Personalization At Scale - Harnessing the Power Of Adobe Marketo Engage Data To Drive Engagement

Transcript

Thank you so much for having me. My name is Kiara Riga and we’re going to talk a little bit today about harnessing Marketo data for better personalization to drive engagement. So a little bit about me. I am a two time Marketo engaged champion and this year I was lucky enough to be selected as champion of the year. I’m also an Adobe subject matter expert. I earned that title when I helped write the new version of the MCE. I am the marketing operations manager at Digital Shadows. I’ve been working in Marketo for over six years and I’ve been certified since 2017. So today what we’re going to cover is first, actioning and harnessing implied data in Marketo engage and then how we some ideas for personalization beyond the obvious of tokens and whatnot. Some use cases for that personalization and then we’ll go into how to automate that. So what are we going to cover today? We’re going to go over harnessing and actioning implied data in Marketo engage. Some ideas for personalization beyond the obvious of tokens and whatnot. Some use cases for that personalization and then we’ll go into how to automate that in order to make your marketing that much more efficient. So first let’s start with collecting that implied data in Marketo engage. So what is implied data and what do we hold in Marketo? Marketo holds such a wealth of knowledge about our prospects and customers, right? But a lot of that isn’t captured in a field. Every action that someone takes when they interact with your marketing tells you something about them, right? And only a small fraction of that gets actually captured into a field. So what am I talking about here? There’s tons of examples, but here’s just a few that I thought of how they prefer to consume content, right? So if you have a prospect who’s constantly only interacting with your webinars despite you emailing them blogs and white papers and things like that, they probably prefer to consume their content via video. Same thing for if they’re only consuming blogs, right? They’re telling you some really important information about who they are and that’s not really captured in a field. Same with the problem they’re looking to solve. Your product probably has several use cases and your content probably only focuses on one or two. And if they’re consistently interacting with content that only focuses on one specific use case, that’s probably the problem that they’re trying to solve with your product. You can also see where they’re at in the buying cycle, right? If they’re interacting with a lot of top of funnel, middle of funnel, bottom of funnel content, that’s telling you something about if they’re only interacting with very high level ungated content at the top of the funnel, they’re probably not ready for a meeting with sales. But if they’re doing your free trial, for example, or requesting a demo, obviously, they’re a lot farther down the funnel and more ready for an opportunity. Which product they’re interested in. So if you sell more than one product and you see them interacting with content for just one product, they’re probably only interested in that one product, right? But that’s key information for us to have that isn’t always captured. And then lastly, if they’re current customers, you can probably see what areas of the product that they need growth in. If they’re consistently interacting with content you put out and interacting in the product in only one certain area of the product, that’s going to be key for you to try and nurture them to use other parts of the product and get a better chance for them to renew. So I said lastly, but these are lastly for my ideas. The sky’s the limit here. You can decide on anything that you would like to see. The sky’s the limit. They’re telling you so much. So sit down and think about what kind of implied data you have in Marketo. So we talked about what this data is. Why is it important? Why do we want to capture this? Because the more we know about someone, the more personalized we can be with our outreach. And I’m sure I’m preaching to the choir here, but I’m going to go over some benefits of more personalized marketing. We’ve got better engagement, higher conversions, lower unsubscribe, shorter sales cycle, higher close rate. I mean, the list goes on and on. I think every marketer knows that the more personalized, the better. So when my organization rolled out this process that I’m about to share with you today, our email conversion rates actually more than doubled. And they’ve held steady for over a year since we rolled this out. So that’s been so huge for us. And more than that, this process is much more automated than what we were doing previously. So we’ve both saved our marketers time as well as increased conversions. So best case scenario. All right, so now we’ve talked about what the data is and why we want it. Let’s talk about how to actually capture it. So collecting the data. Now if you attended my session last year at the Skill Exchange, this might look a little bit familiar. And that’s because I’m going to be taking the product interest scoring use case that I presented last year and using that same strategy because product interest is really just a type of implied information. And so you can apply that same process to anything, right? So first you’re going to want to create a supplemental scoring model. This is not going to replace your MQL scoring. It’s just going to be another one on top of it. Then you’re going to determine which categories you want to score. In this case, and for this presentation, I’m going to be using use case as the majority of what I’m presenting on. But again, sky’s the limit for which categories you can score. And then you’ll want to break those categories into different buckets. And then assign a code to each category, right? So let’s say you have three unique use cases you’d like to score. We’ve got email marketing, data visualization, and content personalization. Your codes could look a little something like this, right? You could have EM01, DV02, and CP03. Your codes can be anything. They don’t have to be these ones that I’ve put here. The only thing that I suggest is make sure that it’s something like a mix of letter and number or something that you know will not show up organically in a program name. And you’ll see why in a minute. Then you’re going to add the corresponding code to the name of the program that corresponds to that code, right? So you’ve got a content personalization program. You’re going to add CP03 to the name of that program. And same thing for your URLs of web pages if you want to be scoring Visits webpage. Now I’ve heard quite a bit from people that SEO teams really don’t love changing web page URLs, which is totally fair, of course. So I have a few suggestions for that if that’s the case for you. If you can fit it into your UTM strategy, start adding it to your UTMs. I think that’s a really great easy win. If you can’t fit it into your UTM strategy, then I suggest creating a program for those web Visits. And then you can do all of this based on program membership. Obviously that’s more manual, but it is a way to capture that data. And then lastly, you’re going to want to create a score field for each category and decide how much you want to score for each interaction and then create score tokens for it. Now your score value can be whatever you’d like. When we first rolled this out, we had it mirror our MQL scoring model, right? So we let’s say webinars were 40 points. We would give 40 points for an interaction with an email marketing webinar in the email marketing score, if that makes sense. But we realized that that was really confusing for sales. They kind of, the numbers didn’t really mean much to them, right? And so that was tough. And so we ended up switching to just one point per one interaction. And that became a lot more clear to people who don’t kind of live and breathe this stuff. They could understand, you know, okay, this one has like three scores that meet or three points. That means you’ve interacted with three different pieces of content in that category. So once you’ve set all of that backend stuff up, we’re going to go on to how we actually collect the data, right? You’re going to have to set up some smart campaigns. So in your smart list, it’s going to look a little something like this. You’re going to trigger on program status changing with the program name containing that code you made. And then you can have your new status be only specific ones. If you only want to give scores, for example, like you don’t want to give scores for just registering or no showing a webinar you really want attended or attended on demand. You can do anything you’d like and kind of filter it down any way you’d like. Then in the flow, you’re just going to change that score to your use case interest score. And that’s it. That’s all you’ve got to do for program membership. And then for web activity, it’s going to be the exact same thing except you’re triggering off of that URL changing. You could also add a constraint here if you do it via UTMs to look into that UTM information instead of the web page containing. But that’s kind of how you would do it. And if you go with the program membership option for web activity, then you’ll just do everything like the previous slide. So that’s it. That’s how you set it up. And then once you’ve got it all set up, you’ve got it all running, you probably want some data. But you also probably don’t want to backfill all of the data you’ve ever had in Marketo because it’s going to take forever. And it’s not super relevant what someone did like three years ago, right? So backfilling that data from every program you’ve ever done is more trouble than it’s worth. But an easy way that I like to backfill whenever I roll out a new category or something like this, I will pick the top programs in each bucket. And then I will do a quick smart list like this and say anyone who’s a member of this program change their score. And you want to have like about the same amount of possibilities for scoring in each bucket so that you’re making sure it’s even across categories. And then you can start to backfill that way so that you have at least some data as a baseline and you’ll continue to collect data as you go. So how to make sense of this data. Unlike demographic or behavioral scores, the data alone doesn’t mean a ton, right? If I’m looking at this, a data visualization score of three and an email marketing score of five doesn’t mean a ton to me where you could look at an MQL score and see, okay, they have a score of 70 and our MQL threshold is 75. So I know what that means. But these numbers don’t mean a ton on their own. The value is really in the comparison here. So your last step to making this actionable is having a field calculate the lead’s top score leader contact. Right. So this is best accomplished in Salesforce, or any whatever CRM you’re using. And that’s how I’ve always seen it done. Let’s say you don’t have those resources, though. A way I suggest you just get this off the ground before you can get those resources is to use segmentations, right? You can say, okay, I want my threshold to be five points, I don’t want anyone having a top use case unless they’ve got at least five points in that category, then you rank your categories. So with these three categories, let’s say I think email marketing is the most important than data visualization, then content personalization, you’re then going to create a segmentation that says if person has a score higher than five in email marketing, put them in the email marketing bucket, and so on and so forth. For all of your buckets, it’s not going to be ideal, it might capture some people as their you know, second top comes in as their first, but you know that you’re capturing people with a high interest in those topics. So it’s a good way to just get started. So using and automating this data. There are tons and tons and tons of applications for this, I keep saying sky’s the limit today, because it really is. So a couple of really big and quick wins you can have is sharing this information with sales and BDRs. Right. So first of all, a tailored sales pitch, imagine giving this screenshot to a BDR, instead of just sending a BDR and MQL and saying, like, go for it, do whatever you’d like, you can give them this implied interaction data and say, okay, their top use cases, email marketing, their top content category is webinars, and their top product interest is marketing automation. So then you can have the BDR send them a webinar that’s about email marketing within your marketing automation product, rather than the BDR going in totally cold. So this can make a really, really tailored and personalized sales pitch to hopefully increase your BDR conversions and meetings. Similarly, you can have alerts about current customers, right. So this one, for example, I was interacting with tons of marketing automation content, I don’t have the marketing automation product. So it’s probably time for my salesperson to reach out and start talking to me about marketing automation, right? It’s probably a pretty easy upsell. And the way you would set that up is just using a threshold, right, you can decide your threshold is whatever you’d like, let’s say, five pieces of content in two weeks, whenever that person has a score, or their score has changed by five in two weeks, then you’ll send the sales owner an alert, and that’s it. And you can help the sales team be really, really aware of what their customers are doing. Then we can talk about automating this data with dynamic content and snippets. This is also a pretty quick and easy win once you’ve got all the bones of this setup, right. So you can create segmentations based on that implied data, that’s going to be your first step. And they’re easy segmentations once you’ve got that field for top use case, for example, because it’s just going to be top use cases, email marketing, okay, their segment is email marketing, so on and so forth. And then you can have like a dynamic suggested content snippet and all of your emails are all the emails where it makes sense at least, right. So then, instead of just having a secondary CTA that you guess about, you can send like a webinar invite and at the bottom have, okay, here’s the content that I know that you’re interested in. And you just refresh that snippet every month, every two weeks, whatever makes sense for you, and add new content is in as it goes. But this is a really good way to send people really personalized content at a broad scale rather than having to create a different sound for people who like email marketing versus data visualization versus content personalization, you can just have one snippet to do it all in every email. Similarly, with dynamic follow up pages, you can do the same thing. I know a lot of us are, you know, our follow up pages will say thank you for registering check out this white white paper while you wait for the event. But imagine you could do thanks for registering and check out these three white papers about a topic that you already told me you’re super interested in. That’s going to be huge to be able to keep nurturing people before that event even happens. And then the last idea that I had here is dynamic email showing the preferred type of content, right. So I know a lot of us are creating like the same topic will have a podcast, a webinar, a blog and a longer form piece of content, we could have more we could have less. But we know that people learn differently, right? Not everyone likes to consume stuff by video. I personally hate watching videos to learn things. I want to read it. But I have friends who are the exact opposite, right? And so if we already know this stuff about people, because we know what they’ve already interacted with us on. So we can just take a snippet or excuse me a segmentation of top way to consume content. And if they’re interested in webinars or video, you send them the video version. If they’re interested in listening, only you send them the podcast. If they’re interested in reading, you send a blog or white paper, right? So you get in front of them with the mode of communication that they know they that you know, and they know that they like to receive. So nurture is another really great way to automate this data. This was where we really started to see gains in this process. So you can revamp your nurture streams by creating a stream for each for example, use case. And then using the top score field that you created, you put people into different streams. And then from there, you can get so creative with it, right? So you can use dynamic content to take other implied data categories, and make an even more personalized experience, right? So you’re already sending content about email marketing. But what if you wanted to send content about email marketing, and you make sure the webinar people get a video and the readers get a blog or a white paper, right? You can do all of that and get so into the nitty gritty and so so personal with it. And the sky’s the limit with how much you can overlay this stuff. So that’s really amazing. And then once your prospects and customers exhaust content, throw them into another stream that you know they’re interested in. It might not be their top, but if they’ve shown over your threshold of score points for that other stream, send them in, give them more information from there. So that’s a really, really great way to automate this. Because I know for a while, especially for ABM, our marketers were sending like we had this information. And we could kind of guess what people were most interested in. And so they would do a different set for email marketing, data visualization and content personalization. That’s really tough. And it takes up a lot of time unnecessarily. So this drip nurture can help you be super, super personalized, while also scaling. So what else can this data tell us? Tons again, you’ve heard me say it enough, the sky’s the limit here. But here’s some examples. We’ve got where to put your content creation resources. So the people who are creating content with us, their time is finite, we wish they could just crank out content 24 seven, of course, but that’s just not realistic. And so you want to make sure you’re putting those resources to good use, making sure the content that you come out with is resonating, right. So um, if you have one use case that 75% of your database is interested in, it’s probably best to keep cranking out data in that use case. Similarly, if you start looking at preferred mode of content consumption, and only like 5% of your database even interacts with podcasts, put it to the wayside right now, you know, so we have all of these different, different data points that we can now start learning more about the people we’re marketing to, and help develop an ICP and then use that ICP to then decide what content really fits in here. Then you’ll know the parts of your product that resonate the most, right? If there’s one where a use case just really blows all of your other use cases out of the water, and even if it’s not everyone’s top, like almost everyone in your database has at least a couple points in that score. That’s probably the part of your product that resonates the most in the market. And so when you’re thinking about other channels, like digital marketing, you know, display ads, things like that, where you won’t necessarily know what that person is interested in, when you’re doing your marketing, take this into account and start using that and making sure you’re speaking to those use cases with that kind of marketing, because it’s clearly a hook for other people. And then you can, and this might be my favorite application for it, you can overlay this information with meeting and opportunity data, right? So you can take, let’s say, everyone from email marketing, who becomes an MQL eventually goes on to be an opportunity. That’s great. But then what if those opportunities aren’t closing, right? That can give you some really valuable information about your product. If you have a ton of people who are super interested in you for one use case, and then they don’t end up buying, that probably tells you something about the product. And it probably tells you about where you should be focusing your time on when you’re, you know, prospecting to people. So you can, the sky’s the limit here, really, but you can really dig into your existing meeting and opportunity data to try and find trends and use that to help with your prospecting, with sales, etc. So I just threw a ton of information at you. And you’re probably thinking, okay, this is all great, but how the heck do I sit down and get started? So first thing you’re going to want to do, this is an optional step. But I think it’s a great one. If you have skeptics on the management team, which I think we all do, you know, marketing ops is a finite, finite resource, same like content creation, like I was talking about earlier. And so if you need to prove out the concept, to ensure that it’s something that, you know, works for the team, do an A B test. And just like I was saying for backfills, where you pick a couple of key resources that you think fit really well into one bucket, pick, you know, people who have interacted with a specific use case, send them, you know, they’ve interacted with three to five pieces of content within that use case, those key pieces of content that you’ve picked out, send those people in the test group, the very tailored content to that use case, and send more generic stuff to the rest of the test group and the control group. And that should be able like you do a couple of those, you should be able to show your your team like, look, these are the people who got the really targeted email, and look how much higher they converted, right. And that’s a really great way to prove out the concept. So once you’ve proven out the concept, here’s some quick ways to get started. You can work with your web team to scrape the website for all of your different URLs and use that to categorize your URLs. You can also export all of your smart campaigns. And that’ll help you find all of your programs in order to bucket those. And then the last step, or a last idea I had, if you have an offer field in Salesforce, or in your CRM somewhere, you can start to sort by offer. Unfortunately, we can’t export offers from Marketo that I know of. But you can do this in Salesforce, right. So you take that whole list out of Salesforce and sort by offer instead of having to do each individual program. So lastly, we talked about this earlier. But if you don’t have that CRM or dev resource, and you really want to just get started, then use segmentations to calculate that field for the top score category. So let’s go over the process in review, you’re going to want to start by thinking about what people are implicitly telling you when they interact with your programs. I gave a bunch of ideas. But you can pick whatever you like, and whatever makes sense for the business that you’re working on. Then you can prove it out with an A B test, if need be, you’re going to set up your supplemental scoring model. And then you’re going to start automating this, sharing this data with sales, and weaving it into your reporting in order to get deeper insights. So that’s all I have for you all today. Again, my name is Chiara Riga. And we’ll move on to the Q&A. So feel free to send me any questions you have. All right. So thank you for joining me, Chiara, and congratulations on being Champion of the Year. Thank you so much. Yeah, that was a great and very informative presentation. I’m going to ask you some questions that we’re taking from the audience right now. So the first question comes from Veronica, who first of all wants to say, I love how insightful this process is. Now the question is, how do you recommend sharing this information with sales? Yeah, so you can kind of everyone knows their own sales team best. So I think you can kind of, you know, figure that out for your sales team. But I think what’s worked really well for us is having the scores kind of lower down on the Salesforce page or wherever your sales team works out of, and then having that top score field just like readily available for them right next to person score or any other scores that you’re kind of sharing with your sales team. Great advice. Thank you. The next question, it’s actually a two-part question or two related questions from Chelsea. And Chelsea is saying, I really love this, Chiara. Giving sales another score to monitor would be tough for some organizations. How do you recommend rolling this out to sales knowing that it’s yet another process change? And then I’ll ask you part two, which is, it’s a similar question, but for product marketing. So they already identify their audience. How should we share this information with them knowing that they don’t have any Marketo Engage experience necessarily? Yeah, so as far as the sales team goes, I think you’re right. Another scoring model is a lot for, I think, most sales teams. And that’s why, like I said with Veronica, kind of leaving those actual raw scores and numbers kind of lower down the page is really helpful. And then just put that top field right there, like that top product interest, top use case, whatever, put that right somewhere visible to them and just give it as something extra. I also like to start with the BDRs and make that field a tokenizable thing if you have sales automation. So if you could use a token to say, hey, are you interested in top use case and have that in the subject line, right? That can be a real time saver for BDRs. And so I think you start there with the quick wins, with the automation. And once you prove out that concept, then you can start rolling it out more broadly. I think product marketing is kind of similar, right? The product team’s not going to be in Marketo and digging up that stuff. So I recommend if you have a monthly content meeting or somewhere on a regular cadence where you meet with product marketing, I think it’s best to have a standard set of reports that you can show on every call and show, here’s what’s really resonating and give them the data in a way that makes sense to them. Because them going into Marketo, it’s just going to be whoosh. But if you show them reports and show them with numbers, with data to back everything up, you’re, I think, going to get a lot more success there. Great. Thank you. Here’s another question. How often do you review and modify the personalization scores to identify what is working and what’s not working? Yeah. So I think in the beginning it was really hands on. I was spending a lot of time really digging into it and just making sure like operationally nothing was breaking and that things were both coming in correctly and being utilized correctly. After that, I’d say on like a quarterly or a half yearly basis, I’ve been in kind of a yearly score change flow recently, but I’m trying to move more to every half year because I think one year is like a little bit too long to go. Cool. I’m wondering over, just thinking out loud, over time, maybe we at Adobe can apply AI and ML to make this more intelligent in the future. Oh yeah. That would be incredible. Yeah. It would be incredible. But maybe someday we’ll come up with ways to do that. Here’s another question for you. Actually, come to think of it, Kiara, how did you go about figuring out which inferred data points you want to capture with this process and for somebody else, how many different inferred data points would you recommend capturing? Yeah. So I really started with what is it that, A, as an organization we are trying to kind of share with a broader org and what as an organization are our goals. And then within the marketing team, what is their goals for personalization? And a lot of that, nobody is going to come to you with a silver platter of, here’s how I want you to personalize everything. It’s going to be little conversation snippets that you hear and kind of file away in the back of your mind. And so I recommend just staying really actively listening in calls like that and figuring out those kinds of things. And that’s sort of where we started really. And I know that’s kind of a fluffy and vague answer, but I think just really figuring out what is valuable to the broader team and really making sure you spend time with your marketers around you and understand their challenges and what can help them succeed is really great. In terms of the number, I think that you could use as many as you can manage, I think is the correct answer here. Nobody needs data just for the sake of it. And this is all going to use more Marketo processing. So you want to make sure that whatever you’re rolling out, you have a plan for it and you’re going to use it. That said, I know I said the sky’s the limit about a hundred times in my presentation, but again, the sky’s the limit. If you can manage 10 scoring models, you can roll out 10 scoring models. I don’t have the bandwidth for that. So I have just a couple, but you can really do whatever is most valuable for your org. Yeah. And my product manager answer would be as many as you need. Yes. That’s a better answer. Yeah. Well, what you said is it makes sense. Just make sure you take the time upfront to understand the needs and the challenges of your marketers. Right. And that way you’re doing what’s really useful. Yeah. Great. Well, you know, I think that’s all the, those are all the questions we have right now. Really appreciate your time. Once again, congratulations and looking forward to chatting with you more in the future. And I know you have an Adobe Summit presentation too. So want to give a quick plug to everyone to attend the Summit very soon in the near future. Thanks a lot, Kiara. Take care.