Knowing your customers through data and experimentation

Kristen Guttas: Hello, hello and welcome to our webinar. We will be starting in five minutes, so grab a coffee, grab a seat and stay tuned. We'll be with you shortly.

Hello and welcome to know your customers through data and experimentation. We're joined today by mParticle and we are very excited to give you a glimpse into the value our many mutual customers have been able to obtain by combining the power of Mixpanel with mParticle. Before we begin, a few housekeeping items... If you're having technical difficulties, try refreshing your browser. That usually does the trick. If you have a question during the Webinar, feel free to enter it into the Q and A widget. We will address all questions at the end of the webinar. Lastly, to enhance your viewing experience at any time, feel free to click into the viewing expander on the bottom of your viewer.

Onto introductions.... I'm Kristen and I'm in Product Marketing at Mixpanel and today I'm joined by Daniel Lee, who is a Strategic Partnership Manager at Mixpanel, and Justin McManus, who is the Head of Solutions Engineering over at mParticle. We are super excited to have both of these experts presenting today. A big thank you to both of them.

Before I hand it over to Justin, I'm going to quickly give us a look at what we'll be covering where I started talking about how you can know your customers better through cross platform analysis for moving on to best practices for ringing Mixpanel and mParticle data into other tools. We will be closing out with a quick demo illustrating or popular joint use cases. Now onto Justin for an mParticle overview.

Justin McManus: Awesome. Thank you so much, Kristin. Hi everyone. Hope everyone's having a great day. To give a little bit of an overview of mParticle...

First, we're here to really allow brand marketers to do more with their data by ultimately solving for fragmentation. This allows marketers to centralize all of their own customer data to build audiences and send events so all their various tools of engagement and tools of record. This allows them to be nimble, agile, and reduce reliance on engineering and ultimately stay ahead of the competition. What we really strive to do is break down data silos, by having a single platform to ingest data from all your digital properties, internal systems, as well as other cloud based tools that may contain customer data. The goal is to build a central customer profile and then activate it all your different analytics and marketing tools.

Brands that have successfully deployed a CDP (customer data platform) are now able to have a single source of truth of all their customer data. A 360 degree view is more than just collecting data from your digital properties. It also has to include data around location, attribution, deep linking, helpdesk data, or marketing automation data... Which will be highlighted in our "better together" demo towards the end of this webinar.

A lot of the principles around having a CDP are certainly aligned to privacy or regulatory needs as well. Whether it's initiatives or compliance needs like GDPR or VPPA, a CDP has to be able to support you and really give you fine grain control around how you identify and build your customer profiles and choose to share certain data with different types of partners.

Lastly, a CDP really allows for integrated marketing and orchestration across all your channels when a CDP is sitting at the core of your customer data and allows you to send audiences between paid media, email, push notification, DMPs, and various different engagement channels, while synching that of course, so your analytics tools like Mixpanel.

If we look under the hood a little bit further, there are really five key steps around having a CDP. Versus that collection piece, we provide a variety of different mechanisms that allow you to ingest your customer data, whether that's through direct SDK integrations, having server-to-server APIs or directly ingesting data from partners like Braze, like Zendesk, like Foursquare, or any other kinds of location or attribution providers.

Once the date is inside of mParticle, there's a ton of fine grain control around identity resolution, how you treat anonymous to known users and what happens on that transition to log-out. Also for scenarios where they're shared devices, where there are many profiles living on a single device too. We basically offer fine grained control around how you identify and manage user profiles, that can be applicable to any sort of regulatory or privacy needs that your specific company is addressing.

On the activation side, it's really important to have that flexibility around sending raw events to different channels or sending audiences across all your different marketing partners.

Lastly, it's also key that CDPs are bi-directional and support the entire ecosystem of partners that you're using today.

In terms of the brands that we work with, we ultimately work with large consumer-facing bands who really want to do more with their data and stay ahead of the competition. There's lots of verticals that we support ranging from eCommerce, multichannel, retail, media and entertainment, all the way down to lifestyle and financial services.

Again, we work with very large consumer-facing brands and that ranges from the Walmarts, Overstocks, Spotify, Viacom, Venmo, and Paypals of the world. And with that, Dan, over to you to give us a little bit of an overview into Mixpanel.

Dan Lee: Great. Thanks Justin. That was an awesome overview of mParticle. So what I'll do is I'll do a high level overview of Mixpanel, kind of who we are, what our core value proposition is, and then we'll kind of go into that joint demos. So everyone in the webinar, you can see what it actually looks like in practice.

So at Mixpanel, we really believe that it's critical to your business and really every function across the entire business to really understand users really deeply. In order to build better products, or to create better messaging, or to deliver better services, it's really important for all organizations to be hyper data driven.

And so for Mixpanel, our data model is actually from the start built around the user, which allows us to track pretty much every interaction and user has across any digital property... Whether that's mobile, or web, or on wearables, or smart devices, or whatever that is. We can easily link all of those actions together around a single unified user profile.

In recent years, we've actually opened up our platform to ingest other types of data. So along with the typical interaction data — like logins, or deposits, or subscriptions, or purchases, or swipes, or tests, or cliques within your digital properties. We're now able to bring in other types of data such as: attribution data, install data, data around — if you're in B2B situation — about your account level data... Whatever else that might be, we can actually bring that into Mixpanel for further and deeper analysis to understand users and mParticle is a great way to be able to kind of do that very quickly and elegantly and easily.

So once that data is actually inside the Mixpanel platform, our goal, you know, at the highest level is to really help our customers and our users, you know, answer questions about their end users faster than they do today. So with Mixpanel, you know, we'll see this in the demo with just a few clicks.

You know, a few taps here and there, you'll be able to build reports, dig into your data, dig into your users. They'll get that insight without any sort of reliance on any technical knowledge. We really believe in the democratization of data so that you know, everyone around the entire organization can and should be really intimately knowledgeable about their users.

We've been doing this for a while now — almost 10 years — and because of this, we've been able to see lots of customers achieve success. We've been able to scale our business both in terms of customers and our infrastructure, which we believe is best in class.

And so over that course of time we've worked with lots of, you know, big customers, innovative customers across many different verticals. Some examples would be, you know, we work with companies like Docusign to really improve their conversion flow or for Starz.

We've helped them collect real time user data to really improve the experience for their on-demand streaming services, and work with companies like US Bank to really understand kind of their in-store payment solutions.

As a part of the webinar, there are handouts, so you'll be able to download kind of some case studies both for mParticle and Mixpanel... The Mixpanel ones have to do with stars and one of our big customers, Viber.

So you've learned a little bit about mParticle, we've learned a little bit about Mixpanel, but we're going to hop into two demos right now. We'll look at both kind of platforms independently, but we'll also touch on some innovative use cases around actually combining all the different types of data and different data sets which should really be unlocked by combining the two solutions.

So, you know, we'll be looking at the typical user behavior from Mixpanel combined with attribution data or data for your marketing automation solution or location data and bring that all together to really understand the user and the user journey as a whole whole concept.

And with that, I will actually hand it back to Justin, to show a little bit of a what the experience is like in, in mParticle.

Justin: Awesome. Thank you so much, Dan. Excited to jump into really the meat of the presentation here and walk you through exactly how we can get data from a variety of different sources that live outside of digital properties such as location data, attribution data, marketing automation data, and get those into tools like Mixpanel to help you understand more information about your users.

You know, here's the mParticle dashboard. You know, really the first step in any sort of engagement with a customer is to go through data planning. Data planning is that process where we take a holistic top down as well as a bottom up approach to really understanding where all your customer data live today.

This live stream is a great tool that allows you to do real time QA-ing to see, again all that data coming in from all your different inputs. So whether it's data coming in from iOS, Roku, Smart TV — or tools like Braze, Radar, Zendesk — you can actually QA the data in real time.

You can also look to see all the different outputs of where that data is connected. You can also filter on different message directions as well as look at particular devices. If you're looking to do QA-ing on a specific dev build.

You can click into individual message records — and this is a little bit more for the technical audience here — but we expose all the raw data associated with any sort of event coming into mParticle. Again, the ideas around full transparency of data and giving your engineering team really quick and easy ways to QA data because again, this is the foundation of the data that's going to be going to tools like Mixpanel and the rest of your marketing growth stack.

Once integration is up and running and you're looking good and feeling good about it, the next step is to go over to the directory. This is the portfolio of partner integrations that we have that come completely out of the box so you can look and see all the different integrations across analytics, data warehousing, and marketing tools, but you can also key in specific service providers.

So I can click on Mixpanel and I can see a lot more information about what the integration looks like. I can see the granular raw data that we're sending over to Mixpanel, but I can also see that we support an audience as well as an events integration with Mixpanel.

If we look at a partner, it's such as Braze, we can start to look at other types of integrations as well. For example, Braze supports what we call a feed. A feed basically allows us to ingest additional data that is not generated by mParticle. This would be information like email sends, clicks and opens. If you're looking at Zendesk, these would be points of data around cases opened or close. If it was Foursquare or Radar, this could be around geofences entered or exited. These are all events that we allow easily being sent over to partners like Mixpanel.

The next step is to manage your air traffic control. In this example here, we're essentially taking the different inputs, whether it's data coming from direct digital properties or if it's coming from partners like Braze or Radar and we can essentially point it over to our Mixpanel project so I can click on Braze here for example, and see that I'm already sending data over directly to Mixpanel.

We also allow for different types of configurations, right, so you can basically determine if you want to send sessions over to Mixpanel. If you want to create profiles, if you want to use Mixpanel's people product, there's a lot of different configurations that are easily managed through the UI as well as also managing things like super properties, which is specific data model over to Mixpanel.

Once you've essentially set up the air traffic control and you're navigating your data from all your different sources to your different inputs and outputs, the next step is to look at the different governance capabilities as well.

Essentially here what you're looking at is the full data model inside of mParticle and all the different partners where we're sending data to. Here on the user screen, I can actually pick and choose the different types of user identities I want to share with these different partners.

On a per event level, I can actually get down to the attribute and specifically select which type of attributes and events I want to share with each of these partners. This prevents engineering teams having to go in and manipulate client-side code.

More specifically around the use cases that we're looking to solve for here today, if we look at Zendesk feed, we can see there's a number of events around cases being opened or closed as well as specific attributes.

This allows me to select the different data points that I want to share directly Mixpanel, so whether this is a Zendesk feed, a Radar feed, these are all events that are being ingested into mParticle that we now have full control over to determine how and when we share that over to partners like Mixpanel.

And with that, that concludes the mParticle portion of the demo and I'd love to hand it back over to Dan to talk about the insights that he can gather directly by mParticle sending the data over to Mixpanel.

Dan: Cool. Thanks, Justin. That was awesome.

So, I'm going to hop right into our Mixpanel project over here, share my screen, and I'm going to take a look at two different projects today. So, I'm gonna use one project, this is a high level overview of Mixpanel and then we'll hop into another Mixpanel project that is actually specifically catered to some of the use cases with mParticle.

So the first project, we have kind of a basic music playing application. We can see that we're collecting some events here, you know, playing a song, even purchasing a song, signing up, upgrading of plan, and there's even the capability of signing up for a newsletter. Of course in this demo environment we're only tracking a couple of things, but it Mixpanel, you know, we could be tracking a typical customer, might be tracking hundreds or thousands of different events across different properties and different platforms and things like that.

So here, what we'll take a look at is first we'll take a look at, you know, the capability of Mixpanel to really look at things in dive into an analysis very quickly and easily. So, actually take a step back and what we'll do is we'll take a look at all of our users. But what I want to do is maybe I'm curious about where our users are actually located for our music app. And so we can see that a of our users are from the US and we do have users from some of these other North American countries like Mexico and Canada, and that's great. But what I really am curious about is, you know, our users genre preference.

So, you know, just really quickly here I'm going to look for are your favorite genre and instantly we can see, we can segment our user group and actually see which genre preferences are more popular in different countries. Right. That might actually dictate our product experience when we personalize our app and based on, you know, user location that might actually change our marketing strategy. Right? So Mixpanel's typical user base has been, you know, catering to product managers, but now we can kind of see how this would benefit all organizations across the entire company instead of just, you know, the product management team.

So that's an example. And of course if we wanted to, we could break this down over and over again. We could take a look at a time series graph, we can do so many other types of queries. So, I'll just jump into a couple other key reports that, you know, are typically the highest use case and highest value reports from the Mixpanel platform.

So I'm going to hop into this funnel here, and this is for the same application, and what we're going to look at a funnel that keys around a signup flow and playing a song. So we have the three step funnel: signup, song play, and song purchase. You know, we created this very quickly, we can take a look at conversions and we can even look at the conversion rates over time and we can see that, you know, maybe we did something in our product development cycle that has lowered our conversion rates.

But I think one of the biggest value propositions of Mixpanel is being able to take this, and you know, if you're looking at this report, you might say there's a 27 percent conversion rate. What does that mean? Is that good? Is that bad? How do I actually take action on that from, you know, people playing a song to actually purchasing? So something that we can do here is we can actually segment, by a property, let's say something like initial referring domain.

Let's see here. And, in this data set, we're actually seeing that we're only referring from one domain, but let's actually segment by something like country. And by segmenting by country, we can kind of see that, you know, for different countries, we're actually converting at different rates and that might allow us to dictate our marketing strategy of where we want to invest, where you want to make product improvements, whatever else that might be.

I'm going to hop over to our retention report here. And you know, of course it's great to understand your users. Of course, it's great to optimize conversion funnels... Whether that's the signup flow, or activation, or getting to some sort of keystone event — like playing the song or even a purchase event.

But one thing that I want to take a look at is, you know, when users are coming into our application, are we able to actually retain them over a long period of time? So here, what I'll take a look at is I want to look at users who played a song and then perhaps played a song again.

And what I'm actually going to do is I'm going to go back for further timeframe and I'm going to segment by a property such as plan type. And so what I'm really curious about is whether, you know, our platform, our marketing efforts, our product efforts, everything that we're doing is really helping retain both our free customers and our premium customers.

And so we can kind of see here as the weeks go on and we can kind of see the drop-off in our users over time and are we retaining them. And we could be segmenting this again by plan type or genre or country or whatever else that might be any sort of metadata about the user or their actions inside the product.

So that gives kind of a high level of review of Mixpanel. Of course we have all these reports, we can look at how users are flowing through our product. I'm doing analysis on, on, you know, are they converting, retaining. We also have capabilities kind of covered in some other case studies around messaging those users based on the activity that they're performing or even using our suite of machine learning products to understand the correlation between events or even predicting users that will kind of perform a keystone event like a purchase or something like that.

Before I dive too deep, I want to spend some time on kind of the purpose of the webinar, which is to really key in on the, on the joint use cases of mParticle and Mixpanel. So in this project, we have a kind of a media and entertainment product, you know, keyed around, you know, viewing videos and things like that in a content platform.

So the first thing that I'll show off here is kind of this full funnel of all the things that we are pumping into this Mixpanel project. Some of these events are actually, you know, Mixpanel events that mParticle is helping us track. And a lot of these events are actually coming from other products across the entire stack... So for instance, we're actually now being able to build this funnel that actually starts with an install event from a mobile attribution solution, like an AppsFlyer or Branch.

Then we actually have this open app event that is probably a Mixpanel event. Same with this registration complete, but in our customer and user life cycle, what we're doing is after they're being registered, we're actually sending them an email to see if they actually start their trial and we're sending them an email. They're opening the email hopefully, and then they're starting to trial, and converting on their trial.

And these two events here — from the email sent an email open — those are actually coming from mParticle through one of their other mutual partners, such as a Braze. So we're taking that marketing data that typically lives inside Braze and moving that into Mixpanel for this whole entire journey analysis. Unlike before we could segment by any of these properties, I'm either coming from, you know, the mobile attribution solution or the marketing solution or data that we're collecting inside Mixpanel.

A couple of really innovative use cases that I think are really relevant to a lot of customers that we're seeing nowadays are a few other funnels that I want to show us. So this one, you know, this wasn't covered in the, in the full stack funnel that I just showed, but you know, we're, let's say dealing with, you know, on this platform they have a movie special or comedy specials for, you know, stand up performers and of course we can watch those videos online, but a key component to the success of a standup comedian or content producers — oftentimes kind of that in-store brick and mortar experience. And so one of the things that we can do is, you know, mParticle opens up these possibilities where we can actually send in location data from a partner like Radar.

So in this case, what mParticle's helping us do is leveraging a platform, like Radar, which can help us understand when a user actually goes to a physical location... That could be a brick and mortar store that could be an event — in this case where, you know, sending an event to radar and Mixpanel panel and mParticle when a user actually goes to a particular event, come venue or concert venue. So we're actually now able to link that data to in-app data, right? So the first event in this funnel is actually visiting the Shoreline Amphitheater and the second event is actually watching that same comedy special. And we can see those conversion rates, we can see those conversion rates over time and really link that in store experience to the digital experience, which is becoming increasingly important.

Another example is being able to see a conversion rates between, you know, a customer support experience and actually churning from the product. So for instance, I'm here, we have this funnel of someone, I'm opening up the case via the Zendesk platform that's going into mParticle and mParticle shooting that over to us. We can see that we have, you know, 30 percent of users who are opening up a case on Zendesk with us are actually uninstalling the app.

And then if we wanted to, we could send additional categorical information about, you know, what does that ticket tagged with and use mParticle to send that over to us and we can segment this funnel by some of those categories. So without going too deep into how these events can be leveraged across all the different applications, I want to bring Justin back into the fold here, so we can kind of start a little bit of Q&A, ask any questions, and dive a little bit deeper into any particular areas.

Kristen: All right, thank you both so much. That was so informative. Looks like we have a few questions and the Q&A box already. And if you haven't submitted a question yet, feel free to do so now.

First question comes from Alex and it says, "Do you see companies moving in direction utilizing deeper behavior analytics to make business decisions? Are there certain verticals that you see that are better suited for this?" This question can go to, to either Dan or Justin.

Dan: Yeah, I mean I can start with that. So I think for us it's really interesting with Mixpanel being, you know, around for so long we typically started with companies, you know, had their entire business be digital, right?

These companies that were very mobile forward, maybe doing some cross-platform stuff with mobile and web. But now I feel like, you know, we work with so many companies across so many verticals where everyone has some sort of digital experience, right?

So for Mixpanel, a lot of our strongest verticals are going to be financial services and going beyond just like the B2B SaaS companies, and consumer technology, and media and entertainment, and working with fortune 100 companies. And some of those examples that I talked about before is, you know, for instance, the idea of connecting digital experiences with actual physical in-person experiences.

That's something that with the help of partners like mParticle and Radar, we can actually analyze it and link that together. And that kind of opens up the value for so many other customers who could be leveraging tools and to become a far more data driven about their entire user journey. Anything to add, Justin?

Justin: That was awesome, Dan. Yeah. Echoing a lot of similar thoughts there. I think the customer journey is always changing. Ten years ago, very much focused on web, a few years later mobile app started getting into the picture. Now there's platforms like Radar and Zendesk that are containing customer data.

We look a little bit to the future and there's voice and these wearables, and I think the journey's going to consistently change, right? The number of devices that people have, it's just going to be increasing more and more every single year. And I think most companies do have data challenges, right? And they do have data challenges which leads to challenges around learning those insights from the data.

So I very much agree with absolutely everything you said. In terms of verticals. Yeah, totally agree with those verticals you mentioned there as well.

Kristen: Thank you, both. Great first question.

Next question comes from David. He's asking,"Once we've segmented users in Mixpanel, how do we then send that segment an email?"

Dan: Yeah, absolutely. So, there are a couple different ways to do that actually, David. And maybe I can actually jump back into one of these projects. I'm not sure that this demo environment is optimized for it, but I can kind of show you at a really high level how that might work. So I'll just quickly start my screen share again.

So for, in this project, you know, we've analyzed our users, we've identified that we want to send our users who are just our free users an email or users who haven't converted through a funnel, an email, or let's say your users who, you know, walked into a store... Like walks into a physical location store but didn't actually, let's say I'm making an online purchase, they didn't make a purchase in the store, and they still haven't made a purchase online. And we want to send them an email coupon to kinda get them to, to kind of engage with us online.

So for Mixpanel, our navigation is set up where we have dashboarding, and we have analysis, and understanding users, and then we have this take action area. So what we can do is really simply, we can go over to take action, we can create a message and we have a couple different channels... So email, SMS, push, in-app, we can also use webhooks if you have a different email engine that you like to use.

So a lot of our customers, you know, we'll send webhooks to something like SendGrid. You can craft something here and one for here, we can actually recreate that targeting criteria based on both user attributes. So let's say the country or their favorite genre of music or their plan type or we can do it based on a series of events so we can say that they played a certain song at least X number of times within a certain number of days. Whatever else that might be.

So, that's kind of how one might take action on their data. We also integrate with other partners if you are, you know, your team is working well with some other marketing platforms that you guys really love. And so, we're not here to, to hog everything. We work well with, you know, the Brazes and Urban Airships of the world and we have a Marketo integration where you can export, you know, a cohort that you've built inside Mixpanel to that platform and engage them and all the campaigns and, and, you know, user journeys that you've, that you've created in these other platforms as well.

Kristen: That's great. Next question. Devin asks, "If you were giving advice to someone wanting to undertake a customer data strategy, where would you recommend it again?"

Justin: Yeah, absolutely. I can jump in and take this one. So one of the first steps we ever take with our customers is do what's called a data planning session and this is where we get all the various stakeholders into a room — product, engineering, marketing executives, anyone that's involved in customer data — and we take a holistic approach with a sense of an understanding of bottoms up.

You know, what data are you already collecting and what do you feel comfortable with, what are the areas of improvement, but also from the top down, right? Working backwards for those KPIs that you're looking to drive across your analytics and marketing practices. Once we have all that information consolidated, it becomes really, really easy just to develop a data plan that aligns to both the short term as well as the long term strategy.

Dan: Yeah. I think our, our philosophy at Mixpanel is actually very similar. We always, you know, I think one thing that we've seen, um, especially enjoying working with mParticle as, as a key partner for us is really the alignment around our services and our approach to making sure that companies are really set up for success long term.

And I do actually agree that like that planning phase is so critically important and now you never know what data that you might need down the road to answer a question, right? With how fast products move and organizations move, you know, new data is always coming to the front and we need to make sure that we're tracking that stuff to be able to uncover that.

I think at Mixpanel, we always, you know, think about like, what are your business goals, what questions do you need to answer to drive those business goals? And then what data do you actually need to answer those questions? Right? And working backwards from what is the ultimate goal here, the goal is never to track as much data as possible, right? The goal is never to like collect all that, it's always to actually drive something forward. Um, and so, um, I think there's a lot of alignment there between like how we both approach that.

Kristen: The next question from Devin actually kind of goes along with what you just said, "In terms of bringing in historical data, is this something that can be incorporated?" I think there's probably a good question for Justin.

Justin: Yeah, absolutely. One of the core fundamental aspects of having a CDP is also having that protection of your customer data. Yeah. It's really just the flip of the switch for us, we make it super easy to import historical data, whether that's analytics data, but also fits things like email addresses and push tokens and any sort of marketing identities that are important for a marketer to really easily just jumpstart their approach with.

Kristen: Awesome. Thank you both. I think that concludes our Q&A and our webinar, which has been great! Thank you both so much for joining. I've learned so much. If everyone who attended wants to take the brief survey that we'll send out afterwards, that'd be great. Thank you so much for joining.