking the, the frequently bought together. I mean, that’s what we call it. They go by many, many different names. I mean, what else works well, with maybe bought together?
David Waterman 11:29
Yeah. Kind of add ons. But you know, Amazon’s the biggest elephant in the room here. So yeah, that’s frequently bought together is the main stain. That’s the one that they use for their, their engines, right. That’s their recommendation engine.
Jon Carmine 11:45
They call something different now. They just changed it. Oh, yeah. They did. They just changed the other day. I’ll look it up. I’ll look it up. While we’re. And by the other day, I mean, it could have been like a week ago. Yeah, we know. Yeah. But I checked the other day. I was like, wait, they changed their wording. But anyway, keep going.
David Waterman 12:01
So there’s like cart upsells. Right. Okay, there’s some, you know, upsells, cross sells, those are typical. Right, they would be recommendations.
I think those are just come to mind. Product upsells. What where are you? Well, I already went over that we product add ons, we think we already talked about that.
Right? You, you know you need this component or piece, you know, buy this and get the extension cord or something. And the ones I mentioned earlier, the action feeds. Right. Somebody else just right.
Frequently Bought Together
Jon Carmine 12:31
So Amazon calls it, buy it with, now? Oh, is that what it is? Okay, buy? Yeah, because it’s like before it was frequently bought together, because it was only based off of like that history. But as time went on, now, they incorporate so much into that algorithm that delivers new that not just order history, that they’re like buy it with it works with works well with, right.
David Waterman 12:54
Okay, I could see that, although I think I think of it still is things that you would buy in combination, or in a bundle, right. There’s some funny stuff out there on the internet. And we will I’ll come back to that. But you know, Amazon has recommended all kinds of crazy things out there. I think they they’ve been well known for that. Maybe that’s why they changed the name, because some of the bad press they got over.
Jon Carmine 13:17
Yeah. Give me an example. Weren’t you saying like, people were like going to jail or like,
David Waterman 13:22
I was looking around for some statistics. And yeah, there was people making bombs in England. And you know, they were, you know, it was recommending bomb components to put together and things of that nature and right. Cyanide, some, you know, some things, some chemicals, I imagined that that make, right, but if you put this list together, you make some cyanide or something to kill somebody. I mean, so they, they I think they just let it go. And they didn’t really have any, they didn’t realize that it would put together things that yet people are building bombs, and now that they’re just recommending it to other people.
Some of it was probably harmless. I mean, I’m not picking on them. But that’s one reason why maybe they changed the name. But those those have been, I think their AI is much better now. I don’t they probably pull some of that stuff out. It seems like that was a little bit a few years ago. But Netflix is another one. I think that comes to mind. All right. Aren’t they big in that area of recommending which shows to watch and things? Yeah,
Jon Carmine 14:21
I think Netflix and Amazon, maybe before we get into the details of each one, the recommendation engine, like that’s kind of what it that’s really kind of the term what it’s called now it sort of is this big umbrella of phrase, it like a phrase that fits a whole bunch of different stuff. It doesn’t have to be just like, you know.
David Waterman 14:42
It’s because they’re completely, different, right? Netflix and Amazon are completely different algorithms.
Jon Carmine 14:48
I’m sure there’s hundreds of algorithms that make it up. I think what’s interesting about Netflix is they will Netflix is driving you to watch cheaper content for them to produce and or that what they purchased.
So they constantly drive you deep, deep deep into their, their library of the stuff that nobody wants to watch on the top level that’s popular to decrease their, their spend. And they’ve actually been very successful. That’s why they have such a large amount of money to spend on new content is because they forced, they push people down. So their rights and royalties and views are lower.
David Waterman 15:32
It makes sense. I mean, how many times have you watched the series and realize it was filmed, you know, 5-10 years ago, and it ended on, you know, fifth series three, and you’re like, wow, you know, I’ve started Googling them ahead of time to feel like figure out, are they new? And are they gonna keep going? Or do they only have a finite number? You know?
Jon Carmine 15:51
Yeah. And they also sandbag them for what journalism they call it sandbagging. Like, I don’t know, if they don’t know if that’s a common term, but sandbagging a story for like a rainy day, like when you don’t have a story, because it’s timeless forever.
You know, they’re sort of saving the series and stuff, or breathing real new life back into life. Right, moderately successful on cable TV, but the good part about what Netflix can do is like, they can target who they show that content to, whereas like cable networks, 10 years ago, it plays eight o’clock, right across the country that you get.
David Waterman 16:30
But speaking of targeting, I think it’s kind of comical to the, you know, if my son goes in there, and he watches under My Account, some, you know, silly teen movie, you know, now I’m getting recommended nothing but teen movies, you know, for another month until it resets itself.
Jon Carmine 16:46
That’s why they have profiles.
David Waterman 16:52
Right, they have profiles, but if you don’t adhere to them, it screws them up.
Jon Carmine 16:56
For sure. And we’re going to talk a little bit later, we talked about the future of where this stuff is going about different models, and how a world would look, if you had more control over what was recommended to you.
David Waterman 17:10
It’s a great point, right? I think that’s where we’re heading right into. But let’s let’s stay with the topics anyway, what is what is frequently bought together? What is the definition of it?
Jon Carmine 17:21
Yeah, and how it kind of works in any commerce. So there’s a couple different ways that that it can be done. And let’s use a very specific examples of a shopping store and E commerce platform. It, there’s kind of a couple different versions of it.
You can get these plugins or extensions that work with your site, and they will generate this content based off of a couple different things. So the first one, I entered into three kind of categories, manual static, like dynamic, and manual is the customer, the store owner, not the customer, the store owner has to go into each product and generate all of the content, they have to say, this product goes this one, this product goes with this one. It’s tedious. It’s not changing. It’s not updating with the times. It’s not updating with product lines or data. It’s just, it’s just sitting there.
Another one is static, that kind of does a little bit where it sort of is generated by the by some sort of logic, usually, it’s not great. These are kind of the stock ones that WooCommerce has.
David Waterman 18:36
WooCommerce some other ones Shopify, yeah. Right. They’re kind of like standard now, right? I think, but they’re not really that good. They’re just sort of, like you said, they, they, they do the job, but they’re not not all that.
Jon Carmine 18:52
Right. Right. And, and a lot of the times, it’s because they’re use a lot of resources, they’re processor intense, and the way WooCommerce and some of these platforms work, they would run on the same server at the store is running on which can cause other problems. or sort of a slew of, you know, they’re not really set up to kind of store this data in a, you know, an efficient way, per say.
There are ways to do it, of course, but true. They’re not set up.
David Waterman 19:25
I mean, they’re not you know, even for the amount of resources they consume, they really aren’t really doing a whole lot right I mean, they only have right data set to work with and they only right
Jon Carmine 19:35
yeah, and they have some weights too, that they can apply and some some sliders to set some probability and chance there. There’s a couple of that but there isn’t really that like learning and improving based off of
David Waterman 19:45
that’s what I mean, right? It’s gonna just stick with it over.
Jon Carmine 19:50
I think the the end goal is the smart sort of dynamic, frequently bought together or recommendations. It doesn’t Just have to be that that is not only looking at past order data is also looking at geography. Where is that person located? Seasonal time of year, there’s so many different things. And then also the changes of time maybe because of a pandemic or something you’re selling more of another product will let How can that? How can that be put in? Into the recommendations to kind of long better?
David Waterman 20:28
Right? Yeah. No, I think you’re right. I mean, I could see it. I’ve seen a lot of poorly done, frequently bought together, but I think those are, as you noted, the standard, you know, built in, because, you know, not all these sites are, you know, you know, WooCommerce, or Shopify, there’s a lot of, you know, older shopping carts out there. And then I think I’ve run into a few where they’ve given me very bad recommendations, you know,
Oh, yeah. So I have to think that they were more of the plugin type or the, you know, the built in one, and then maybe the store owner didn’t have the time and energy or whatever reason didn’t tend to fine tune it even. Because I think, you know, I can gardening and stuff like that.
That’s a really good example where, where I’ve noticed it, because there’s companion plants and things you shouldn’t put together and whatnot, and it’s just recommending things all over the board, you know, things that would not work, you know, insecticides and things.
I’m like, Really, you’re recommending that for this? And I’m realized that no, it’s not there. Nobody put any thought into it, you know? So I see that a lot in my, you know, it’s, you know, yeah, but, and those sites are all over the place, they’re all different shopping carts, you know,
Jon Carmine 21:43
And something interesting that, that we’ve kind of been working on with our frequently bought together. Ours lives in the cloud. It’s completely external from the site. In fact, like, there’s one line of code that goes in the head of your site, and it, it sort of trickles through all the different products and shows up and disappears based off of based off that one line of code. Um, but what’s, what’s sort of cool about ours is it kind of starts at a certain level, and then we can throttle it and change it, it’s kind of meant to grow.
So initially, you might look at some of these recommendations and be like, that’s sort of odd recommendation. But what the system’s kind of doing is putting that information out there, because for some reason, it has found a correlation when it’s running through the algorithm. And it will put that there. And some people are like, Why, but for some reason it works. So then, now tracking that information, and then having it changed dynamically.
David Waterman 22:48
But isn’t that the essence of why you want some kind of machine learning? Because it’s going to put things out there that you wouldn’t have thought to try? And, and it will find new, right? I mean, that’s the idea behind it.
Yes. If we, if we have our, you know, I’ve set up the whole store, well, then it’s my personal my personal recommendations, but I would then never change it, because that’s my opinion, right? And I would have no, I wouldn’t take chances or do other things.
Jon Carmine 23:16
I think there’s a misconception sometimes with store owners. They’re like, Well, why would that be purchased with that? And we’re like, we’ll ask your customer because they did like you they are. That’s what they did, like. So I think that it’s interesting to see .
David Waterman 23:34
I think you have to do that you need a machine that’s learning not necessarily and trying new things, not some static and pre, you know, predefined list of things that you think it should be right.
Yeah, I agree with that. That’s probably a huge problem in the industry, you know, people have a bias, good or bad. And it’s just going to show up everywhere. Yeah. You would never do that. Well, look at all your customers who are doing that. What? I get it. Um, well, that’s interesting, though. So is is so it’s smart. And it’s changing. Right, John? I mean, it will. Yeah. So if it sees something, it’ll adjust over time, and like, if there was like, like, in the pandemic, when there’s a run on something, i it’ll recognize things that are, you know, inventory could do anything. Really, I imagine that.
Jon Carmine 24:22
Yeah, ours updates every 24 hours. And we’re pushing it toward being even more.
David Waterman 24:29
But yeah, that’s a resource intent. But you’re saying you’re taking that data off? Right?
Jon Carmine 24:35
All of our AI components live in, in the cloud?
David Waterman 24:39
Isn’t that the new way of doing things? I guess?
Jon Carmine 24:42
Yeah I mean, it’s we’ve put a ton of time and energy and, and effort into developing a serverless sort of architecture that runs our platform in the cloud, and then we run the AI components pretty much serverless.
David Waterman 25:02
I mean, I think you’re following a path, you might be leading the pack. But that’s where everybody’s gonna they can’t do all this crunching on the same web server.
Jon Carmine 25:10
Truly. Yeah, I mean, it’s, it’s there, there are just too many variables and how it, it will spike up to do the data, and then it sort of sits there.
David Waterman 25:19
So I I mean obviously, throwing more stuff into it right throw social media to it, throw weather into it. I mean, I can go into, you know. Go back to the gardening one, right? What’s the time of the season and what you’re recommending me? You know, they do impact it’s too late to plant that where you live? You know, I mean, why would you recommend it? But, but I mean, the more you throw into it, right? Isn’t it going to take more resources and more and more research? If it’s staying on that server? I have to believe that that’s gonna impact things? Oh, yeah.
Jon Carmine 25:50
I mean, a lot of these servers just can’t withstand that they’re not designed to, because a lot of the components that protect the store from like, attacks or protected from crashing, are this influx of, of data and this influx of, you know, content information. So we would obviously have to be written for the architecture, I think that in a lot of these hosting environments, that just, there’s too many of them to sort of make a good call on which way to go.
So having it live externally, for us, was a way that we could preserve the consistency of the product and also not weigh your site down. You know, when you look at a server log, or when you look at a record of what’s heavy on your site, you’re not gonna see one of our products on there because it’s not.
David Waterman 26:42
You can’t control that with your, you know, your customers or WooCommerce, right. So they have, sometimes they have their own hosting, you can’t control, you know, their server anyway.
Jon Carmine 26:53
A lot of our stuff is also we do a lot of WooCommerce work. A lot of our stuff, though, is sort of the E commerce platform agnostic in the sense of it can really live anywhere. So yeah, and we have some some plans to expand it out down the road. Shopify? Oh, yeah, definitely. Yeah. Cool. Well, that’s cool. That’s what makes I have a note here about, you know, what makes a good recommendation and what makes a bad recommendation.
David Waterman 27:26
Okay, a good to come up with a lot of bad ones. What’s good one would be? Well, a good one would be a lot of combination of things. Right? Would it be, you know, not not relying on just order data? Or what I ordered? Right would be?
Also, you know, I think, well, here’s an example. So if it’s constantly telling me this the most popular item, then I’m always going to get fed that that information, right. And so I would say,
Well, that sounds like a positive. And actually, over time, you know, I’m not getting to try anything new. Right? You know, it’s keeping me in. The other hand, I would say, in some cases, I would want that, like, if I was, you know, would recommend well, you last time you bought this particular product, right? Or toilet paper, right? Not sure. You know, don’t forget it right? And be like, Yeah, cuz that’s my favorite brand. Right? I mean, there’s probably, how do you?
How does it know when to give you, you know, get out of your comfort zone a little bit, I can recommend some other things that you wouldn’t have thought of.
Jon Carmine 28:31
When to diversify your palette, and have you move on to another another product line that you have? It’s the great, right. And they and they do that kind of through a different when I say they, I mean, the algorithms that are kind of styles of it. There’s a kind of a couple main main three right now. I mean, there’s tons of them, but collaborative filtering, content based filtering, hybrid recommendation system to sort of, kind of do a couple these different different pieces and cool them pull them together.
David Waterman 29:09
That’s kind of what I’m envisioning. Right. Yes. Model One. Right. Yeah. I mean, well, you know, I’m a big fan of, you know, I hate to keep going back to but if I could just ask one or two questions, you know, I think sometimes these you know, the algorithms are doing way too much work trying to guess it.
But if they just asked me a question, you know, for example, you know, something pertinent, then I think they could get much closer to it as well.
I think that just hasn’t that interactive with the AI is really there yet with the, you know, but if you had virtual assistants or you had, you know, someone helping you shop online, or, you know, VR, as we’ve talked about before, you know, in that case, when it says well, what are you looking for today, or in this case, you know, are you going to use this, you know, in the next week or two, you might be a different answer right then if you If I had said to you just show me, you know, you would pick something that based on either my past history or some algorithm, and I think you hit me.
Jon Carmine 30:07
Yeah. Oh, a lot of times I think customers and marketing, I say customers, I’m in like our clients, they’ll say, you know, they’re trying to categorize people automatically and kind of dropped him in buckets. But I think we raise a good poin. Why don’t you ask them? Is it that? Is it so bad? Is it so bad? Like sometimes, yes, you don’t? You know, you don’t want to but if somebody is willing to give you that information? Absolutely. Why not? Why not take it?
David Waterman 30:38
I know, I, there seems to be this is the probably the next revolution in my mind, there’s going to be where there’s slightly some interaction with, and it will be combined with AI, because there’s no way a human can keep up with it.
But it will every time you come to a site or you come to this, you know, the market to the brand, however you want to say it, you know, it’ll ask you something or interact with you in such a way that back to your models, right? I mean, maybe that’s how it builds a model of you.
Yeah, I think trying to guess and scrape the internet for my social media content, you know, is just not a good idea. It’s privacy reasons. And I Yeah, we’re gonna be right, you need to ask.
Jon Carmine 31:18
It’s gonna come to a place where we need some sort of broker, sort of that acts like a bank. And you know, that they the currency is, is money right? In the bank, in this instance, the data models of you. That’s the currency. So if we can make that a standard, and then you can also say, I want to, I have private pieces of my model, and I have public facing pieces. When I go to shop online, when I go to a store, I pass off like a handshake to that store, what my models are.
David Waterman 31:53
Like your cookie. Yeah, I think liquefying it. But yeah, like, here’s my cookie, No, give me a cookie, here’s my cookie, it tells you everything about me, and that I want to share, so that it can help you guide me to the right product, but I’m not giving you everything I’m giving you what I publicly want to share kind of like your profile on social media, you get to pick and choose, right?
Jon Carmine 32:15
Yeah, and there’s also a private component to that the models on the back end, and that the broker, right can sort of act a little bit on your behalf. And even though it’s keeping some information private, when information comes in, it can also give other information without giving all of your private stuff away, you know, we can say that you like this, or you like this without giving all of that background and regulation.
David Waterman 32:41
Oh, I see as a broker. So it’d be sort of like the website or the E commerce or where, however, would ask it a question, and it would still keep your info private, but you may be answered in a certain way that would be helpful, but without letting loose the secret,
Jon Carmine 32:56
Right. That’s relevant to what they need it for.
David Waterman 32:59
That sounds really well, that does sound like where we have to go. Because I think just I’m trying to scoop up every piece of information and big data on you know, is, it’s not gonna really, it’s not who I am today is what I was yesterday. And I don’t think you know, by the time they, you know, calculate it all down. I think some countries are trying to do stuff like that, you know, so they can monitor.
Jon Carmine 33:23
I’m sure it could very well be a government agency.
David Waterman 33:28
But I mean, for my shopping experience, you know, and I don’t know, I think I would rather, like you said, have a model. I would like love to subscribe to that idea. Are you building that?
Jon Carmine 33:40
Is that right? Please? Yeah, wouldn’t that be awesome. And contextual information is so important to knowing all of these different things together and how they connect and how they relate, and how much they weight is important. All the features of a person, all the features of the things that you like, siloed individually, they’re less valuable, but together with contextual perspective, you gain so much more.
David Waterman 34:08
Absolutely. Understanding of it. And I think that’s why I’m, I mean, to bring it all back. I mean, that’s why I think the frequently bought together or how it recommendation engine, however we want to go backwards, even is, is, you know, got has to improve and it has to have some AI in it, I think and that’s where, you know, to be any to be effective. Really, right.
Jon Carmine 34:28
Oh, totally. Yeah. I mean, and then.
David Waterman 34:32
I just think that’s just like the baseline. Now it’s, yeah, that’s what it is. That’s 2022 right now, you know, you can the review stuff, I think they could use the AI in to helping, you know, go through the review.
What are the problems with reviews, right is and I hear this all the time is I don’t have time to go and approve them all. I don’t have time to review them all.
You know, it’s a time consuming so they outsource that but like if you If the outsource company had some, some really good machine learning in there, then they could really, maybe do a lot better with the reviews, right?
Jon Carmine 35:08
Yeah, there’s some of them that are popping up now especially for like explicit content to try to like filter them out ahead of time.
But being able to do something valuable with those reviews, and act on them to is even more important, you know, maybe you’re able to get that review, help that customer automatically, and then have them automatically send them another one to rewrite a review. And you can sort of fix that problem. There’s some of that stuff now. But it’s not always done in the smartest of ways.
David Waterman 35:41
We could do a whole episode on reviews and how to, you know, totally, yeah, you’re managing your neg negative reviews is one thing I consult on all the time. It’s not always the positive people will talk about the positive, but it’s really managing the negative. I mean, to be honest,
Jon Carmine 35:53
I think so, on closing this one out, I think something I want to I want to point out is that these the recommendation engines and starting to develop these recommendation libraries, and this data on your product base, it doesn’t always have to just be used for product recommendations. T
here’s a whole world and of things that you can do with that information once you have it. And kind of getting it and having it like frequently bought together, this is a really good place to start and kind of get in on that early so that you’re able to reap the rewards of that down the road.
David Waterman 36:35
I know I hear you it’s like preaching to the choir. It’s sort of like we were saying with if you want to start sending SMS messages, right, you need to start asking for phone numbers. You know what I mean? Start asking now. Yeah, now’s the time to get into some some machine learning and AI and start that process because there’s probably lots of areas you can use it.
