SUMMARY KEYWORDS:
data, customer, treasure, cdp, insights, customer lifetime value, stage, run, audience, based, purchase, marketers, product, building, segment, platform, browse, increased, collect, machine learning models
00:02
Hi, my name is Shawn Hong. I am a Solutions Consultant with treasure data. Been with the company for about four years now and today I’ll be walking you through a brief product demo of treasure data’s customer data platform. So let’s get to it. So treasure data we were founded in 2011 as a cloud data warehousing solution. But what happened was around 2012 2013, a lot of our customers started requesting customer centric marketing focus use cases. And so our founders at that time, they saw an opportunity to pioneer the enterprise CDP space as we know it today.
Now, today, we’re 100% backed by Softbank, we’ve got a rapidly growing footprint with 500 employees worldwide with offices in Mountain View Vancouver, Europe, Japan, Korea and India. And thanks to our cloud data warehousing background, we’re managing 170 trillion data records at any given point in time in the system, with more than 180 out of box data connectors. And we’re also highly reliable with a 99.99% uptime. We also adhere to some of the world’s strictest security and compliance standards, including some regional ones like GDPR, and CCPA. And we’ve also got more than 400 customers, some of whose enterprise logos you can see below on this screen.
So you’ll see here we service organizations from all different backgrounds agnostic of industry, so whether you’re b2c or b2b, retail, auto gaming, etc. As long as you’re using customer data in some way, shape or form to drive returns, you will find tremendous value with trended data CDP. Okay, so next, I’m going to walk through a typical retail customer journey that we set up for our customers. And this is just so we have an idea of what we’re working towards here. And then afterwards, we’ll go into the platform to see what’s powering all this underneath.
So let’s say I’m a customer coming to a brand’s website to look for T shirts. Now we’re on our E commerce site and the treasure data web tag, it’s set up in the background to capture all events, clicks, and interactions on the site in real time. Now, this web tag can be used in conjunction with any of your existing analytics tools, such as Google Analytics, or Adobe analytics. This widget down here at the bottom left, it gives us a direct line of sight to what the CDP knows about me.
Currently, we don’t know much, however, treasure data, it’s already trying to generate insights, such as my propensity to purchase so we’ll see how these insights become meaningful over time. Everything is streamed over to our real time dashboard. And we’ve got actionable insights for marketers to optimize each customer journey. So even though I’m anonymous at this time, treasure data will continue to collect everything I do, and then use these behavior signals to reach powerful insights that will push me through this journey funnel down here.
We can also see the homepage that I just landed on as well. Now I want to browse t shirt. So that’s exactly what I’m going to do at this time. And so men’s T shirts, scroll scroll. Now as I browse, I see the Steve McQueen t shirt that catches my eyes. So I’m going to click on this right now. We land on the product page, and we can open up the CDP box again. Treasure data immediately captures this behavior and moves me to the next stage of the funnel informed because I clicked on a specific product. And you can see the propensity to buy has increased to 10. And I also see what category of product I’m interested in as well men’s T shirts. But now let’s say I need to run to my next class and I close my browser when I pulled the UCB again.
So even though I’m still an anonymous user to the CDP, there’s still a ton of data that we can use, such as interest words, which is collected through our content affinity engine. We see the last product that I just viewed as well. And finally, again, we see some of our newer page visits. Okay, but let’s say I’m bored later in the day, and I decided to browse Facebook. Okay, so you can see I’m now being targeted with an ad here because treasurer data has sent this information to Facebook based on my browsing behavior. So even though I’m still anonymous, we can still power personalization across the internet. And the goal here of course, is to ultimately drive me back to the website and complete my conversion journey.
So I’m going to click on that ad right now. That sends me back to this t shirt that I browsed before. And if I open up the CDP box, I’m now in the favorable stage because I clicked on a Facebook ad for a product that I previously viewed. We can also see that our propensity score has again increased to 40. And we see of course that Facebook here is our second touch. So back in the UCV we can see that the CDP has logged my second interaction with Facebook. That’s of course essential for attribution and and understanding what drives a customer to ultimately convert.
04:52
web activity has been updated as well. Okay, so an important point in any customers journey is when they provide something key information about themselves, in turn allowing us to know who they actually are. So now that our insights have determined that I’m pretty darn likely to make a purchase, we can nudge me to submit my email address in exchange for a discount.
Okay, so as I navigate through I see the offer, I’m going to input my email address at this time. Subscribe. Okay, so we now see that my likelihood of purchasing has increased to 80. But also, we’ve not only moved to the consideration stage, because we’ve subscribed for that newsletter, we suddenly see a million other new attributes attached to this treasure data ID. So what’s going on here? Previously, I was completely anonymous, with just a few behavior signals in my profile from my current session. But now we’re seeing a complete 360 degree view of who I am.
So our propensity scores have again increased, suggesting I’m super close to making a purchase. My email has been added in so our ID unification engine, it’s backstitching data attached to that email into the role level ID. Next, we can see that my touch points have updated as well. So we’re now seeing a holistic view of how I’ve interacted with the brand historically, as associated with my email address, which further fine tunes or attribution models.
We also see now our next best product, Levi’s 512, slim tapered jeans, we’re going to come back to this in just a second customer lifetime value, last purchase date, what I last purchased, where my last purchase was, and that I’ve given marketing consent, all of which can be used to drive better personalization experiences. And so yeah, all this information is key to building a customer data, a customer centric data strategy that’s focused on building deeper relationships with your customers, and building customer lifetime value. And if we go back to the homepage, treasure data is next best recommendation engine presents me with a pair of those very same Levi’s 512 jeans that we saw earlier in the UCV. Okay, so that’s kind of like a prologue. And it’s going to become clearer in terms of how this is enabled. But with that, I think we can switch back to the platform to showcase the framework of what we just saw here.
Treasure data is comprised of four distinct steps when it comes to your customer data collect, unify, analyze, and activate. So going through each of these steps are one by one starting with collect, we have more than 180 connectors out of box, including those two, cloud and file storage tools and databases like Amazon s3, redshift, Azure, Google Cloud, secure FTP, etc. But say if we were to connect to a commonly used tool like Salesforce, we can simply connect using OAuth. And then once we log in, we’re ready to start ingesting files from our Salesforce instance.
So let’s say we’ve collected data from all these different sources, probably the most important step in the sequence. Step two is to unify your customer data. Now treasure data, we automate this process for you with the goal of creating these golden customer records these unified customer views of your prospects and customers, where we can tab through a composite containing aggregate metrics, like, you know, customer lifetime value, predicted lifetime value, interests, keywords, these are all generated by treasure data’s analytics engine, we can look at customer behavior, track our customer journey, as well as all the individual touch points that make up that journey in store iOS website. We can look at how well our campaigns are performing.
And then finally, once again, next best action. So this is all interesting information. But until we do something with this info, it’s all virtually useless. Which means the next step is to make sure we leverage all of these spectacular insights when we’re creating segments of audiences to run our marketing campaigns. And all this happens inside our audience studio. So we can click into one of our many segments here, Facebook acquisition, edit segment rules.
We’re then presented with our segment rule builder, it’s got a drag and drop interface to make it easy to use for everyday marketers. Now this right here is an acquisition campaign as per the title. Okay, and it looks like the rules here fit that use case orders are at zero, we’re looking for customers that haven’t ordered anything. And pageviews are also greater than zero based on a content affinity for Pinot Noir. We have 2400 profiles that fit that segment. So we’ll then send the segment over to Facebook to run either a look alike campaign using this audience as a seed audience or targeting these individuals directly using their pie.
09:53
Okay, so you can create single segments like we’ve just done, or you can stop multiple segments together to create an omni channel customer experience where each stage is being activated to a different channel, according to the customer’s time channel and content preferences. So stage one who adds stage two, Facebook custom audiences. Stage three MailChimp or message gears for email. And so we’re basically getting more and more direct with our outreach efforts as we move down this funnel all the way until the final point of purchase down here. And even beyond that to things like upsells, cross sell, abandoned cart, retargeting abandoned session.
These are all very common use cases that we see deployed on top of treasure data. Lastly, I want to talk about our predictive scoring capabilities. So first, we have a UI based predictive scoring tool, which is built to be easy to use for everyday non technical marketers. So we have something in the system for all levels of technical expertise. So in this UI based predictive scoring tool, you know, here we’re looking to see who among our audiences are most likely to churn in the near future based on other people that have churned in the past.
But marketers will be able to go into this tool and create look alike audiences for any dimension of behaviors in the seat audience, and apply this to applications like next best product action, offer preferences for time channel content, how likely people are to subscribe or visit a physical store or in this case, again, we’re going to use it for churn prediction. Beyond that, we also have an integration to a machine learning library called hive mall, that’s actually going to let you run machine learning models natively in our query editor. And this is all, you know, fully documented online in a documentations page.
So this is our query editor, running a machine learning function. And here is our documentation. Okay, you can also bring in your own machine learning models and attach it to a project by using our Python Editor. Okay, you might even choose to use this editor to build out your own custom integrations to your tech stack. The platform is truly infinitely customizable, which is important for our enterprise clients, who always have their own unique needs, which require their own custom solutions. And to help with customization. We started last year, this initiative called treasure boxes, which is essentially a catalog of code snippets for frequently requested custom use cases with contributions from our customers, engineers, and even partners who through treasure boxes provide value add services like data enrichment, integration services, AI, ml, data management, etc.
Okay, so that’s really the whole loop here. The gist of it is, you have a bunch of data sitting in separate silos. They’re not designed to communicate with each other because that’s not what the vendors originally intended. But with treasure data, we will unify and get rich insights out of these platforms that will then send to your destination channels so that you can run fully informed campaigns on any channel based on the best time plays and personalized content using all the data at your disposal. Okay, but that’s it for for an introductory walkthrough of treasure data CDP. Thank you so much for your time.