Data Stream Prioritization

Transcript

My name is Jazz Singh and I’m a senior customer success architect here with the Dobie. And I help our customers overcome their existing challenges or roadblocks to achieve their marketing. North Star with the Adobe Digital Experience stack. And my expertise, like Katie mentioned this within the campaign orchestration and customer journey domain. So let me do this. Let me the camera and let spoke just on the content. Okay. So today we’ll be talking about data and specifically one of the challenges around around data. That’s too much data, right? So that is data stream prioritization. What do we do when we have too much data or agenda? First things first, right? Let’s we’ll talk about what is data stream prioritization. Then we will go over. Why is that important to even consider? And we should go over how to achieve data, assume prioritization. And last but not the least, what are the best practices around data synchronization? So by definition, interesting prioritization. It’s the process and ability of a system to prioritize and manage data from different sources based on certain predefined rules or criterias. And this criteria, it can vary depending on a few factors. For example, the significance or importance or the urgency of that data. The Adobe Marketing Technology stack, it frequently encompasses that collection of data from different sources, a very diverse set of sources such as CRM systems, loyalty programs, the web data, mobile data through apps and SD cards and other digital points, depending on the vertical or the industry that you’re part of the enterprise of the OF and this data acquisition process, it can occur in either batch or real time formats. Real time. Okay. Streaming and different types of data streams may include first party data, second party interactions, web visits, ad impressions, and many more. Now, there are a few things that make up that data stream prioritization to what it should be and to be able to explain this concept a little bit better. I’ll be using Adobe Experience platform as a central point of reference, so that would make a lot more sense since it has a lot of data sources coming in in order to create that complete customer profile. So first thing, real time data collection whenever we talk, CDP is specifically APB, we talk of real time because it has some of the capabilities, some other data management platforms or data platforms they don’t have. So the Adobe Experience Cloud, it often involves real time data collection for different sources, like I said, websites, mobile apps and other digital points and these different types of rescues may include customer interactions and ad impressions and more. So it is the prioritization criteria. So the criteria or the rules for these prioritization, they can vary and they can be based on the importance of that data, the urgency of that data. For example, customer interactions during critical marketing campaign that’s going on might be assigned a higher priority than the routine bed. But at this best example could be like the holiday seasons or Thanksgiving sales. Next is business rules and objectives. So the prioritization is typically aligned with the business rules and objectives and different businesses or enterprises that may defined these rules regarding whichever data should be processed or analyzed first based off their own organizational and strategic goals. Next is optimizing the resource utilization. That means basically that data stream prioritization, it’ll help optimize your resources utilization and ensure that critical data is being processed promptly. It can also prevent delays in processing high priority data, allowing your enterprise in the business to respond even quicker to some customer interactions that just pop up. And overall market trends. Next, big customization and flexibility. Now, like I mentioned, Adobe Experience Solutions, they may provide customized customization options and allowing you to tailor your data stream based on whatever your unique requirements are for particular use case and the users they might be able to configure and adjust their priority settings to adapt whatever you have. Changing your business models based on, let’s say, the holiday season or business goals. Yes, so on, so forth. Next is integration with other Adobe Solutions. So the Adobe Experience solutions, like I said, it consists of API, Adobe Analytics, campaign Target and many more. So prioritization settings. They can extend across these solutions to ensure that seamless flow of data and insights. So when we talk about best practices, I’ll talk a little bit more deeper into that. On why it is important now let us dive into why is the testing prioritization even worth considering when it comes to campaign orchestration and customer journey? So prioritizing that data streams, it becomes essential and you need to prioritize the construction and updating of that customer profiles and segments for that effective campaign orchestration and activation. You need to ensure that the most pertinent, precise and timely data is given precedence, especially for those real time use cases. The objective of prioritizing data streams, It’s to optimize the quality and the relevance of that customer data within the platform. But this not only doesn’t stop you’re like an intern in a lawyer or to elevate your customer insights and tailor your marketing efforts to enhance overall customer experiences. And that way you’ll have the most valuable and up to date data at your disposal. And once you have that, the marketers have that the activation and the personalization just you can scale it very well. All right. So we talked about what data prioritization is, what’s important, but let’s talk about how to achieve this. The data protection now, the process of achieving a data stream prioritization, it’s basically necessary. It’s right, a blend of getting technology. You need strategy, you need tools. So we’ll begin by delineating the metrics and the criteria that ascertain the significant science of data streams with your specific context. Of course, now these may encompass the variables like the customer engagement part, the transaction values, or if the use case has to be real time. So there’s some real time relevance factor there or any other indicators that is specific to your business. So what you need to do is you need to harmonize that data synchronization with your overall arching business goals and objectives. Now these alignments, they’ll ensure that your strategy, your prioritization strategy is is in sync with your with your org or enterprise, broader strategic initiatives and goals. But you don’t stop there, right? You have to regularly you have to put like scrutiny on that performance of how that strategy, that data suite prioritization strategy, working out for you and continuously assess how well your systems will respond to this dynamic conditions, Right. Marketing. It’s interesting and there’s changes coming all the time. So it has to be dynamic. It has to be able to respond to the management conditions and better align to your business goals or not. And you can always adjust as necessary. Right? So if you see some change coming up, you should be able to adjust and necessary and engage in a continuous process of iteration and reiteration and continuously improve your data and their data stream prioritization of strategy. Now this will involve incorporating the feedback, right? So you need feedback. You need to analyze the performance of the metrics and then overall adaptation to those changes in the business requirements. This approach is both flexible and adaptive and is crucial for have the have that prioritization strategy effectiveness over time. Now, I know this can be overwhelming or tedious process, right? So far what we talked about, what it is, why is it important, how to achieve it. It’s a lot of strategy work, but we’ll talk we’ll talk through these on how to start well, with a few pointers, one by one, number one being source reliability. Now, there are so many different data sources that will vary in terms of their accuracy and reliability, like their ability. You can ask any data engineer that’s their first problem, like they’re not getting the IDs that are needed or the data is corrupt by the time it reaches the management system. The CDP. Now some sources they may provide trustworthy data than others. So that’s that’s another reason to prioritize one over the other and allow the CDP to assign higher priority to sources that have been historically more accurate and reliable than the others. The second one is timeliness. Now, the freshness of that data is essential for effective customer engagement. Why? Because that timely data update and it’s crucial for understanding and responding to a customer behavior, you need that latest data, especially for the real time use, case it and like I said, to ensure that real time or near real time from certain sources is processed and incorporated into customer journey profiles way more quickly than historical data. A little bit double click into the source reliability is the actual data quality. So the prioritization to make considered the quality of data provided by different sources could be, you know, it could be high quality data free from errors and inconsistencies overall, and it would prioritize over low quality data. And the cdhb itself may have mechanisms to identify CDB in this case being AP rate it have its own mechanism to identify and prioritize the data sources that are consistently providing accurate and informative information versus some stale ones. Another thing to look at is the business rules or the overall criteria, right? So the rules there often aligned with whatever your business or goals are at hand. So for example, if you’re making a marketing campaign, go the data from sources related to the campaign interactions might be given a higher priority than some other ones. And we’ll talk a little bit more. I have an architecture and a use case example. Later in the slides we’ll talk a little bit more at an example at hand. It’ll make more sense. Then the next one is a user defined rules. So the category itself, it may provide you the flexibility for your own users to define their rules for the beta. Prioritization like this will enable your team to tailor the system to their own specific use cases and preferences. Last but not the least, is the integration with the third party system. So the data it may be a source from various internal or external systems, right? So the prioritization will extend to integrating data from third party systems as well. So we need to ensure that the relevant external data is considered in building those customer profiles. Let’s dig right into the challenges. So while data source prioritization, especially with the CDP, is it’s essential for managing that diverse data streams, it may comes with its own challenges. So first one being the data quality variability. Now different data sources, they may exist exhibit different varying levels of data quality. Not everything is going to be at the same level. So we need to ensure that consistency and reliability across the sources that can be and that can be challenged. It’s not easy. So that’s why it’s one of the top targets that you run into because low quality data from, you know, certain selected sources that it can negatively impact the overall accuracy of customer profile. The next one is real time data integration. So prioritizing the real time or near real time data, it can be challenging, especially when dealing with large volumes of data. So that is something you need to look into. You need to ensure that timely integration and processing of that data without compromising anything on the system performance. It’ll require robust infrastructure and technology and quite honestly, AP does a really good job at that. Next one being data governance and compliance. So complying with that data governance and privacy regulations, while prioritizing data can be complex. So that is something you need to look into. Like do do you meet the regulations or you need to prioritize the data. So as there’s always meet the regulations and there’s always a balancing that’s needed for prioritization with this requirement, and you need to adhere to that data protection laws too, because it poses a challenge for the CDP rate. Yeah, but the data governance within the CDP and the labels, CDB, A.B. does a really good job of that. Next is dynamic business environment. So different businesses, they often experiences changes in their own internal strategy, use campaigns and overall priorities. So you need to adapt your data source prioritization of data screen prioritization rules to reflect that dynamic business environment. And that can that can be a continuous challenge because business requirements, especially my marketers on this call, they can they can confirm that it keeps changing costly. So it’s a constant challenge at the same time and ensuring that the prioritization strategy it remains aligned with that evolving business goal is it’s crucial. Next is integration complexity. So the KPIs, they often they’ll integrate with various systems, both internal and external, and managing the complexity of that. Integrating data from different sources requires a great amount of planning and coordination. And this is where your EA’s and your data engineers, they come into play. The changes itself that happen in the source systems or additions of overall new data stream, it needs a trigger to be make adjustments to your prioritization strategy. So that’s a constant challenge as well. But if everybody is on the same page, especially like I mentioned, the years and the delay, this can this challenge can be definitely smoothed. Next one is user adoption and training. Now what that means is as users, they’re responsible for configuring and managing data source prioritization goals, and they need to understand the capability of CDP thoroughly because if they’re not trained well, if they don’t know what the club is capable of, it’s really hard for them to actually manage the the data source prioritization well. So the like I said, adequate training and support, they’ll be very crucial to ensure that the users who are managing the data platform, they can effectively utilize and prioritize the data using the features of the CDP. The next one is scalability. So as the volume of the customer data grows, which certainly I would say 99.9% cases, that will be the case, the volume of the customer data will grow. You have to ensure that your prioritization strategy in mechanism, it can in fact scale effectively and does not become a challenge down the road. So scalability issues, it may arise by dealing with increasing number of data sources or larger datasets. So that is something you need to almost like nipped in the bud. If you think that is something you will face down the street when your data scales interrupts. Last but not the least, data silos. So in enterprises and then organizations with multiple departments or business units, data may be siloed. And we see this in practical examples all the time. Marketing has a different data on loyalty as a different data. The CRM systems, they’ll have different level of data. So data siloed can happen and are very common. So our job is to ensure that that relevant data from different parts of organization is in fact considered in that prioritization process. And that can be challenging. So that’s why I tend to call out here. But after like you see all these challenges, but overcoming these challenges will require a very robust technology and well-defined process, and it’s a proactive approach to be able to adapt to these changes in the business and the landscape that’s happening. So what you need to do is regular assessment updates to your prioritization rules and overall, just focus on that data governance part as well. And they’re essential for that successful data source prioritization strategy within a CDP. All right. So getting a little double, taking it a little bit more practical examples on what this is data stream is about. So this is how a typical architectural representation is with Adobe Experience platform being at the center of the data flow and you can see the data flows from left to right, so you can see the dotted red box. I call it out just so you see it on the left. Those are the potential data streams or sources that get get fed into the app. As you can see, these could be either web or mobile data streaming with SDK or even batch data sources like email, data, CRM, MDM, loyalty and whatnot. So what we can do is last we let’s do a use case example using this architecture that will make more sense on how data stream prior to what it is and how it works. So the use case is had a typical use case, right? A non customer to a gourmet restaurant chain. It’s being presented with some targeted ads on about a certain menu item and the user or the prospect in this case they’ll click on the ad and based on the UTM parameters that are collected from the ad, the user will be presented some relevant creative throughout the ordering experience. So based on the referring ad, a certain creative will appear on the here page and overall menu list item. Now when the user will scroll down the menu. The menu item from the ad, it should go on top of the list. So a very typical use case. I used a restaurant example because this has been resonating with me lately because of GrubHub and DoorDash. It’s just so easy to order. So and I’ve seen this practically happen. On the next slide, we shall look into how the data flows given in this example and what data source need to be prioritized for the simple use case. All right. So like I said, based on the use case, the data attributes needed to show that relevant menu item at the top of the page, we will need the particular UTM parameter that are collected on the web through that ad. And since this is a new or unknown user or customer, like I said, truly a prospect at this point we don’t we don’t need or even have any of that data from the customer loyalty or previous interactions. This is a very net new prospect for us. So hence there’s only one data source. So we only get to prioritize this. And before I get into a little bit, make this case a little bit more complex, I want to quickly walk through the flow of the data, how this interactions works. Going to AP and the target and then coming back. So just follow the steps. If you see the bullets in the yellow colors, those are steps number one is the user will see a paid media ad on a paid media channel. The ad includes a link that has that UTM parameter attached to the URL and it’ll persist through the restaurant website. It’ll be excuse me, it’ll be set up that way. So that UTM campaign code, it’s set to add to that data layer and it gets picked up with the rest of the data layer into the initial load of the page and the first call to the adobe edge. For that, I say again, personalization. So the visitor context now is a set at the edge, the edge segmentation engine. It’ll qualify the user for any segments based on that event data received and in the request along with the context data. So here the user gets qualified into the let’s call it a new customer segment based on those UTM parameters at the edge, the experience will start orchestrating a call to Adobe Target Edge with visitors Context now evident. This is number five. Sorry, I should have set the numbers before biology’s number five within the Adobe Target Edge. The target receives user segmentation from that segmentation and to start analyzing and add any additional parameters sent from the request from here, the number six, when the experience is identified for the product prospect, the content itself, the selection that is made, it gets delivered to the experience edge. And finally, number seven, the experience says then finally delivers the content to the requesting application for the rendering. Now this is, like I said, very simple use case. We only had one data source, so we only prioritize that one. But let’s assume that this was a known customer with the previous orders and interaction and related loyalty program data or even previously presented offers the whole nine yards, basically. Now, in that scenario, we will start prioritizing the data streams to be brought into API based on the criteria defined by the new use case, for example, new use case, meaning if we need a segment for returning customers that free presented, let’s say offer X in the past, we will pull in all the data streams except loyalty data because we don’t care for loyalty data and this particular segment is not required. But if the use case was a little different and our segment were able to pull all the gold level returning customers that were presented with offer, why in the past and days in that case, we need to bring all the data sources that I mentioned earlier, loyalty interactions previously previously presented offers and whatnot. So yeah, but like I said, Rome wasn’t built in a day, right? So neither was this use case. So you don’t start with that. You start with the MVP and we’ll talk a little bit more about that on how to approach this data. Set the same prioritization strategy step by step in our next section, which is basically the best practices, right? So like I said, implementing that effective datas impact decision is very crucial and you need to maintain that accurate and valuable customer profiles. So best practices first things first, define that clear prioritization criteria. So you need to precisely outline the standards for prioritizing data sources. You need to engage the stakeholders from different departments to guarantee a comprehensive and comprehensive list of priorities. Next is assess and monitor data quality. So consistently you need to evaluate the accuracy of that data from diverse sources. You need to introduce data cleansing and validation procedures to rectify any problems. If you see for the related data or any inconsistent or inaccurate data, the next one is consider real time and batch processing. So you need to have a need to be able to understand the difference between the real time and batch processing requirements according to your business needs. So I want to take a moment here and go back to the example that we talked about. So here, so in here, the first use case was very much real time, right? It needed real time data, but the following complex use cases that I talked about which needed, you know, certain record member customers in the past X days, it’s certain X offer that is not really that real time and that can be suffice with just based data. So that is the kind of prioritization you will be able to do as a best practice means based on the use case, selecting the data data stream prioritization. So you need to strike that balance right between that necessity of that real time versus the corresponding batch processing based on what infrastructure also demands. Next one is aligning prioritization with your business goals. I can’t stress this enough. You probably heard me say this a few times already in this session. You need to guarantee that the prioritization of that data source, it should be in harmony with any existing business goals and strategies. And you also have to periodically reassess and revise and adapt to these prioritization goals in order to monitor those shifts within the business goals, priorities, campaigns or objectives. Right? Next is implementing user defined goals. So you need to grant all the users or marketers, in this case, the freedom to create and personalize their rules for prioritization of data sources. Because if you empower these different teams to tailor the prioritization strategy to suit their individual use cases, it will help evolving into the request. It helps while they will involvement of the requirements. Now, at the same time, you also have to make sure they are on the same page because you don’t want one team to be making changes that negatively affects the other team. And in that case you need to encourage collaboration across different departments. So promoting that collaboration among different diverse departments for sharing those insights and priorities, that will certainly help and is one of the top priorities when it comes to best practices for datas and prioritization and having that guarantee and the inclusion of data from different sources across the organization and departments to be able to construct a complete view of the customer. Next is establishing a governance framework. So we talked about data governance and how it should be number one priority when it comes to designing and having your data stream prioritization. It’s because having and establishing that strong data governance framework overall. In short, as you adhere to the privacy regulations and even internal policies within your organization, there shouldn’t shouldn’t be any trade offs for this. This is a no no negotiation situation, and you need to clearly outline the roles and responsibilities of the management while updating these prioritization rules. Next is prioritize the consistency across all different channels. So as is when you’re using a CDP in the middle, the activation is happening on so many different levels and channels. So you need to sustain that uniformity in the data setting prioritization across the sources, across the diverse channels and touchpoints. And you need to also guarantee that consistent updating of customer profile, irrespective of the channel that they should be, that they are having touch points on or where the data is being gathered next is providing training and support. So you need to provide that extensive training to the current users or marketers that are tasked with the configuration and or overseeing overall data source prioritization and the rules and to be able to deliver continuous support. Right then you be able to handle any challenges or any inquiries that may be arising during the implementation itself. And you need to be able to maintain these prioritization rules. Next is regularly reviewing and optimizing. So it’s a constant, I won’t say challenge, but definitely a constant process. So the challenges are handled slowly. The the plan of priority assessment of that data source prioritization rules, they need to pinpoint areas for enhancement, as you see as you see fit. The the fine tuning of this prioritization strategy will come with some feedback addressing any building up or evolving business needs or adapting to any of the changes that the data landscape may have. Last but not the least, is scaling incrementally. So you need to prepare for scalability across the board and this can be done by incorporating gradual changes or upgrades. Now keep a close eye on the overall system performance right, and keep modifying that prioritization strategy as necessary to be able to accommodate the increase in data or user activity or changing data sources and so forth. So basically, by following all these best practices, the enterprises organization, they can truly enhance the overall effectiveness of the data prioritization strategy within their CDP. And it also leads to more accurate and actionable customer insights. The regular doing reviews on a regular basis and to be able to adapt to changes that will help contribute to this ongoing success or the prioritization strategy. All right. So this is going to be a summary real quick about what sort and now what so what being what did we talk about today? We talk about design prioritization and its definition. What it actually is and go encompasses thought is basically why is it even important? Why you talking about it? So we talked about the importance of it, discrete prioritization and the revenue related challenges. And we also talked about the architectural representation and use case examples. Oops. And now it means what are the next steps and best practices. So in the last couple of slides we discussed and went over what could be the best practices when you are considering data and prioritization for your own organization. This is towards the end of it. So these are certain sources, these are hyperlinks and I believe Kate even later on shared the deck in this content with you so you can click on them. These are sources for their experience league website. Some of the concepts that were accused in writing these slides, you can revisit them here and that was it from my site. Katie, over to you. Wonderful. Just thank you so much for taking us through all of that today.