Speaker 0: Alright. I think we are live. So, uh, welcome, everyone. Um, wherever you’re joining us from, I hope you are having a great Friday and a great day three of MarDreamin. Um, what a week it’s been so far, and it is not over yet. Uh, so I am excited to introduce our next speaker, uh, and topic, uh, how to deploy data for success, scale, and customer experience. Uh, so please join me in welcoming Jason Court. Jason, take it away.

Speaker 1: Hi, everyone. Good afternoon slash evening from, uh, from The UK. Uh, thanks for joining me today, and thanks for all the sponsors for MarDreaming. For all accounts, it’s been a great great, uh, few days, so really great stuff. Um, so, yeah, as the title suggests, gonna talk about some, uh, data modeling, uh, how to do it successfully and scale, and how to do it with your customer experience in mind. So before we kick off, a little bit about me. Uh, I’m a CRM technology lead at the Corp Group here in The UK. I’m one of the cofounders of howtosfmc.com, uh, four times Salesforce multi cloud certified, and a twenty twenty till today Salesforce marketing champion. And before we get into the nitty gritty, it’s a bit about what to expect today. And so today, I’m gonna go through why why I’m picking this topic to talk about, um, what I believe effective data modeling in Salesforce Marketing Cloud is, and how it looks, and some of the the fundamentals around how to scale for success and, uh, as easily as possible, I guess, is the crux of it. So let’s, uh, jump straight in into the the why now. So this was originally inspired by the Salesforce state of marketing report from 2021. Those of you who are paying attention will note that there is a Salesforce marketing or state of marketing 2022 that came out about three days ago. So, uh, some of these figures may have been slightly superseded. But, essentially, what we’re seeing is less than half of marketers are satisfied with with data and how they can use their data, how they can access their data. Um, You know, with 42% only satisfied with the quality and the hygiene of their data, only 37 satisfied with how timely it is. And with 40% of businesses expecting their data sources to increase between 2021 and 2022, you know, more data coming, we need to start being as efficient and scalable as we can with how we process data and make data available to our marketers. And another key thing is that only 39% of marketers are satisfied with the accessibility of their data. So our way of defining success is how we improve these metrics. Yeah. As I say, since putting this presentation together originally, Salesforce have released a new set of marketing, and some of these have got really similar themes. And some of the predictions from last year have come true. In 2021, it was 10 data sources, and 40% of marketers predicted more. Now the result is 15, and the prediction is it’s gonna be 18 next year. So it’s not slowing down in any way, shape, or form. Another key thing that’s coming out of the state of marketing report for 2022 is trying to do more with less and tightening budgets and making sure customer expectations are continued to be met. So if there’s one, uh, I guess, mantra to take away from this session, it’s this. Your data only has a limited value until you can reliably and repeatedly act on it. It’s all well and good doing something once, but that’s the value gone. You want to be able to act on and repeatedly repeatedly and reliably use your data as many times as it is gonna add value to your customers. And, I guess, with that, let’s move on to what effective data modeling in Salesforce Marketing Cloud means and could be. So to start off, just gonna do a bit of a summary of what is a data model. Uh, lots of you will be familiar with terms like data schema, database layout, data structure. Uh, in this instance, when I’m referring to a data model, it’s the structure of attributes in tables or data extensions in Salesforce Marketing Cloud and the relationship between them. It’s essentially the way that your marketing cloud users and you yourself will interact with your data and then the way that your data will interact with your customers. So you want to make this as streamlined and as easy to use as possible and give as much value to your colleagues, yourself, and your customers. So what is effective data modeling? So an effective data model and effective data modeling practices cover these five key, uh, elements. They should be enhancing processes. They can’t define a process. The data model is only a component of a process. You know, having a really effective and efficient data model may help alleviate some of the problem of a cumbersome process, but you can’t expect, no matter how good your data model is, it to solve, uh, any procedural issues or operational problems. It can make things easier, but it won’t fix a bad process. You should be using it to empower your users, give the right people access to the right data so they can give you the right results, give your customers the right experience. It should be satisfying your customers. Customers expect more and more from digital experiences every year. And if you satisfy and delight them, they’ll come back. Uh, it should drive business value. Whether you’re looking to do more campaigns with better segmentation or just manage the cost of selling products by offering a lower discount to people who are likely to transact anyway, You know, you should be doing something to, you know, really enhance the ROI of your marketing activities. And, uh, I guess, the final one is that it doesn’t stop. As we’ve seen in the predictions from Salesforce in the last two years, we’re gonna be almost twice as many data sources in our marketing compared to 2020 at the 2023. So data modeling can’t stop. You can’t build a model if you go, right. I’ve done data modeling. It’s completed. And then leave it for the next few years. You know, if you do that, you end up stagnant and you end up not with something scalable and sustainable. So in terms of what I believe are the data modeling output best practices, um, you should have a well defined data dictionary. And this is a very personal one to me because I once inherited a data model that had four fields that could have contained an email address, and none of it was documented. We had an a field called email that was a Boolean that meant consent and was not usable. We had a field called email address, which was also a Boolean that meant that an email address was present. We had a field called customer email that was again a Boolean that meant the email address that was available related to a customer. And then we had customer email address, which was the actual email address for the customer, which whilst that doesn’t sound like a hugely problematic thing to have to deal with, If you’re dealing with it on a 100 attributes and you can’t work out what means what, your data model isn’t very accessible, and we go back to that first point of, you know, 40% of customer or marketers not being able to access their data, uh, uh, especially well. You should be using your data model to enhance personalized experiences. You should, you know, enhance the out of the box tooling of the platform you’ve got available. You don’t need to build custom tools and solutions every single time. It should be delivering business objectives and operationally efficient. The last thing you want is every campaign to need 15 new pieces of code to be written when you could do it through gran and drop. And where possible, you should standardize your attributes so that they mean the same thing every time they are referenced. In terms of building your, uh, your data model, you don’t want it to be repetitive. You don’t wanna have the same attribute lots of times unless it’s something that matches a relationship. You want it to be consistent in terms of its results. So no matter who’s using your data model, they should all get the same result. Uh, ideally, you want to build something that’s quite modular. You don’t wanna build a a monolith because every time you need to change that, you need to dismantle a monolith. When it comes time to change, it’s gonna make things take a lot longer, so you’re not gonna scale. You don’t want your model to have lots of dependencies. You want it to be as dependent on as few elements as possible. And combining that with your modular design, you know, on a given day, if you’re missing transaction data, for example, you could still run your campaigns that don’t need it. But if your model needs tran transaction data to be complete, you’ll have a whole day without campaigns. You ideally want it to be pretty iterable. You know, I’ve worked in orgs where you’ve had data extensions with 200 fields. So when you want to iterate something, you’ve kind of got nowhere to iterate into. And then in a lot of cases, you can do a lot with configuration versus pure code in the marketing cloud. You know, at the heart, I’m a developer, and I’m a huge proponent of the value of code. But I’m also an advocate for being lazy where necessary. You know, the option to turn something on or turn something off is a huge benefit, um, because it means that you can scale things out very quickly and scale things back just as quickly if you need to. It could take a little bit of effort upfront, but, uh, the efficiency gains from it can be amazing. So we want how to make the most of your data, and there’s a number of considerations when it comes to making the most of your data. You need to consider the purpose. So what function does the data perform? Who’s using it? What’s their skill set? What tools do you have at your disposal? You need to consider how your your, uh, data is gonna impact your customer experience and how it impact your customers, where your data model lives in your processes, and does it add value to the business or the customer. I’m not gonna particularly touch on customer experience and value because that’s quite subjective and can be very different in different businesses, uh, but I will touch on, uh, the rest. So in terms of your purpose, what does your data do, uh, in your data model? You can have data that’s for segmentation purposes, for operational purposes, like, uh, like an email address. You can have segmentation like an age, personalization, like a name or analytical, like some kind of stratified control group, um, and where you can have all these different values with different purposes. Some will overlap them. You know, you may use age as part of your segmentation. You may also use it in your in your personalized experiences. But each fundamentally, each attribute will fall into one of these four typical categories. Uh, your users. Uh, each business is different. So, you know, I’ll talk about a few kinds of uh, users. But every business is different, so take some of this with a pinch of salt. Uh, you may have some people that can do everything, may have really siloed processes. But if we look at, you know, these three typical marketing cloud users, you might have a campaign developer who is a wizard of HTML and really competent with AMPscript. We have a campaign analyst who knows the raw data in and out and can find you out. Anything that you need to know can write amazing sequel and put together audiences. And you may have a digital marketer who’s got the eye of the the eye for opportunity and is really close to how customers can be satisfied with campaigns. Do you need to take your users into consideration when it comes to building out your data model? Your processes. Where does it live within your processes? Where does your data model live and what does it do? You know, an example process for that team that I’ve just highlighted, you’ve got digital marketers coming up with their ideas, passing it through to an analyst to work out whether the campaign’s even worth doing before it gets to the campaign developer building it and getting out the door, and then campaign reporting on strategic analysis. Each of these points, your data will not gonna influence. And tooling, you’ve got a load of tools in marketing cloud that most of you will have access to if you’re in marketing cloud already. You know, there are four key tools around SQL filters, journeys, and amp script. You know, if you’re already using Journey Builder, then you’ve probably got all of them. So think about what skills and what each of your users will be doing in marketing cloud and how they can interact with your data and the data what you present for them. So we’re gonna start working an example with the team, so I think it brings some of this to life. So we have a team of three digital marketers, um, one campaign analyst and one campaign developer, and we’ll just throw that onto the process that I’d laid out before. And you can see quite quickly there’s some points for bottlenecks and things that could make their process a little bit cumbersome and take a little bit longer to get things out the door. So the question is how does this scale? Because right now you’ve got a bottleneck. But by modeling your data in a way that is accessible to your digital marketers, you can empower more users and and really alleviate some of those potential problems. So let’s have a quick look at a new process that your data model can help, uh, introduce by making your data model more accessible and your data more accessible through effective modeling. So in this instance, what we’ve suggested is actually the campaign validation could be done by the digital marketers if we model the data in a way that is accessible to them. So in this instance, we could do a few tweaks to a data model and make it work through data filters, so there’s no need to be competent with code. It can all just be through drag and drop and self serve through data filters, which means there’s less stuff going to your analyst and less stuff going through to your developer. And, actually, only the campaigns have been proven or shown to have some kind of value or sufficient value get out the door, and you don’t have to wait for a campaign analyst to work that one out. So let’s start with how you would go about this and start thinking about, right, how can we drive value by putting the right code in the right places to make a debt model accessible through data filters? So first things first, you would start with your contactable concept. So whether that’s, uh, a lead or a contact or a person account or something that you’ve pulled over from Snowflake or whatever. You start with your contactable content, and you put that right in the heart and center of your data model, your person. And then you add a few details. You can actually start doing email marketing with just this. I mean, it wouldn’t be best practice to just email everyone who was opted into email, but, fundamentally, email marketing starts with find me opted in people with an email address, and segmentation is the art form of finding a smaller group of opted in to email people to deliver the same results so you can do something for everybody else. Then you have another direct extension. We have nothing too massive. But, fundamentally, now you can segment lapsing shoppers, people who’ve opted in and not transacted for three months. What about opted in, not transacted for three months, but historically spend a lot of money? You’ve got a few different campaigns you can run there with just these four attributes. And then add a few more. You could throw in a bit of personalization now and say, hi, Jason. It’s been a while. You used to spend a load of money with us. Why don’t you come back and spend some more? And it keeps going and keeps going until, eventually, you get to a point like this, where you’ve got a person in the middle, and you’ve got all these attributes glued to this individual that you can use data filters to go, right, find me the people who are like this. And if you factor this into the process alone, digital marketers can quickly and reliably get the size of an opportunity for a campaign they want to run-in Salesforce Marketing Cloud using data filters, which means that your campaign analyst will have more time to spend on more complex campaign opportunities or even bigger strategic analysis projects. You know, by and by putting your data into this sort of structured way of working with, you know, drag and drop filters, your campaign developer may have to write far less code because looking up against the data extension with AMPscript is the same if the if the data extensions are the same. So putting the time in to structure your data in a way that is really accessible can really enhance the way your business operates and really enhance the way that, uh, your users can interact with customers. Now it wouldn’t be a session with me if I didn’t talk about AMPscript a little bit in a bit more detail. Um, this is my AMPscript toolbox. Every marketing card developer probably has their own little AMPscript toolbox. These are my my go to functions for pretty much everything, uh, for a starting point. Um, and these are you know, if you wanted to move on something more advanced, you can add a for loop. But But if you do a AMP script and this is something that’s you’re not familiar with, uh, these function will get you going at a pretty good pace. You know, learning how to do lookups and apply logic, uh, to do to to make your campaign look differently or do different things for different users. Um, absolutely vital to getting you started. Um, I’m not entirely sure what v means. All I know is that it’s v, but to output the data, uh, all the decisions that you’ve you’ve come to with your your logic and your if else and end if. Um, formatting the way that the output appears. Title case is a good one because, um, um, if you’ve got data for more than one source, you’ll probably find that lots of people have or different systems will have things in all upper case or lower case. Some will have them in proper case. Um, these kinds of answering functions are super useful. And then treat as content is just a a useful one for developing because, uh, it forces out to be evaluated as it’s written. Um, so it’s a a really good catchall, uh, part of function to to use and leverage. Um, but if you’re already comfortable with that, so you probably know most of these already. But if it’s new to you, these are the the first sort of clusters. I realize I’ve slightly cheated by having if else and if as one, but we can’t have one without the others, unfortunately. Um, but as a rough example, if you want to personalize based on your structured data model, you don’t have to have all of the details in all of your audiences, which means that your audience the work on getting your audiences into a campaign is lessened, and you can actually use lookups to go and get the data you need. So in this instance, we’re saying, hey. You’ve not shopped with us since when however long. Are you still fully stocked? And the output of that will be, hey. You’ve not transcended us since whenever. Are you still fully stocked? But gonna move on to a little bit of how to scale and the things to consider when it comes to scaling, uh, now that we’ve talked about data modeling, uh, as a whole. So scaling is quite a tricky conversation and quite a tricky thing to to try and verbalize. Um, but first things first, if you want to scale, you need to standardize. You’ll look at any kind of mass produced product and it’s because lots of things are are standardized. It doesn’t have to mean make it simple. Doesn’t mean you have to make it boring. Doesn’t mean you have to stop thinking, but creating a really robust pattern that enables you to scale will mean that you are able to do far more quicker. It allows your people to do the stuff people are good at, which is think of cool new ideas, think of the ways that you want to deliver things, think of the experiences you want your customers to have, and let your technology do the things that the technology is good at, like the same thing over and over again. You should consider what you can automate and and how to automate things. Don’t have to automate everything for every hour, but if something needs to update at 10:00 in the morning every day, that’s really straightforward. Using Automation Studio and Marketing Cloud, you can do an absolute boatload. Automation is a massive topic, um, and I would love to get into the nitty gritty of it at some point. But if you are new to automation and marketing cloud and you’re looking for some way of understanding it more, um, there is a great book that was written early this year by Greg Gifford and Jason Hanshaw. If you’re unsure of how to automate in Salesforce Marketing Cloud, it is basically has become my bible for automating in Marketing Cloud, and it’s a recommended read from me. Um, so if you can get a hold of it and you do want to learn about automation, do get a hold of it. It’s a great tool and a great book to to help you leverage marketing cloud in a way that the documentation for the platform doesn’t necessarily allow you to. Um, but finding what you can automate, we you know, as it as it stands, I, in a current org, I’ve got about 400 things that automate every single day that would otherwise be manual jobs that don’t need to be because we’ve got the standard pattern for them, and they automate off the back of that pattern. So you can achieve a heck of a lot with a small team by automating in a truly strategic way. You should also be iterating. You know, don’t get stagnant. You know, I mentioned don’t be I’ll be iterable before. Um, lots of projects that start out scalable end up in monolithic state and unable to scale because the actual concept of scaling isn’t iterated on. You know, an example of this is my team right now, we’re on our fifth or sixth iteration of how we actually build a data model. Um, each one of these times we’ve picked up the process, we’ve made a few improvements. But it means that what was a, you know, couple of months process two years ago is now a couple of days process today. And it means that when we’re able to scale, we can do far more with our time, and we could deliver far more value. Um, I think the the big thing to to wrap up with beyond each of these points of the things you need to consider when it comes to to scaling is that being able to scale is as much a culture shift as a technology shift. You know, you can have the best technology available to you to to scale. You can have all of the tools in the world. But if you’re not in a position where you’re able to embrace the culture of scaling and putting scaling at the heart of what you do to make sure you’re delivering really good experiences for your customers, you’re gonna be limited. You know? No matter how good and expensive your software is and capable your software is, if you want to do things manually, you will never scale. So embrace what scalability is and what the opportunity it brings is. Um, Salesforce multi cloud and the Salesforce ecosystem is great for being able to scale. You just have to use it right and be prepared to embrace what scalability means and how to scale, uh, as a whole. And, yeah, that’s that’s me. That was everything I wanted to go through with regards to data modeling for for scale success and putting your customer experience at the heart. Um, thanks for coming along and listening to me, uh, talk about something that I’m really passionate about.

Speaker 0: Thank you so much, Jason. That was an awesome session. Um, lots of lots of great insight there for sure. Um, we are about out of time, so not sure if we’ll get to q and a or not. But if you guys do have questions for Jason, you are always able to reach out, uh, directly to him in the event chat, um, and and link up that way.

Speaker 1: Absolutely.

Speaker 0: So just wanna thank you guys again for joining us. Uh, one last special shout out to our sponsors for their support. Without them, Marjorie Mann would not be possible. Uh, so make sure you pop into their sponsor booths and, uh, show them some love. We still got some great sessions coming up, uh, this afternoon, evening. Um, so be sure to head over to the agenda to check out the full list. And, uh, thanks again, Jason. Thanks again, everybody. Um, enjoy the rest of your day.

Speaker 1: Cheers, everybody.