Thanks, Anthony. All right, welcome everyone. My name is Alex Mie. Uh I'm here on behalf of Twilio. So many of you might know the communications APIs that Twilio has. Uh today, we're actually gonna be talking about another product that is in the Twilio portfolio called Segment. Uh so you're gonna learn about customer data platforms and how you can leverage those uh using AI to actually drive business outcomes for your organization.
So a quick note on the agenda, we're gonna talk first about this problem of system centricity versus customer centricity. Uh we're then gonna talk about unified customer profiles and how you build those. Uh and then we're gonna talk a little bit more about customer data platforms and I promise we're gonna spend most of our time on an actual demo so that you can actually see what this looks like.
So real quick by a show of hands, how many folks here are in the data space, spend a lot of time uh actually cleaning, preparing data. Uh couple of folks. Ok? Uh by a show of hands, uh wishes it was the other way around. Uh how many of you have a last mile integration problem, you get all the data prepared, you build some great models. Uh and then to actually get it into marketing's hands or your support team's hands. Uh you have to make sure to integrate it in some downstream tool. A couple of you. All right.
So this is part of the problem that we call system centricity where over time companies will buy a particular solution for a particular part of the business. And the problem is that everyone ends up with a slightly fragmented view of who their customers are and different sorts of truths. Uh instead of being customer centric uh and really having your customer at the heart of your, your customer journey and being able to understand all the different touch points that a customer has with your business, whether it's through your applications or if it's talking to your support team, your sales team, uh all the different places that you interact with the customer.
So a golden profile is really a way to take all of that customer data, whether it's uh clicks on your website, uh mobile application, um installs uh data from email interactions, SMS, interactions with your customer and all and unify that into one place. And so Twilio Segment is a customer data platform. Uh it's a platform that plugs into your existing AWS stack and helps you do just that uh it helps you get a real time understanding of who your customers are so that business teams can then act on that uh and drive their respective business KPIs.
So we'll talk a little bit more about real time identity resolution and how that works. But that's essentially this process of stitching together all these different identities, whether it's a mobile uh phone number, if it's an email, an in store point of sale ID, uh that's a big part of what we do and how we make that data actionable and then a quick note and then we'll hop into the demo.
So, uh Twilio Segment and AWS have been partners since 2018. We have over 2.5 1000 joint customers. We just learned last night that we are one of the finalists. We're Twilio's partner of the year in the I think the digital experience category. Uh and a lot of our customers. Hopefully, some of you today already have AWS stacks that integrate uh with something like Segment like uh AWS Redshift, uh or maybe you built your models using SageMaker. Um or maybe you're trying out some of the, the new generative capabilities we heard about in the last session.
So let's hop into the demo so that we actually see what all of this looks like. So first up, we're going to take a kind of common ecommerce website where you can click around, you can wish list products, you can add things to cart. So if you've set up Segment, the proper way, you can create these different types of data sources. So in this case, this ecommerce website has a, a source for its website interaction data, which i just showed you you can also connect your mobile apps uh through our SDKs. Uh and this starts the process of being able to unify all this customer data together.
Uh so through this debugger feature that we have, you can actually see that this is the exact event that i just performed, so we can go back, we can wish list another product and all of this data comes in in real time. Uh and kind of starts to create that unified view of the customer. So the next piece is how do you actually tie all of these different touch points, whether it's from the website mobile app, from email interactions into a single profile.
So let's quickly take a look at uh the profile that i that i just was clicking around on. So right now, i'm an anonymous user on the website. I haven't provided my email or phone number, i haven't done any purchases. Uh but we're already starting to track all of these events that users are performing, like showing intent around wish, listing certain products and kind of starting to build that user profile. And so over time, what you end up with uh is this uh complex user profile. And a key part of setting that up is what we call identity resolution.
Uh so you can have email identifiers, phone number identifiers, identifiers from your mobile applications. You can even bring your own custom identifiers if you already have those. Uh and through the identity resolution, you can configure things like um ID rules value limits uh be able to actually block certain values from merging profiles together. And so the net result of all of this is that you have these rich profiles that your business teams can then leverage.
So let's hop into an example of how you can do that. So uh there's a lot of attention around artificial intelligence predictive modeling. So one of the capabilities in our platform is that you can actually create your own uh prediction model. Uh and so there's two ways of doing this. If you're a more business focused user, you can build those, those models directly here in the UI which we'll go through in a second if you already have your own scores sitting around in S3 somewhere, or you've already kind of invested in that you can then also import those scores and solve these same use cases. So 22 approaches that meet your company where they are today.
So with a custom predictive goal, there's essentially two things you can figure who are you making the prediction for? So let's say everyone that viewed a product in the last, let's say 90 days. And then what are you actually predicting? What is the target event uh so in this case, let's predict whether or not this customer is actually gonna complete a purchase. So you can calculate this and then once that's ready, uh you get this prediction that's kind of ready to use for your business teams.
Uh so in this case, you have a distribution of, of scores for the users. Uh and then you also get to see which of your uh your tracking events are actually the most predictive. So in this case, adding to wish list is a pretty high signal uh around a user's likelihood to convert.
So up until now, we're creating that data, getting it ready for the business teams. So how would a marketer actually leverage this going forwards? Um so this is where the business user kind of takes over. Uh and so if they want, they can actually create their audience uh directly from this prediction. And so you effectively have these prebuilt conditions around this likelihood to purchase score, you can preview, make sure that you have the right users, the right reach for your campaign. And then once you've built that audience, uh you can send it to any of the integrations that Segment supports, which is several 100 integrations across marketing sales uh CRM for the more data focused.
Uh folks here, you can also send it to like your own Redshift, you can send it to like a Kinesis stream if you're building an actual application on top of this data. Uh so lots of flexibility in terms of how you then want to use uh these predictions or these audiences. And so once you go through this this flow, the end result in, let's say a Facebook uh ads manager is that we create these audiences directly for you and we keep them up to date as users enter and exit the audience. So Facebook ads is one example, uh email marketing is very common, having the same data for customer support can help you route leads more efficiently. Uh if you have someone that has a higher likelihood to turn, so lots of ways that you can use this data.
Uh so now I want to talk a little bit more about um how we're building on top of the current capabilities, but actually integrating some of the data warehouse uh that customers already have. So we've noticed a trend where a lot of our customers are already investing in uh in consolidating all of their data into a single data warehouse, maybe it's Redshift, maybe it's Snowflake, maybe the bricks. Uh and so what our customers want to do is actually be able to create these same campaigns that you just saw but directly computing on top of those data warehouses.
Um so let's take that same use case that we were just looking at. Uh and let's actually build a what we call a linked audience so we can build the same uh starting the use case where we're gonna look at profile traits, uh where let's say users have a likelihood to purchase, that's, let's say greater than 0.5. And now let's say that you have some other data that's in your warehouse that is not user centric. Uh let's say that you don't want to actually send a campaign to uh to customers that have wish list of product where the product is not, not actually in stock. So that type of data, the product inventory data usually is sitting around in some other system, maybe you've already done the hard work to kind of pipe that into a warehouse. And so with this capability, you can create a uh what we call an entity relationship. And so you can actually um you can actually then filter by some of those products.
So in this case, we have a wish list products um table that is sitting in your data warehouse and you can directly pull from any fields that it already has. So we can check whether or not the product is actually in stock. Um and then through this flow, you're now targeting not only the folks that have a high likelihood to purchase your product. Uh but you're also ensuring that the product that you're actually going to retarget them on or send an email campaign about uh is actually available.
Uh and so customers that have used this combined uh flow have been able to improve things like cost per acquisition for paid ads, teams, uh conversion rates for life cycle campaigns, uh reduce customer support time uh for resolving customer tickets.
Uh Segment is this kind of data layer that plugs into the rest of Twilio, the rest of your applications uh and unlocks a lot of those use cases. So that's it from my end. Uh we are uh located just behind theater five. If you guys have any questions, wanna learn more, want to see a demo from one of our sales engineers. Uh we have a couple of minutes left over. So if you guys do have questions, I'll take questions now. If not, thank you for, for coming and sparing your lunch hour too good.
Ok.