← The Agile Brand with Greg Kihlström®: Expert Mode Marketing Technology, AI, & CX
The Agile Brand with Greg Kihlström®: Expert Mode Marketing Technology, AI, & CX
#699: Being real about synthetic personas with Narek Vandanyan, Prelaunch
00:24:29
Loading summary
Transcript71 lines
- [00:00]
Greg Kilstrom
What if your customer experience and culture strategy didn't just create value but actually funded your growth? I'm inviting you to join me in a powerful mastermind group I'm co leading with Journey Spark Consulting. It's called the CX and Culture Connection, creating a self funding growth flywheel and it brings together CX and culture leaders who want to drive change, build stronger internal collaboration and actually reinvest efficiency savings into growth driving initiatives. This isn't just theory. You'll get monthly virtual sessions, one on one coaching, quarterly workshops and access to the Value Accelerator tools focused on strategy, cultural alignment and voice of customer. Plus you'll connect with leaders across functions from marketing to product to ops who are facing the same challenges and pushing towards the same goals. Up to four team members can participate from your organization so you're building alignment while building momentum. Want in, learn more and sign up at journeysparkconsulting.com that's journeysparkconsulting.com mastermind let's turn your CX investment into a growth engine. The Agile Brand welcome to Season seven of the Agile Brand where we discuss the trends and topics marketing leaders need to know. Stay curious, stay agile and join the top enterprise brands and Martech platforms as we explore marketing technology, AI, e commerce and whatever's next for the Omnichannel customer experience. Together we'll discover what it takes to create an agile brand built for today and tomorrow and built for customers, employees and continued business growth. I'm your host Greg Kilstrom, advising Fortune 1000 brands on martech, AI and marketing operations. The Agile Brand podcast is brought to you by Tech Systems, an industry leader in full stack technology services, talent services and real world application. For more information, go to teksystems.com to make sure you always get the latest episodes, please hit subscribe on the app you listen to podcasts on and leave us a rating so others can find us as well. Now onto the show. Marketers spend a fortune trying to understand their audience, conducting surveys, focus groups and market research.
- [02:20]
Narek Vardhanyan
But what if they could just ask.
- [02:22]
Greg Kilstrom
Their ideal customer directly? With synthetic Personas? That's now possible. But how reliable are they and how.
- [02:29]
Narek Vardhanyan
Far can we really take this technology?
- [02:31]
Greg Kilstrom
Joining me today is Narek vardhanyan, CEO of Prelaunch.com, a platform focused on helping businesses validate their ideas. Before launching, prelaunch.com recently introduced customer Persona, a tool that allows marketers to chat with AI generated Personas built on real audience data. Narek, welcome to the show.
- [02:50]
Narek Vardhanyan
Hey Greg, thanks for having me yeah, yeah.
- [02:53]
Greg Kilstrom
Looking forward to talking about this topic with you.
- [02:55]
Narek Vardhanyan
Definitely. It's something top of mind for me. I know, and many others. Before we do though, why don't you start with giving a little background on yourself and your role@prislaunch.com.
- [03:05]
Yeah, sure. So I've started a marketing agency 10 years ago, it's called TCF. It was really successful. We helped many companies launch their innovations into the markets. Like about more than 1,000 products helped them do more than millions of dollars of sales. So we're always into the innovations and we're trying to help companies to innovate better. And we're seeing that many of those products were failing. So this what led us kind of to start pre launch, which is a really interesting idea and really transformative concept on how you validate specifically innovations like products that hasn't been into the market. And the concept is actually really simple, which is the foundation of the platform itself. On one hand we have brands who are placing their early stage concepts on the platform and then on the other side, potential customers place small deposits like $1, $5, $20, depending on the price of the products. And then we understand the real buying intent, we take the data, analyze it, compare with our benchmarks, and basically try to estimate very accurately the market potential of those products and. And who are the customers.
- [04:10]
Greg Kilstrom
Yeah, great, great.
- [04:11]
Narek Vardhanyan
Well, yeah, let's, let's dive in here. Definitely fascinating topic. So why don't we start by maybe even taking a step back and understanding, you know, what exactly are synthetic Personas and how do they differ from traditional market research?
- [04:27]
Yeah. So synthetic Personas are these AI generated Personas, as you said, and which basically simulate the real life scenarios and actions of people on what they would say, you would ask a person about it. And there is a lot of controversy in the industry. Like there is like pros and cons, a lot of people in the industry as are pro synthetic Personas and part of the industry is against them and we stand in between, to be honest. Right. So we see that there are many cases that they can be really helpful. Like you can ask AI and based on all the previous data, answers and everything interactions, they can provide really accurate insights on some of the topics that you're asking them.
- [05:13]
Unknown
Right.
- [05:13]
Narek Vardhanyan
And on the other hand, there can be topics which are not appropriate to use synthetic Personas for those are specifically related with innovations and something that hasn't been done before. So in these cases you better use AI, but you need to put at its core the real customers because if that hasn't been done before, then the AI probably hasn't learned really better from available information.
- [05:39]
That's interesting. So what would you say are the biggest advantages of interacting with synthetic Personas instead of relying on some of those more traditional methods like surveys, focus groups and things like that?
- [05:53]
I think the biggest advantage is the time, of course. So they are like instant 24 7. You can talk as long as you want, nobody gets tired. They are always available. So it's.
- [06:06]
You don't have to feed them, right?
- [06:08]
Exactly. Yeah. You don't need to convince them, you don't need to bribe them. So they're always available. And that's a nice thing because the industry has always been kind of struggling with this time sensitive data, especially with the focus groups, is taking really a lot of time, etc. But you need to be really cautious about it because as I said, it really depends. It really depends on the topic. It really depends on the type of the products that you're asking. And ultimately at the end of the day, it really depends on how this synthetic Persona is trained.
- [06:43]
Unknown
Right.
- [06:43]
Narek Vardhanyan
So what, what's the foundational data that is used to train this, this Persona? So it's not that much of a good or bad. It's more of a. Like what data was used to train this or that Persona. Like if you have been training this, this AI Personas on really foundational, really reliable data, then most probably the AI version of it will also be really reliable and provide very accurate answers.
- [07:09]
Unknown
Right.
- [07:10]
Narek Vardhanyan
And on the other hand, if you're like just scraping Internet, providing very generic answers, etc. And your AI Persona wouldn't provide much of the value. And again, like if this is a very traditional product, like you are just, you're producing another packaged food that's ultimately in the market that, that has been there before, or a TV or a fridge or something that people really relate to and has been there for years, in these cases the Personas will be really good and reliable. But on the other hand, if this is something that hasn't been in the market, like it's a flying car, it's a really new concept, really new gadget that people haven't seen. In this case, what we are doing, we are basically mixing real humans with AI. So we would again put the concept of those, those products on our platform and real people will come and place small deposits like $5, $10, $100.
- [08:06]
Unknown
Right.
- [08:07]
Narek Vardhanyan
To be in this wait list of. We call it basically a wait list where you would receive it. If this product is actually released in the markets.
- [08:14]
Unknown
Right.
- [08:15]
Narek Vardhanyan
And in this case you have this actual person who is in the waitlist and they really want to buy it. And then we pop up a survey and we ask them, why did you do this reservation? What's your main motivation, your incentives, et cetera, et cetera. So it ends up being really useful data and really trustful data because you're ultimately getting this data from the future customers so you can really trust them. Now what we do, we clusterize with machine learning with this data and try to find different segments of audiences, like people who have this use case, people who have that use case, and then people who have totally another use case based on their real answers.
- [08:53]
Unknown
Right.
- [08:54]
Narek Vardhanyan
And on that segment we generate synthetic Personas, customer Personas, where you can talk to people and get some of their answers, which is very, very much correlated with what actually people said. So it's kind of basically becoming a convenient way to just gathering and understanding all these insights that were coming from people. And it's less inventing something that hasn't been said actually by an actual person.
- [09:23]
Yeah, yeah. So maybe to paraphrase or summarize what you're saying. So, you know, in looking at it sounds like there's potentially two different scenarios. I mean, one is a purely like interacting with synthetic Personas. But those are most successful when there's some kind of existing data or use cases for that AI to train. I mean, I've seen this in other use cases as well, is like generative AI is incredibly helpful at kind of telling us a lot about what's already known. But it's not great at coming up with brand new, you know, novel ideas. That's where I think humans, you know, can. Can come into play. And so then that's. The first scenario is, is kind of using purely synthetic. The second one which, which you're describing is using synthetic plus humans to kind of stretch a little bit maybe to, you know, synthetic can provide some. So to use the flying car example. So we have planes and we have cars. So there's some existing just to very be very like gener generalized about it. So there's some data to train a model on. But you know, the human and their imagination is going to tie those two together. So synthetic plus human is that where. So in other words, the two use cases would be if it's a. If it's an incremental innovation, probably purely synthetic could be used. But if it's a bigger innovation, probably synthetic plus.
- [10:53]
Exactly. Yeah, that's exactly what I'm saying. And the analog was very, very similar. Like again with the LLMs, with the ChatGPT, with all other kind of foundational models. Like if that's kind of an existing information that has been in the Internet, you can ask a lot of questions and get a lot of interesting summarizations, advices and everything right? And it's really useful. But as long as it's a really novel idea that hasn't been in the Internet, you can't really get the very useful answers right or it's a very niche type of information that hasn't been shared in the Internet. You can't really get a really good data. And same with the synthetic Personas, right? If it's an incremental innovation or no innovation at all, like it's another product, you want to test your packaging, you want to test your ad creative, et cetera. And it's very kind of known industry known category that has been in the Internet over the years and a lot of data about it, you will get really good answers. As long as it's a whole new breakthrough innovation. You don't really need to rely on the synthetic Personas more more use them as a convenience which basically will summarize the existing customers answers.
- [12:08]
Unknown
Want to learn more and join the discussion About Marketing and AI? Attend a premier conference dedicated to marketing and AI. That's Meacon, the Marketing Artificial Intelligence Conference. From October 14 through 16 in Cleveland, Ohio. Meikon brings together the brightest minds and leading voices in AI. Don't miss this opportunity to connect with a dynamic community of experts, visionaries and enthusiasts. The Agile brand is proud to be the lead media sponsor of this important event. Register today@MarketingAIInstitute.com that's MarketingAI Institute.com and use the code AGILE150 for $150 off your registration fee. I can't wait to see you there.
- [13:00]
Narek Vardhanyan
So then how should someone look at.
- [13:03]
Greg Kilstrom
Like what are the measures of success here?
- [13:05]
Narek Vardhanyan
I mean obviously launching a product and having it be successful is be the ultimate measure of success. But you know, how should, how should users of synthetic data and Personas look measure is. It is reliability and even a metric, you know what, what are, how do.
- [13:22]
Greg Kilstrom
How should they look at that and evaluate platforms?
- [13:25]
Narek Vardhanyan
I think that's a really good question. And, and that's the, that's the biggest unknown probably that someone yeah might face when they're using the synthetic Personas. A lot of companies who started to actively use the, the synthetic Persona start to evaluate like they, they ask the same question to synthetic Persona and then they do the classical research and they do some other ways and then they try to see how much is the correlation, whether there's a synergy and then like how much is the overlap.
- [13:57]
Unknown
Right.
- [13:58]
Narek Vardhanyan
Basically. So a lot of companies are doing their own research to try to evaluate like how reliable is this Persona for their purposes. I think generally there is no this kind of ultimate benchmark that you can set and say this is very reliable or this is not reliable because as I said, it's really depend on the use case. In some cases it can be really highly reliable or any other cases it might hallucinate and provide not reliable answers. So you basically do your own fact checking and measurement and if you have the other ways of using it, you basically use it in parallel with synthetic Personas to see the reliability of among them.
- [14:41]
Yeah. So I mean, given that, and that totally makes sense, and in some cases I would assume with the bigger innovations, it's hypothetical, it would almost be impossible to do it. Like you could probably do a real apples to apples comparison if it was synthetic with a real product in a real situation. You could do as close as possible, an AB test, so to speak. But with some of these, it's difficult. So I guess how should companies and marketers and others be looking at these to use tools like this in the best possible way?
- [15:19]
I think it all comes down to how much data was used to train these Personas.
- [15:26]
Unknown
Right.
- [15:26]
Narek Vardhanyan
And that ultimately leads to some kind of reliability also. And I know that many big companies are currently focused on generating their own Personas. So those are companies that have pretty much 50 years of data. And that definitely can lead to very reliable kind of predictive analytics in their case. And in some case, if there's just a company that's, that's using the data which has been like not statistically significant, very, very like low amount of data, in that case, the reliability will, will suffer for sure. So I think the research on how, on what Persona to use, what method, what company to trust for, et cetera, it really depends on the details and on like what was used to train those, those AI models, AI Personas. So you need to do your own research and try to dig deeper to understand the relevancy.
- [16:19]
Unknown
Right.
- [16:19]
Narek Vardhanyan
It's not more of a, it's not that much of a good or bad, it's more of a how relevant this will be to your specific use case. Because if they haven't specific data which can be really useful for your activities, then probably they might have a really high accuracy, but on a very specific niche Types of products.
- [16:40]
Unknown
Right.
- [16:40]
Narek Vardhanyan
So which might not be useful to you. So in this case, the main advice I would give is to basically dig deeper into the methodologies, into the data, into the historical kind of availability of those data, and also the reason and the goal on why you are using it.
- [16:58]
Yeah, well, and I guess one other. I'm going to go off script a little here, but like one other thought that comes to mind. So you know, predictive analytics, certainly you mentioned like the company that has 50 years of, of, of data which can be incredibly valuable in, in some ways and yet, you know, we, we live in uncertain times and we could probably always have said that, you know, for, for decades, but you know, things seem to be volatile. Let's just say, you know, as of late is using generative AI and, and synthetic Personas plus predictive. Is that a better way? You know, because predict predictive is very reliant on what's happened in the past and the way that it's happened. Are there some benefits to some of the ways that synthetic Personas are developed that also kind of mitigate against what happened five years ago would happen tomorrow as well?
- [17:53]
I think the correlation is, is very high. Because Personas use predictive analytics. In a, in a sense they as well.
- [18:03]
Unknown
Right.
- [18:03]
Narek Vardhanyan
So because they predict based on all the previous interactions, what's the best next answer to, to provide you? So that's, that's also kind of a predictive analytics in there. So sometimes it can be really good. It can be really good if you have a really good data. And yeah, like they might have some correlations associations. And if you're training them well, you'll see and sometimes be really surprised on how they're providing you this much of accuracy. Yeah yeah. And in other cases you will see that they don't have much of specific data and that will mean that there wasn't enough data on this specific topic to train it so they could predict the next best answer for this specific question.
- [18:47]
Unknown
Right.
- [18:48]
Narek Vardhanyan
So in a sense it's very similar concept. I mean like companies are using predictive analytics in many areas and customer Personas are also basically predicting the, the answer based on all the trained data that, that they have been. They have been trained for.
- [19:05]
Yeah yeah, that makes sense.
- [19:06]
Greg Kilstrom
Well, so where do we go from here?
- [19:09]
Narek Vardhanyan
You know, what do you see? I mean, certainly, you know, your company's doing a lot of innovative things already, but you know, what's on your, your radar and where do you see synthetic Personas evolving in the next few years?
- [19:20]
Yeah, so you know, with our case, we're really focused on innovation. So there always be a human factor on pre launched and there always be humans which will kind of tell their, based on their emotional, kind of logical and other feelings, what they like, what they didn't like, etc. And as we are basing our insights on the actual behavior of people, that gives us a lot of reliability. I believe that synthetic Personas will evolve in that regard, that for different types of product they will be incredibly useful and valuable. And for some types of products like innovations, they won't be that much useful. They still will be useful, but in a sense that they will become a convenience store where you can ask, get a lot of answers, et cetera. But all those answers are backed by real people.
- [20:16]
Unknown
Right.
- [20:17]
Narek Vardhanyan
And it's not really inventing answers, but it's more of a kind of conveniently transforming and telling you what actually people said in there. And in some case it may invent answers, but it's really confident that these insights are really reflecting to the high level of trained data.
- [20:35]
Unknown
Right.
- [20:36]
Narek Vardhanyan
So I think they will continue to evolve towards that regard. They'll become more useful, more convenient. Like you'll meet your future customers and you'll get insights from your future customers instantly. And that's really great because ultimately you can have an idea and just discuss it on the way, but you need to be really careful not to just fully rely on it. And especially if you're inventing something new, you need a real human to back it up.
- [21:04]
Greg Kilstrom
Yeah. Yeah. Love it.
- [21:05]
Narek Vardhanyan
Well, Narek, thanks so much for joining today. I've got one last question for you before we wrap up. Like to ask this to everybody. What do you do to stay agile in your role and how do you find a way to do it consistently?
- [21:18]
Yeah, I mean today it's, it's all related to AI and there are a lot of fascinating and exciting opportunities with AI. So like my main focus is to use all the current innovations and everything that comes out very extensively and then try to implement it into the platform and into the vision that we're having. So that's the ultimate way to win.
- [21:41]
Greg Kilstrom
Great, Great.
- [21:42]
Narek Vardhanyan
Well, again I'd like to thank Narek.
- [21:44]
Greg Kilstrom
Vardenyan, CEO of Prelaunch.com for joining the show.
- [21:47]
Narek Vardhanyan
You can learn more About Narek and.
- [21:49]
Greg Kilstrom
Prelaunch.com by following the links in the show notes. Thanks again for listening to the Agile brand brought to you by Tech Systems. If you enjoyed the show, please take a minute to subscribe and leave us a rating so that others can find the show as well. You can access more episodes of the show@theagilebrand.com that's theagile brand.com and contact me if you're interested in consulting or advisory services or are looking for a speaker for your next event, go to www.gregkilstrom.com that's G R E G K-I H L S T R O M.com the Agile brand is produced by MissingLink, a Latina owned, strategy driven, creatively fueled production co op. From ideation to creation, they craft human connections through intelligent, engaging and informative content. Until next time, stay curious and stay agile. The Agile Brand.
- [22:53]
Unknown
Before we continue, I wanted to share a key strategic resource that a majority of the Fortune 500.
- [22:58]
Greg Kilstrom
Are already aware of.
- [23:00]
Unknown
Finding the best technology, business and talent solutions is not easy. With business demands and competitive pressures mounting, you need to be able to design, deploy and optimize your technology to provide leading customer experiences while driving business growth. Those of you that have been listening to this show for a while know that this podcast is brought to you by Tech Systems, a global provider of technology, business and talent solutions for more than 80% of the Fortune 500. Tech Systems accelerates business transformation for their customers. Whether you're looking to maximize your technology roi, drive business growth, or elevate customer experiences, Tech Systems enables enterprises to capitalize on change. Learn more@techsystems.com that's T E K systems.com now let's get back to the show.