Adobe Analytics: Analyzing the Data

Transcript

Hello all, this is Ashok Gaurupadi, Product Manager, Adobe Analytics. Now that you’re familiar with the Adobe Analytics user interface, we will dive just a bit deeper into using the interface to start mining your data. By the end of the session, you will see how incredibly easy it is to drag and drop your way to decision making in Adobe Analytics. You will learn all about building tables, applying and creating data ranges, and using the incredible power of segments to slice and dice your way to data-driven insights. Analysis workspace is incredibly powerful and almost suspiciously easy. But the one prerequisite for it to be useful is that you start with a business question you’re trying to answer. As soon as you decide you’d like to get to the other end of questions that typically start with the which, what, when, how many, rest assured you’re usually a drag and drop away from the answer. Let’s look at how you would build a freeform table and start populating it, all in service of the questions you have of your data. All you do is select the freeform table option from the rail in the left, which also lists other components you can use for your table. You would rename the table to whatever makes sense, and you select a date range you would like to apply to the table, say the month of March. You would then drag and drop metrics from the list of components, metrics like wizards, to start building your table. Think of metrics as components that help address the how many part of your question. I can choose the number of items I want listed in the table, like say 10. I can then add a dimension, such as page, to the freeform table, which would help me address the which and what aspects of my question. I can keep dicing my data up into even finer slices, like breaking down my page by the entry page, and even using the right click option to break down my entry page even further by say the browser type. So I’m now breaking the entry page down by the browser type dimension. I can continue to manipulate the table endlessly. If I’m looking for say a specific item in the list that showed up, like say the product details page, I can simply click the filter icon and filter the table using the phrase before sorting the column to get to a list of my top product details pages in terms of widgets. So I hope you’re getting an inkling of how easy and flexible Analysis Workspace is, and how you can drag and drop your way to answering questions, like say which page was the most visited in the month of April. All you do is select the when from the date range selector, drag the how many or in this case visits as a column onto the freeform table, drag the which part of the question page in this case as a dimension and that’s it. You now have a list of your top pages in terms of visits for the month of April. Here’s another example. As long as you can construct a question that’s a combination of the what, the which, the when and the how many, you can be assured that the answer is a freeform table away. In this case, you would set the date range to match, you would drag visits onto the column area of the freeform table, you would drag the browser dimension onto the row area of the table and that’s it. Given how ridiculously easy it is to drag and drop your way through your data, and how the freeform table offers an endlessly flexible canvas to structure your question and paint your answer, these drop zone guides appear as you drag and are about to drop an element onto the table. I personally find these guides incredibly useful since I know exactly what is going to happen when I drop the element before I do it. For instance, in this case, I know that I’m adding the online revenue metric to the table. Whereas in this case, I’m replacing the day dimension that’s already on the canvas or row with page. Here I’m filtering the online revenue metric by new visitors. It’s worth noting that only the top of the online revenue metric is set in blue, indicating that it is a filter that is being applied. Whereas in the case of the replace drop zone guide we saw earlier, the entire metric was outlined in blue. Another drop zone guide that you will see surfaced as you get familiar with Workspace is the breakdown option that would break a previously added component, like the thank you page in this case, by the campaign vendor component. Moving on to date ranges. Like everything else in Workspace, applying a date range onto the canvas or your data is exceedingly easy. The calendar option within Workspace lets you specify actual dates or select a preset from the options in the dropdown, which includes some of the more commonly employed presets like the last 30 days, this year, last year, and so on. When you select a date range, you have the option to restrict that selection to the panel within the freeform table you’re choosing the date range for, or apply the date range to all the panels within the table. Do note how if you do not choose a date range, but have populated your freeform table with components, Workspace does apply a default date range. You can also drag and drop date ranges listed under the time section of the component rail on the left, directly onto the canvas. And then you do have the option to choose whether the date ranges you apply are rolling or not, if you wanted to ensure that the data in the project is updated based on when the report is run and viewed. You can also create custom date ranges and save them as time components. This option can come in especially handy if you wanted to create a date range, for instance, spanning a campaign you ran this year and wanted to use that in your reporting. You would create a date range by navigating to the new date range option under components. You can also do that by clicking the plus sign next to the time section in the left rail. In the screen that pops up, you would simply select the start and end dates, decide whether you want your date range to be fixed or rolling, give it a self-explanatory title so other users within your org can find and use the date range, and click save. This custom date range would now appear as part of the time section of the left rail. If you’re talking time ranges, Workspace offers another nifty feature to make the comparisons across different time ranges super painless. You can start with any column and easily add a comparison column next to it, set across a different time range like year over year, quarter over quarter, or month over month, and so on. The way you would go about adding a comparison column is simple. You would create a freeform table with any dimension, as in rows, and metrics, as in columns you want to compare over a time period. You would right-click a table column and select the compare time periods option. The options that show up depend on the time range you’ve applied to the table. As soon as you make the selection, Workspace automatically adds a new column with the metric counts now spanning the new time range, and an additional person change column so you don’t have to. That it does useful things so you don’t even have to think about it, let alone do it, you will hopefully see is a recurring feature across Workspace. Here’s a table with a month-long date range applied, and I’m selecting the compare time periods option, which would let me choose from either the month prior or the same month last year. I can also choose to compare to a custom date range. Selecting the appropriate option adds two new columns, one with the metrics values across the comparison range, and another with the percent difference between the two ranges. You can also simply add a time period column by choosing that option instead of the compare option. This would add a column with the metric values across the other date range without the percent difference column. Here’s an example of the option being used to answer what the percentage change in visits to my homepage was between February 2019 and January 2019. Another example of the feature being used to answer what the percentage difference in revenue was from a Google browser between March 2019 and February 2019. And finally, segments. Segments are arguably the most powerful feature within Adobe Analytics. They let you add an incredibly specific lens with which to understand and decipher your data. They let you answer questions like who are my most valuable customers? How or whether I’m engaging my first time visitors? Am I faring better with my mobile app customers related to those on the desktop? Adding segments to your freeform table is just as easy as adding any other component like metrics, dimensions or date ranges. You just drag and drop the segment found under the left rail onto the canvas. You can use drop zones to guide where you drop the segments on the table. You can also, as we will see in just a bit, add segments as dimensions. That’s just another way of saying segments will be listed as rows in a table and also break down dimensions by segments. Say I wanted to answer a question as convoluted as this one. What are my top countries when it comes to visits and revenue that can be attributed to new visitors from mobile devices? Don’t get alarmed because segments let you do just that. Once I’ve confirmed the date range, I’ll simply add visits and total revenue, which were how many and how much parts of my question as metrics as in columns. Then countries, which would be the which part of my question as the dimension item. Then I’d look for these two segments, new visitors and mobile devices as segments. I’ll simply drag them onto the panel. As simple as that. We can also add segments to columns, which would restrict the metric count to the segment applied. The question I’m trying to answer is which of my pages attracted the most visits from mobile and non mobile devices. All I do is drag the page dimension and the visit metric onto the table. I then search for the segment I’m trying to apply, visits from mobile devices in this case and drag that onto the metric. Guided by the ever helpful drop zone guide, drop it onto the metric. I’ll then add another column with visits, but this time search for and drag and drop the visits from non mobile devices onto the metric column. We can also use segments as dimensions, which simply means we can add segments to the freeform table so they appear as rows in the table. Let’s look at an example to drive this home. Let’s create a freeform table for the month of April and use visits and page views as the metrics we want. You can break these metrics down by segments added as rows. You simply drag and drop the segments onto the table, just like you would if these were good old fashioned dimensions. Once you’ve confirmed the date range and added visits and page views as your metrics, you can search for or add the segment to the table. The drop zone guide will help you hone in on exactly where you can drop the segment. Once you’ve dropped the first segment, you can simply go about adding the others to the top of the table so they appear as rows breaking down your visits and page view metrics. Segments are also extremely useful in helping you answer questions like this one. Once you’re accustomed to breaking down a question like this into its constituent components, building a freeform table to get to the answer, I swear, becomes second nature. So Feb 19 is the when, page is the what, visits is the how many, and United States and mobile devices are the lenses you will need to apply. This is what the table would look like in Workspace. With US and mobile devices added to the mix as segments. Applying the same approach to this question, March and April would be the time ranges to be applied, visits and page views the metrics, and visits from search engines the segment to be applied as a row in the table. This kind of ties together everything we’ve walked through in this session, right? Applying metrics, applying date ranges, computing person changes across time ranges, and applying segments as dimensions. So that is the end of this track. I hope that was useful. And if there’s one thing I’d like to reiterate and leave you with, it’s the fact that any analytics tool out there is only as good as the questions you ask of it. And Workspace is no different. But once you have your questions figured out and framed, Analysis Workspace makes it ridiculously easy to drag and drop your way to insights you can trust and decision making blessed by your powers of deduction. So good luck and thank you. If you have absolutely any questions about what we walked through today, I’m going to stick around. So please keep the questions coming. Thanks. Thank you Ashok. Phew, we are getting deeper into our analysis now, aren’t we? And I hope everyone in the audience, you are really starting to appreciate how powerful and yet incredibly easy Adobe Analytics can be. And of course, as Captain Obvious says in the beginning, it all starts with asking the right questions. So to answer some of your questions, I want to invite Senior Solutions Consultant Gaurav Kumar also joining us from Singapore. Gaurav, the stage is all yours. Thanks Tanvi. Thanks for the great introduction and hello everybody. So I’m here to further take you down the route of answering some of the questions that around the things that you have been learning across multiple sessions. So let’s take some questions. You know, the first question that is coming, which is a very common question that we get as we get down into analyzing and into details with Adobe Analytics is that if there is a number of rows for a particular report or analysis is more than say 50K or X number, that’s a question coming from Sachin. What do I do? How do I get the report data out? So to explain this, there are multiple ways of taking data out of Adobe Analytics, right? So one is at the aggregated level that is with the reporting, you can run all kinds of reports, build the dashboards. Now once you go beyond the limits of what is allowed in the reporting from the perspective of keeping an easy UI for uses, you have next level of tools that you can use for taking the data out. The first one is an Excel plugin that is report builder. You can of course set and schedule a reporting structure. You can keep on getting data in an Excel sheet format out of Adobe Analytics. Going further, you have a tool that is called data warehouse. Now data warehouse gives you data more number of rows, but also at a higher level of granularity. What does it really mean is that if you are looking at an aggregated report in some of the reports in Analytics UI, if you wish to get more data in terms of the breakdowns, in terms of the dimensions coming together, you can set up a data warehouse request. And how you do that is you go into the tools in the menu of Adobe Analytics and you will have a data warehouse request that you can make. And the data comes within 24 hours based on where you want to receive the data. Do you want to get the data in an email? Do you want to get the data on an FTP server? Really depends upon your delivery mode. So you can pretty much use data warehouse. You can use report builders. But if you really want to access the data that is at click stream level, so if you want to go at the most granular hit level data, you can also get data at that level through what is called as data feeds. So if you go to again tools and admin, you will be able to set up something called as data feeds. Now, data feeds are the raw most data that you can get out of Adobe Analytics. And for doing that, you need to set up again the delivery mode. Do you want to get data into a folder? Do you want to get data into a cloud storage such as Amazon S3? You can set up click stream information to flow through from Adobe Analytics at the most granular level. There are many ways of getting data out. Depends upon frequency, depends upon the size, and also depends upon granularity of the data that you are looking at. I hope that helps, Sachin. The second question that is coming here, which is again kind of a very common question amongst the analysts as they go deeper into Adobe Analytics is, what is the meaning of unspecified when you look at data? Right? And this can be disturbing for an analyst when they’re looking at numbers. So let me try to explain this. Unspecified values in an Adobe Analytics report depends upon few factors. One of that is that it might be an issue with tracking. For example, if you are tracking a data for marketing channel and in the marketing channel values when you are putting the tracking codes, some of the code data is either not right in terms of format or the data is not getting populated because of various regions. The links are not working on the creatives and things like that. So you might get what is a broken or a missed data and that gets classified as unspecified. And the reason for that is that if you look at any Adobe Analytics report to make sense out of the dimension that you are looking at, it is important that you look in the totality of the data. Right? So if I’m looking at, say for example, overall widgets and I’m breaking it down further by marketing channels, even though some of the data might be missing into marketing channels, one of the channels, but when you look at the overall report, you still need to get the right aggregated number. That is why this unspecified dimension or value in the data is logical. Right? So it is not an error. The second part of this can also be the region for getting an unspecified data item or row item in your data is if you are breaking down something which doesn’t really correlate in the way that it should. Right? So while freeform analytics in Adobe Analytics gives you all the flexibility to break X by Y. Right? So you can break anything by anything, but in some cases the data might not correlate. Right? It doesn’t make sense business wise. It doesn’t make sense analysis wise. And in those cases as well, if there is a value that is broken down by uncorrelated value, you might get unspecified data. So yeah, that was an interesting question. Now another question that I’m getting here from Prateek, he asked that, what is the future of Report Builder? Well the future of Report Builder is very bright. We are not going to get away from the tool. It’s pretty useful for people who use and for reporting purposes. And some people are really, really comfortable for Excel with Excel sheets. So we definitely need to keep Report Builder. Another question connected to this is that sometimes when you pull data from Report Builder, the data doesn’t match between Adobe reports and Report Builder. Yeah, that’s a great question. And to be honest, in the terms, in the way that you have set up Adobe Analytics report versus a Report Builder report, because it gives you a lot more flexibility in terms of what you set up, what you don’t, you could see differences in the number. There are a couple of regions around this. If you are looking at the most recent time periods in terms of getting data in Adobe Analytics versus you are setting up say a weekly refresh of a Report Builder data, what happens is some of the data, especially in terms of classifications, in terms of mapping some of the EVARs, that is preprocessed data. The preprocessing happens over a period of time. So the data that you look into Adobe Analytics reports at say right now, for yesterday’s data, that data might be slightly incomplete. So in terms of aggregate, but in terms of say classifications and things like that. Report Builder is a completely baked data that comes out of at a fixed frequency that you use, refresh the dashboard. That some data can actually populate really depends upon when you are looking at Adobe Analytics reports versus looking at Report Builder. It also depends because some of the dimensions that you pull in terms of the breakdown, again, really depend upon some of the post-processing rules versus preprocessing rules. So it really depends upon the lifecycle of data pipeline, if I can call it. Both of these products, like the reports at UI versus the Report Builder Excel plugin, has been set up, you’ve got to set it up at different level or range of when you want to get data out of Adobe. So that’s why it could be slight difference, but you need to look at it at case-to-case basis, whether you are looking at discrepancies at an aggregate level for certain dimensions, is it around marketing channel data, that needs to be analyzed and used carefully. Cool. Another question that I’m getting here from Gaurav, and this is though sounds basic, but it’s a very, very important feature, right? So the question is, how can we compare date, the data between two dates? Though it’s the simple requirement of an analytics tool, I can assure you that in Adobe Analytics, this feature is one of the strongest. And the reason is that there are multiple ways in which you can compare time periods. So the first one is very simple. If you have gone through the earlier sessions, you would realize in the workspace, you can pretty much any dimension or metric, you can right click and break it down by compare time periods option, right? So when you do a right click, you have that option of comparing two time periods. So there are some fixed or fixed or pre-configured values in terms of what you can compare with. So for example, if you want to compare today’s data with last year’s same date data, that’s an out of box configuration in terms of time period comparison. You can also have custom time period that you can define. So you can also say compare data between this date to this date this year versus another date and another date last year. Another way of doing this is also if you go to the components, you can create time period components. And that’s where you can actually set up time periods, date time periods for comparison the way you do segments. So you can create a time period that says always two months before the current date. Now that’s kind of a custom logic to write a time period. So when you go to tools, create a time period, you can create a time period based on your logic and you can also save that to be used by other team members, for example, like the way you use segments. Now those time periods can always be dropped onto any element of data and you can get that comparison based on whatever structure of your report or analysis that you are working on right now. Okay. So there are a few more questions. There is a basic question that’s asked about what is the difference between EVAR and PROP? Okay. Let me try to explain this in very simple words. So EVARs are basically persistent variables or they’re also called conversion variables while PROPs are traffic variables. So the simple explanation is that PROPs only live onto the page where you are collecting data. For example, if you want to only get the counts of page views, right? Or particular page views for a particular page, then you store it into PROP because you don’t want to store that data across the journey of the visitor. At the same level, if you want to have an analysis set up for saying that people who are coming from a specific marketing channel, how many of those convert? Now in this case, you need to store this data, marketing channel information to live through the session, right? Or live through the life of the visitor on that particular session. So in that case, you store the information into an EVAR because EVARs are persistent and their values live across the pages. The reason why there are two types of variables is simple because we wanted to optimize the way you collect, store and govern data, right? So you can’t be storing and then there are limited number of different kinds of variables for optimizing cost as well in terms of server calls and things like that. So that’s the basic difference. Couple of more questions coming here. Can I connect on-prem tool to extract data from Adobe Analytics? That’s a great question. So I think I kind of explained this in terms of different ways of getting data out. So for getting data out from Adobe Analytics, of course, you can get the most granular data and you can analyze the data connecting with your on-prem data, right? The ways of getting data inside Adobe Analytics, of course, because Adobe Analytics is quite centralized around digital analytics, but there are still ways of getting data in, right? So for example, if you want to get CRM attributes within your Adobe Analytics and do some sort of analysis, there’s a feature called customer attributes and you can actually pull or upload the customer attributes data and do that analysis. So yeah, that’s around the CRM attributes. And there is one more question that I can take and say, why sometimes data doesn’t match between two different systems? Well, that’s again a very common question. And the simple answer is that the two different systems, whether you talk about two different analytics systems or you talk about comparing your data between Adobe Analytics versus some of the other tools, have different ways of collecting data, right? So Adobe Analytics collects data through its own tags, right? So the way the data is being collected and is stored in Adobe Analytics can be very, very different from another tool. So when you’re looking at reconciling data across two different kinds of system, there could be some allowed delta that you are looking at and you can have or live with that permissible. You can also determine which of the tools, analytics tools is going to be the source of truth for your data in terms of when it comes to say reporting versus going into analytics, right? So there could be difference in two systems and this is not something which is kind of new to here. So two different systems just because of the reason that they can collect data in a very different way. And you also need to determine which is the source of truth versus which is more of your analysis tool. So Adobe Analytics data, I think is a complete enterprise analytics data collection tool. So you always have a syringe that the data that you’re collecting or reporting for digital channels Adobe Analytics should be the source of truth. But when you compare other tools, it really depends upon how the tools are connecting data and how do you really match and compare. So with that, I think that was the time that I had today to answer your questions in this session. I hope that the learning journey continues and you get to know and ask more questions. And with that, I will give it back to Tanveer for the next sessions and I hope you enjoy the rest of the day.