← The Agile Brand with Greg Kihlström®: Expert Mode Marketing Technology, AI, & CX
The Agile Brand with Greg Kihlström®: Expert Mode Marketing Technology, AI, & CX
#619: Knowing which of your marketing dollars are working, with Mike True, Prescient AI
00:27:03
Loading summary
Transcript71 lines
- [00:01]
Greg Kilstrom
What's the state of your website? If you're looking for a new Digital experience platform or DXP to run your digital experiences, I have the book for you. The Agile Brand Guide to Digital Experience Platforms is part of my best selling series of MarTech books. In this book I explore and demystify DXPs and look at the roles a digital experience platform should play, the different types on the market, as well as how to initially evaluate platforms, then how to best implement a DXP once you've selected it. The book also features a forward from Rupali Jain, Chief Product Officer at leading DXP provider Optimizely, as well as several other thought leaders in the industry. Learn more and get a copy in.
- [00:39]
Mike True
Print or digital now by going to.
- [00:41]
Greg Kilstrom
The Agile Brand guide website at www.agilebrandguide.com.
- [00:48]
Unknown
The Agile.
- [00:56]
Greg Kilstrom
Welcome to season six of the Agile Brand where we discuss marketing technology and customer experience, trends, insights and ideas with enterprise and technology platform leaders. We focus on the people, processes, data and platforms that make brands successful, scalable, customer focused and sustainable. This is what makes an Agile brand. I'm your host Greg Kilstrom, advising Fortune 1000 brands on martech, marketing operations and CX, best selling author and speaker. 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 now let's get on to the show.
- [01:40]
Mike True
Are you sure you know where your marketing dollars are? Making the biggest impact? In an omnichannel world, pinpointing where your ad spend is most effective is tougher than ever.
- [01:49]
Greg Kilstrom
What if you could use AI to make it easier?
- [01:52]
Mike True
Welcome to today's episode where we're going to discuss how to optimize marketing effectiveness through marketing. Mix modeling AI and predictive analytics with Mike True, CEO and Co Founder of Preschen AI. We're going to explore how to ensure.
- [02:04]
Greg Kilstrom
Your marketing dollars are delivering maximum return.
- [02:07]
Mike True
On investment and the role that AI plays in getting it right.
- [02:10]
Greg Kilstrom
Mike, welcome to the show.
- [02:12]
Mike True
Hey, thanks for having me, Greg.
- [02:13]
Yeah, looking forward to talking about this with you.
- [02:15]
Greg Kilstrom
Before we get started though, why don't.
- [02:17]
Mike True
You give a little background on yourself and your role at Prescient AI?
- [02:21]
Yeah, so my name is Mike True. I'm the co founder CEO of Pressing AI. Been in the tech space since 2010. Jumped around from IBM Oracle, went in the mobile analytics space with a company called App Annie now Data AI and then, you know, got the itch to try to take all the Learnings over those years and met my co founder and you know, started off on this journey day to day is a variety of managing, you know, our investor relations, you know, company culture, product and you know, just making sure that everything is going in the right direction.
- [02:49]
Yeah, great. Well, yeah, so definitely looking forward to talking about a few things here with you. So I want to start with just the overall topic of just how important it is to know where your marketing dollars are going. I know CMOs and marketing teams are under increased pressure to be able to show this. You know, why is it from, from your standpoint, you know, why is it so critical for businesses to have this deep understanding of where their marketing dollars are most effective?
- [03:18]
Yeah, I think it comes down to being able to do, to do more with less Marketing organizations are under a tremendous amount of pressure to deliver. It's a highly competitive market and having a strong understanding of where to allocate their media mixes and having confidence in what is that predicted outcome going to deliver that incremental growth with their marketing budgets that they have set each year.
- [03:42]
Yeah. And so, you know, some of this might be pretty straightforward to those that are, they're doing this day in, day out. But you know, could you talk a little bit about, you know, what are the, the risks of not having clarity on where ad spend is generating returns? You know, particularly, you know, we're talking omnichannel, we're talking, you know, new channels seeming to pop up out of, out of nowhere, consumer preferences changing, you know, all that kind of stuff. You know, what do you see as the risks here?
- [04:09]
You know, it's risk. I kind of associate this like gambling, right. When you are, you're gambling with dollars, Right. Would you want to make the most educated bet that you're going to, you know, deliver the results? You know, marketing in the measurement space has just rapidly evolved, you know, over the last 10 years with the ability to, you know, scale across different channels. Traditionally you've seen a lot of brands that have, you know, really grown their business on, you know, Google and Meta you but there's a requirement now to start finding new audiences, finding new customers and that requires you to go into more top of funnel channels, things like linear CTV, podcast radio, YouTube, TikTok and having a pulse on the performance of those channels, knowing where to double down. If you don't have that type of clarity, you're going to miss opportunities and you need to keep up and it needs to be fast and you need to be continuously optimizing and so not having the right Pulse on that and confidence is a really risky know exposure for any business.
- [05:03]
Yeah. And so I think that's a great segue here of talking about a way to do this. So marketing mix modeling with the goal of optimizing that return on ad spend or roas. So can you talk a little bit about that's, that's a lot of acronyms there. Mmm, roas. You know, can you explain a little bit for those less familiar, you know, what is marketing mix modeling and you know, why is it so important for optimizing return on ad spend?
- [05:30]
Yeah, the notion of an MMM or market mix modeling was really defined to help allocate budgets across and within various channels. And it's using historical learnings.
- [05:41]
Unknown
Right.
- [05:41]
Mike True
So it's looking at all the historical data. The inputs are things like sales, the advertising spend, the platform reported metrics, seasonality, component, word of mouth. And it's really trying to make sense of the impact or the statistical relationship between, you know, your spend in whatever KPI you're looking to measure and optimize against doesn't require any pixels or user level tracking where you would mostly see with like a multi touch attribution. It's trying to look at the holistic picture and trying to figure out like where should you make that next bet? And then what was the predicted outcome of deploying budgets within those channels and campaigns.
- [06:18]
Got it. So then, you know, I want to talk a little bit about how you, how you approach MMM at prescient AI. So could you talk a little bit about that and you know, how have you seen it transform some of the marketing strategies for your customers?
- [06:35]
This is my favorite topic and so this specific question actually. So the MMMs historically in the past kind of went full circle, right. In the 60s they came out and you have called a retailer, you know, you're selling a product, a razor blade within a, a brick and mortar store. They built these models to do media planning.
- [06:58]
Unknown
Right.
- [06:58]
Mike True
And so they would run once a year. You gathered as much data as you could and then it would produce this.
- [07:03]
Unknown
Report and it would tell you, hey.
- [07:04]
Mike True
Here'S where you should, you should allocate X amount on radio, billboard, tv, catalogs, newspaper, so on. There was no online then.
- [07:14]
Unknown
Right.
- [07:14]
Mike True
So it was all offline. And it's trying to figure out, well, which one of these channels was driving and was most effective. I got really lucky when I met my co founder, Cody. I like to consider him as like a Michael Jordan of research in this space in the sense of his level of talent, but also his competitiveness and so a lot of the MMMs that you'd see evolved would leverage existing research papers and they would be enhanced. Well, Cody questioned the existing research papers and says, well, why does an MMM have to run, you know, once a year? Why does it have to run, you know, twice a year or even quarterly at a channel level? Why couldn't you make a more dynamic MMM that was more retail, was more like real time and more granular? And mmm, it runs every single day at the campaign level. It's Omnichannel. And so, you know, we've helped brands like Hexclad, Good American Jones Road Beauty that are really omnichannel brands. Having a insight into how is my paid spend on Connected TV driving sales to their D2C store, to their Amazon store. And now we're offering a retail model as well. So you're, you're allowing to use an MMM for more dynamic optimizations. And so we've built what we call halo effects. And essentially a halo effect really helps brands scale top of funnel, where we're essentially taking credit from the bottom of funnel. So last click and we're redistributing that up to the top of the funnel where we're saying, hey, we have the highest level of confidence that the awareness from this TV campaign or this TikTok campaign that was not clicked actually was what was the driver downstream for somebody that might have went to an Amazon store or went directly to their website and converted from there. And so we wrote our own research. It's fast, it's granular, it's omnichannel and it can be used to do really dynamic media optimizations versus a traditional mmm where you're thinking of as more like long term budget planning has been incredibly effective for, you know, for the consumer and retail ecosystem.
- [09:09]
Well, yeah, and it, it also just seems like it, it's. That way of thinking is built for the way that marketing is being done more and more often. I mean you mentioned like TikTok. So you know, is the same TikTok video going to be playing for, you know, three months at a time? You know, so like doing, mmm, quarter over quarter even gonna work in a TikTok environment versus, you know, you're talking, if you're talking daily. I mean that's, that's about probably the attention span of, of most TikTok videos. Right?
- [09:41]
You, you hit that spot on the, the. We've seen it. What's been very interesting for us is like, you know, more of the enterprise brands, they, they really have been Honing in on, you know, a lot on, on Google, on branded search. And they just want a little bit of confidence. They know they want to start scaling into these new channels, but they just wanted a little bit more confidence to do that. We've seen some more of the, let's say the modern day brands like the, the hexclads of the world, which are growing incredibly fast, but they've taken this approach of adopting multiple forms of measurement. So you think of an MMM that's going to tell them how to reallocate their budget, do you know, more dynamic optimizations and then triangulating with other forms of measurement like incrementality testing to do holdouts to validate some of those. And so we're seeing MMMs just take this really new role inside of, inside of organizations and adopting it in a way where they can be more proactive and reactive at the same time. Using these models alongside other forms of measurement.
- [10:40]
Yeah, yeah, well, and of course, you know, attribution, nothing's easy, but some things are a little more difficult than others.
- [10:47]
Greg Kilstrom
Right.
- [10:48]
Mike True
So you know, when we're talking about attribution in an omnichannel environment, you know, this stuff can be particularly challenging. I wonder if you could talk a little bit about that part of it and just how, you know, even how MMM plays into that. But you know, why is it particularly challenging to understand attribution and effectiveness when we're talking about, I mean, omnichannel and marketing environments?
- [11:12]
Yeah, I feel the, the core source of attribution over the last years has been through ga, through the platforms and through a multi touch attribution solution where they are using a pixel, they're very clicked based deterministic forms of measurement. And when you think of an omnichannel brand, you know, if you see an ad on, you see a YouTube ad, right, and you don't click it, but they go over to Amazon and make that purchase. Traditional forms of, of measurement make attribution incredibly challenging.
- [11:45]
Unknown
Right.
- [11:45]
Mike True
And so there's this trade off of a probabilistic versus deterministic where you can tell, be very confident with a small subset of your conversions, but you're missing the larger picture, especially for omnichannel brands with an mmm, is it not, it's not deterministic and it's probabilistic, but it's giving you that holistic picture I like to think about. Like an MTA will or multi touch will tell you a whole bunch about a little and an MMM will tell you a little about a Lot with our mmm, with the granularity and speed, we're able to tell them a lot about a lot essentially. And so I think that's been primarily challenged of attribution. And then obviously you see things with like iOS 14.5 cookie dedication, you know, ad blockers where these traditional forms of measurement were relying on these clicks and probabilistic, I mean deterministic journeys. It's just created what I like to say is like moving towards an anonymous Internet. And so I think that's why you've seen the role of MMM plus incrementality really, really coming to the forefront today.
- [12:50]
And so how should businesses be thinking that about this, this tracking? I mean certainly it's a, it's a different approach to do multi touch versus mmm. But how does a business still ensure that accurately tracking which channels are driving the most value?
- [13:07]
Yeah, I like to say there's no silver bullet with this. I, I, you know, I've heard like, hey, this, this source of measurement is my source of tr. I fundamentally disagree with that. I'm saying that the source of truth is the marketer, right?
- [13:20]
Unknown
They have a good pulse on every.
- [13:22]
Mike True
Aspect of the business associated to all their marketing spend. And there's different forms of measurement to help you triangulate with that. Where, you know, an MMM is not trying to tell you that, you know, Greg saw this ad, this ad, this ad and then converted where it's trying to take more of a holistic approach. So mmms, if you're going to look to scale into top of funnel, an MMM is perfect for you. So if you have built your business and you're heavily on meta and Google, right. You have a really good pulse for that using platform GA and an mta. But when you start going into these, you know, more view based channels, out of home channels, right? This is where an MMM should be applied when we work with our clients, right? You just don't onboard with an MMM and say, all right, click a button, we can run a simulation in 45 seconds. It's going to give them a predictive media plan compared to their existing spend and just go nuts with it.
- [14:10]
Unknown
Right.
- [14:10]
Mike True
You know, you really want to understand which channels you're looking to scale and be very methodical about, you know, how you are reallocating these budgets and not making dramatic shifts and starting to learn. Let the, let the models learn, let the AI learn. And the more you start to make these changes, the smarter the MMM gets. And you can start applying this into new strategies as you expand your channel.
- [14:31]
Mix yeah. And so you kind of just touched on this, but want to dive in a little deeper on the role of AI here because, you know, certainly marketing mixed modeling, you know, it's predates some of the recent AI hype. But as you've already shared, you're using AI and MMM in some innovative ways here. So can you talk a little bit about that? How does AI improve that process and what advantages does it offer over more traditional marketing mix modeling?
- [15:01]
Yeah. So specific to the role of AI within MMM and our AI is when you ingest all of that historical data that we talked about, those inputs, the AI is designed to learn that statistical relationship between your spend and your revenue. But there's other factors that are going to be driving revenue and conversion and performance. Things like seasonality of the business, promotional periods. What is the consideration cycle of a product? If you have, you know, a $20 price point, the consideration cycle of people seeing that and then the likelihood for them to convert and, you know, just a click and convert is higher. But what if you have a product that's, you know, $1,000 price point where there's a longer consideration cycle. And so AI is used to take all that historical data and really try to understand what those relationships are and being able to quantify that in a way that the marketers feel confident in that forms of measurement. MMMs, they do not like to see consistency. We love to see change. And so this allows them to go test new channels.
- [16:01]
Unknown
Right.
- [16:01]
Mike True
So if, you know, you start playing around with your different budgets on, across your different channels, it's going to start to learn what is that sweet spot by identifying diminishing returns or saturation plots. And what do I mean by that is if you spend $2 and you make 10 on this campaign, you can't expect to spend $2 million and make $10 million. There's going to be some sort of, some form of saturation. And so, you know, for us, as we've really approached the AI side where traditional saturation plots to find those sweet spots, we're using linear regression models. Those linear regression models, they assume that the shape of saturation is the same for every channel, the data is the same.
- [16:43]
Unknown
Right.
- [16:43]
Mike True
You have your spending impressions and revenue. It's plotted the same on an XY axis. But we've created some technology in AI where we're trying to find the shape of the data, which sometimes these saturation plots looks like camel's backs, camelbacks, but it allows us to get lost. The AI get to get very specific and very precise on that direct point of saturation across all of your campaigns. And as you start to make these changes, it learns and it only gets smarter and smarter. So it understands that really that sweet spot of where you should be spending.
- [17:14]
Yeah, yeah. And so, I mean that really kind of highlights the next topic I wanted to talk about, which is just, you know, the power of predictive analytics in general. So, you know, certainly traditional modeling is looking backwards and building on that, but I think there's a ton of power in prediction and predictive capabilities. Can you talk a little bit more about that? I know you just kind of touched on it, but can you talk a little bit more about how should businesses be looking to predictive analytics to not only maximize revenue, but also looking at things like profitability?
- [17:51]
Yeah. When you think about implementing predictive analytics specifically with an mmm, it's going to tell you based off of your current spend. Right. What do we predict that's going to happen over the next 30 days? And then you start layering in optimization models on top of that based off of those saturation plots and saying, hey, if you implement these changes, we predict with a certain level of confidence that the incremental growth compared to your existing strategy will be X. Right. So when think of predictive analytics, I also think of prescriptive analytics.
- [18:28]
Unknown
Right.
- [18:28]
Mike True
And so you can leverage AI to say, hey, we need to, we're going to prescribe you to increase your top of funnel spend during this time period of seasonality because you need to start filling your bottom of funnel during this Black Friday Cyber Monday season. And this is when you should start to implement these changes. And so there's a prescriptive nature based off of the predictive outcome, what should be some sort of value add or incremental growth on profitability compared to your existing strategy. So there's a nice blend in between taking actions based off of the confidences and predictive outcomes. And in my opinion, I think the world is going to move towards a, you know, more of a confidence based automation. So hey, we predict this is what's going to happen if you make these changes and just kind of let the machines take care of the activation on that side.
- [19:20]
Yeah, yeah, definitely agree there. I think things are definitely moving more. I mean there's just, there's too much to do and not, not enough resources to do it right. So to, to be able to automate to that degree is seems, seems to be the, the right way to move. So you gave an ex one example of being able to use, use prediction to enhance a marketing strategy. But I wonder if you could give maybe another example or two of, you know what, what are some of the insights that predictive analytics can reveal about a, a company's market marketing strategies?
- [19:53]
Yeah, I think a lot of it comes down to a lot of these businesses are heavily season seasonal.
- [19:58]
Unknown
Right.
- [19:58]
Mike True
So you have your, you can predict what's going to happen based off of the spend. Right. But there's other variables that are going into that are alongside just how you're deploying your marketing budgets.
- [20:08]
Unknown
Right.
- [20:09]
Mike True
And so when you start thinking about, hey, how can I predict the impact of the seasonality of, of my business and how that correlates into our marketing spend on the retail side, right. Like are there le leveraging things like in store foot traffic or in store promotions alongside your paid spend? And so it gives a much more like broader view of being able to have the finance team and the marketing team really align of all things that are tied to a marketing budget outside of just how you're traditionally making your spends. Right. So it can get very granular into things like promotional codes, you know, for a specific retailer by a dma. Right. And so I think using predictive analytics to tell a higher level story for media planning and optimizations, but getting more granular in real time and leveraging data points and inputs that are just not tied to your spend that can impact the business.
- [21:04]
Yeah, yeah. So, you know, looking ahead in the coming months, you know, how do you see AI and predictive analytics continuing to shape the future of how businesses shape their marketing strategy, their ad spends? You know, what, what should we be keeping an eye out for?
- [21:23]
Yeah, I think there's, I go back to this triangulation a lot. We have multiple forms of measurement that are being combined into one model and being able to understand the impacts and what the, what those forms of measurement are trying to tell you and really coming out with like very prescriptive recommendations that can be applied across any marketing channel, across any industry. Where I do think the future is really going to be held in is generative AI with creating creatives. Making creatives is a buddy of mine who just started a company and you take one picture, it turns it into, you know, an influencer, into 100 different videos that can be applied there. And so, you know, the combination of using measurement plus creative, I think the AI is going to, is going to tell a powerful story and then tying it back to the automated side of things is, you know, could you, could you reallocate resources to focus on more tactical things a creative strategies versus having to do, you know, having a human execute campaigns where, you know, something we're really focusing on is, you know, that can you automate portions of media buying based off of confidence scores and predicted outcomes?
- [22:29]
Unknown
Yeah. Yeah.
- [22:30]
Mike True
Love it. Well, thanks again for for joining today. I've got one last question for you. Like to ask everybody on the show, what do you do to stay agile in your role and how do you find a way to do it consistently?
- [22:43]
You know, on the personal side, I'm a daily runner. It's where I go. I live down in Miami. I run around Brickell Key and just really process what's going on in the day, what we're going to be doing tomorrow. I love to read books and I love to jam out with my co founder on, on different ideas and just really, you know, level set and making sure that, you know, we got the right culture and you know, just, just a lot of, you know, a lot of time to focus on the health but also making sure that, you know, the business is running in the right direction with my co founder.
- [23:13]
Unknown
Yeah.
- [23:13]
Mike True
Love it. Well, again I'd like to thank Mike True, CEO and Co Founder of Prescient AI for joining us and sharing his insights. You can learn more about Mike and Prescient AI by following the links in the show notes.
- [23:26]
Greg Kilstrom
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 more easily. You can access more episodes of the show at www.greggkillstrom.com. that's G R E G K I H L S t r o m.com While you're there, check out my series of best selling agile brand guides covering a wide variety of marketing technology topics. Or you can search for Greg Kilstrom on Amazon. The Agile Brand is produced by Missing Link, a Latina owned, strategy driven, creatively focused, fueled production co op. From ideation to creation, they craft human connections through intelligent, engaging and informative content. Until next time, stay Agile.
- [24:16]
Unknown
The Agile Brand.