Undertstanding metrics- May 2023 APAC Adobe Analytics Skill Exchange

Transcript

Thanks for having me. I’m Jennifer Workmeister and I’m a Senior Product Manager for Adobe Analytics. Ashok just walked you through the basics of the user interface and now I’m going to show you how to uncover actionable insights. First a little about me. I’ve been a product manager at Adobe for three years, but I’ve been focused on delivering data-driven insights for almost a decade now. I joined Adobe after getting my MBA from UC Berkeley, which is when I moved to Salt Lake City. My hobbies include almost anything outdoors, but I especially enjoy trail running, gardening, and painting landscapes. Before we jump in, there’s a quote I want to share with you. One accurate measurement is worth a thousand expert opinions. We know that timely, data-driven insights are essential to business success, so I’m going to show you how Adobe Analytics is here to make it easy to uncover those actionable insights. I’m going to show you how to build a freeform table, how to create date ranges, apply segments, and lastly we’re going to explore some panels that can speed your discovery of those actionable insights. Before we dive into analyzing our data, let’s first define some of the foundational metrics that we may want to use in our analysis. These three metrics form the foundation of analysis and workspace. First a visitor. So this is going to be a single user, a single person. A visitor may have multiple visits, and a visitor KPI might be called something like people, users, devices, etc. Second is a visit. So a visitor may have more than one visit to a page during a single sitting. That sitting is referred to as a visit or a session. A visit is always tied to a time period, so the most common way that a session ends is if the user has been inactive for more than 30 minutes. Lastly a page view is counted for each server call that is sent. In the analytics world you may also hear this referred to as a hit. As mentioned there will be multiple page views associated with a single visit or session. So let’s recap with an example. If I visit my favorite retail website, I will be counted as a single visitor. When I visit that site I’m going to browse through several different pages looking at items I’d like to buy. I’m still counted as a single visitor with a single visit, but the page view count increases with each page I visit. If I return to that website the next day, I am still only counted as a single visitor. But now the number of visits associated with me has increased to two. And the number of page views will again increase by the number of pages I visit in that second day. Now that you understand the foundational metrics that we’ll be using, I’m going to show you how to build a freeform table. A freeform table is one of the easiest and most flexible ways to get answers from your data. Let’s start by creating a new blank project and workspace. From the landing page that Ashok shared with you, click create new project and select blank project. A blank project typically opens with an empty freeform table. If you need to add a freeform table, you can navigate over to the left rail and select the panel icon and then drag and drop the table over into your workspace. The first thing you’ll want to do is select your date range from the calendar in the upper right hand corner. Now we’ll select a metric to create what’s called a time series or trended table. I’m dropping in unique visitors. As you can see, the freeform table automatically adds the data dimension to complete my table. If I go to the top, I can control how many rows I want to view at one time. I can also use the arrows to go back and forth between paginations so I can skip between different data points. Now to swap out a dimension, all I need to do is drag and drop a new dimension over into the column header. To create a breakdown, I can drag and drop another dimension into the table. Or I can right click on a dimension and create a breakdown from the list or by searching for it. Each time you create a freeform table, think about these three key requirements. When, what, and how many. When is going to be your date range. What you are analyzing is going to be your dimension. How many is going to be your metric. Second, think about what you want your finished table to look like. Typically, you want the metrics to appear as your columns and the dimensions to be your rows. When you drag and drop, drag and drop your metrics over to the right half of the table and your dimensions over to the left. Let’s try what we just learned with an exercise. Let’s think about how we would answer the following question. What page was the most visited in the month of April? The when is the date range from April 1st to April 30th. The what is the page name dimension. And how many is the metric visits? Putting it together in my data set, I get a table that looks like this. I have a row for each page and I can see how many visits each page received in April. It looks like my home page was the most visited page. Let’s try another example question. Which browser did your visits originate from in the month of March? The when or the date range is the month of March. The what or the dimension item is the browser type. And the how many or the metric is the visits. And here is what that table would look like with my data set. Now that we’ve mastered building our table, I’d like to point out some tips. You may have noticed the blue prompts show up as we built our table. These are drop zone guides and they will pop up and help you as you build your freeform table. The add guide, like what is shown here, will help you find the right spot to drop a metric or a dimension. The replace guide will appear when you are dropping a dimension over another dimension. And the breakdown guide will appear when you break down one or more dimension items by another dimension. Lastly, the filter by guide will appear when you filter the table by segments, which we’re going to go over in just a little bit. In this example, online revenue will be filtered by new visitors so that only revenue from new visitors will show up in the table. Now let’s talk about date ranges. While choosing our date range from the calendar in the last example was pretty simple, there are several flexible and powerful options for creating just the right date range for your analysis. The calendar analysis workspace lets you select a date range from a calendar or from a drop down menu of presets. When you select a date range, it applies only to the panel you’re working with. But you also have the option to apply the date range you’ve just selected to all the panels within the workspace project. If you don’t specify a date range, workspace will default to a current month. Dates don’t have to be limited to a panel. You can also drag and drop dates and time dimensions into a workspace project. For example, in our first example project, unique visitors were broken down by the day dimension. Lastly, rolling dates allow you to create reports based on when you ran the report. For example, you can have a report that always gives the data for the last 30 days. You can select preset rolling dates or create your own. Date ranges can be provided by Adobe or can be custom created. They can be applied in the panel calendar or created using the date range builder. You can find pre-created date ranges in the calendar in the left on the component rail. You’ll notice various date components in a separate date range section in the left rail. You can drag and drop these date ranges into a workspace project. You can also create your own custom date range that will show up in the calendar as well as the date range section of the left rail. Let’s learn a little bit more about those next. You can create a new date range either by going to the components menu and selecting create date range or by clicking the plus button next to the date range components in the left rail. Let’s create a custom date range with a rolling time period of one month, two months ago. To do that, select the one month, two months ago and then select use rolling dates. Open the details to make sure it’s set up exactly the way you want. You can use the date math to add or subtract the number of months, weeks or days that you need for exactly the perfect date. Hit apply. Now all you need to do is name it and save it. Now the date range will show up on the left rail where you can drag and drop it into your analysis. A common use of date range components is when you want to compare a metric over two different time periods. This workspace lets you compare metrics year over year, quarter over quarter, month over month, week over week, etc. Or between any two custom date ranges. It also has a date comparison feature that automatically includes a different column which shows you the percentage change between two time periods. To try this, head on over to the freeform table. Here I have visits for pages during the month of May. Right click on the metric column to navigate to compare time periods. The three options that show up depend on the table’s date range. You’ll notice that workspace creates the date comparison by adding another column with visits and filtering that by a custom date range of the previous month. Also a new column with the percentage change between the two metrics is auto created. This feature can quickly help you with comparing any metric over time for your analysis. Here’s an example. If you want to know the percentage change in revenue for each of your products between May and April, you’d start with a freeform table with your product dimension and revenue metric and set your date range to May. Then you would right click the revenue column and choose the time period comparison prior month to this date range. Now let’s talk about applying segments. Adobe Analytics lets you build, manage, share, and apply powerful focused audience segments to your reports using analytics capabilities, the Adobe Experience Cloud, Adobe Target, and other integrated Adobe products. In this section, we’re going to learn about how to use these segments in our analysis. Segments allow you to identify subsets of visitors based on characteristics or website interactions. They’re designed as codified audience insights that you can build in for your specific needs and then verify, edit, and share with other team members or use in other Adobe products. Segments can be loosely grouped into four major types. You can segment visitors based on attributes, so that’s browser type, device, country. You can segment visitors also based on interactions. That describes how they interact with your touch point, so campaign, keyword search, or search engine. You can also segment visitors based on entry and exit, which define where they are entering your website from, where they land, the last page they were at before they exit your website, and then lastly, you can segment visitors around custom variables like product ID, form field, et cetera. You can find segments in workspace in the left with your other components. They can be dragged and dropped into any panel you may be working on. The segments will filter the data in your tables and visualizations and can be used with dimensions to compare against metrics. You can also break down dimensions by segments. So that’s a lot of power and flexibility. Let’s see some examples. There are a couple of ways that you can use segments to filter data in your panel. The first is to apply this segment globally, so it applies to all the data in the panel. Here I have a free form table and I’m going to pick my segment and drop it into the segment drop zone at the top. As I’m adding these segments, you’ll notice that the data in the table is dynamically changing and is being filtered by the segment applied. I can add more segments here too and stack them so that they will filter down to only the data that exists in all of the segments at the top of the panel. Another way to apply segments is to apply them at a metric or a column level. This is helpful when you’re trying to compare a metric filtered by some condition against the same metric without the filter. So let’s say I want to compare my visits against all my pages from the mobile phone to visits against all my pages from not a mobile phone. I’m going to start with my free form table here with pages the dimension and visits as the metric. Now I’m going to search for a segment I’ve already created, visits from mobile phone. I’m going to drag and drop it over here to the metric column. Now I’m going to add another column of visits. Now that I have that, I can search for my other segment, visits from non mobile phone, and drag that over to the new column. Now I can easily compare and contrast for every page the visits that I have from a mobile phone and the visits that I have from not a mobile phone. Another really great way to use segments is to use them as dimensions. I have a table here with visits and page views. I’m going to go ahead and search for my segment. Now I’m going to replace the page dimension with that segment. Now it’s been replaced, but I want to see a few more segments in the list. I’m going to pull over visits from social and another visits. As I drop segments into the header, it adds another row allowing me to easily compare visits and page views across the segments I choose. So let’s try another example. How would we use segments to find the most popular page visited in the United States from a mobile device in July 2021? So I’m going to start with a free form table and the date range is of course set to July 2021. Visits is my metric and pages my dimensions. Now I’m going to stack the two segments to get the answer that I need. Drag and drop the segments visits from mobile devices and the segment USA to find the most popular page visited in the United States from a mobile device. All right, let’s try one that’s a little harder. Let’s find the month over month change in visits and page views from visits from search engines between June and July. To get this table, we’d start with a free form table with visits and page views as the metrics and July as the date range. Use the segment visits from search engines as the dimension. Then you’re going to use the compare time period feature twice, once for the visits metric and once for the page view metric. This is going to give you a great month over month view of visits and page views from visits from search engines and you’ll see that it’s created two columns. One’s going to show you the visits from June and one is going to show you the visits from July. We talked at length about the most popular panel in Workspace, the free form table, but there are several more that can help provide insights. We’re going to explore two more today that are particularly helpful. First I want to introduce you to the quick insights panel. It’s specifically built to answer business questions quickly and it’s a great tool for beginners. Clicking on the panels icon in the left will open up the panels menu. Then you can drag and drop the quick insights panel from the left to start using it. When you first start using Analysis Workspace, you might wonder what visualizations are the most useful, which dimensions and metrics are the most useful and where to drag and drop your items. Quick insights can help with all of this. It uses an algorithm to show you the most relevant dimensions, metrics, segments and date ranges based on your company’s usage. In fact, you’ll see dimensions, metrics and segments tagged as popular in the dropdown list. Quick insights is great for helping you properly build a table and an accompanying visualization. It’s also great for learning the terminology and vocabulary for basic components. It’ll help you to do simple breakdowns, add multiple metrics or compare segments easily within a table. It’s also great for trying out new visualizations. You can change the visualization type to help you see which ones work best with your data. When you first start out, go through the short tutorial that teaches you some of the quick insights panel basics, but I’ll take you through it right now. In the quick insights panel, you’ll start by selecting your building blocks or the components. These should be the same components that we just used to build our freeform table. You must select at least one dimension and one metric for the table to be built automatically. You can drag and drop your components from the left rail like we did in the freeform table, or you can use the dropdown. Notice that these first dimensions are labeled popular, indicating that others in my company have been using them also. You can also search for a dimension right from the dropdown menu. If you know what you’re looking for, just start typing and quick insights will fill in the blanks for you. We’ve added at least one dimension and one metric, the following will be created. You’ll get the freeform table with the dimension, here it’s marketing channel, and the metric, here it’s bounce rate. And you’ll notice the segment, iOS, has also been applied to the metric. You’ll also get an accompanying visualization, in this case the bar chart. The visualization that’s generated for you is based on the type of data you added to the table. Any time-based data, such as visits per day or month, defaults to a line chart. And any non-time-based data, such as visits per device, defaults to a bar chart. You can change the type of visualization by clicking on the dropdown arrow next to the visualization type. Quick insights can also make breakdowns easier. To get a breakdown, just click on the add breakdown option below your dimension to select what you want to break it down by. You can use up to three levels of breakdowns and dimensions to drill down to exactly the data that you need. You can also add more metrics. You can add up to two more metrics by clicking add metric. You can also add up to two more segments and choose to either combine those segments or to compare them in the table. The second panel we’re going to explore is the attribution panel. So a given customer journey isn’t linear, and it’s often unpredictable. Each customer proceeds at their own pace and they often double back, stall, restart, or engage in other non-linear behavior. These actions make it really difficult to know the impact of marketing efforts across the customer journey. And it also hampers efforts to tie multiple channels of data together. Attribution gives analysts the ability to customize how dimension items get credit for success events. So an example would be a visitor to your site clicks a paid search link to one of your product pages. They then add the product to the cart, but they don’t check out yet. The next day, they see a social media post from one of their friends, click the link, and then they complete their purchase. Attribution gives analysts the ability to customize how those marketing efforts get credit for the success event completing a purchase. The attribution panel is an easy way to build an analysis comparing different attribution models. It uses attribution IQ to give you a dedicated workspace to use and compare models. To start, click the panel icon on the left and drag the attribution panel into your analysis workspace project. First, you’re going to want to add a metric that you want to attribute. Then add any dimension to attribute against. So an example would be marketing channels or any custom dimensions such as internal promotions. Then select the attribution models that you want to use. The model describes the distribution of conversions to the hits. For example, first touch or last touch. Next, select a look back window. The look back window describes which groupings of hits are considered for each model. It should be the amount of time before a conversion in which you want to look at the touch points. As you can imagine, attribution models that give more credit to the first interactions see larger differences when viewing different look back windows. Once you’ve chosen your inputs, click build. The attribution panel lets you look across the different models of attribution to see what dimension is making the biggest difference in your conversion event. The top of the panel returns your total metric value, in this case revenue, and a visualization. In this example, I can see that whether I look at it from a first touch, linear, or algorithmic attribution, the display ads are the biggest contributor to revenue, followed by email. Below the visualization, there’s a table that’s going to help you better compare and contrast the results for the various attribution models selected in the panel. That wraps up the analyze section of today’s session. I hope it helps give you a sense of how easy it is to get answers for business questions when you have the power of Analysis Workspace. And welcome back, Servi, together again for some more insightful Q&A. Hello again. Yes, looking forward to answering some more questions from Jennifer’s section. Excellent. Well, we’ve had quite a few coming through. So let’s get started with our first one. It’s from Esther and she’s asked about hits, which I think was mentioned earlier on in Jennifer’s section. So under what circumstance do we apply hits? I’m assuming that’s based on hit-based segments. So hit-based segments are based off occurrences of page or link requests. And if you have to compare that with a wizard, wizard casts a wider net comprehensive for entire visitor web section, essentially. So while creating segments, we’d use hit container when you want to define, let’s say, page hits that you would like to include or exclude from a segment. It is mostly to narrow down of the container available to let you specify specific clicks or page views where a condition is true, letting you view a single tracking code or isolate a behavior from a particular section of the site. And also you may want to pinpoint a specific value where an action occurs, such as, let’s say, marketing channel where an order was placed. So these are some of the scenarios where you’ll use perhaps like a hit container. Awesome. Thank you. I hope that answers your question, Esther. Second question around segments. So you just mentioned segments there, but in the demo it says two segments were added and it looked like they were added with an AND operator. So, for example, USA and mobile. Is there a way to use other comparisons, for example, USA not mobile to create that segment? That’s kind of coming from Mike. Does that make sense? Yeah, so Mike, we can’t, in the operators, they are like AND, OR, and THEN. But when we drag, let’s say, if you want to say not mobile, so when we drag the mobile devices, it can still be like AND, OR, and THEN, but within mobile devices, you can select does not include mobile. So it’s just the other way of creating that segment perhaps. So in that option, you’ll have contains, exists, equals to, does not contain, et cetera. So that’s how you can just add that not scenario to that segment. Perfect. I think that’s what he wanted to know. Next question about, I’m assuming this is sort of an annotation kind of question. Can we add description and text to the workspace report? Yes, so of course, in the workspace report, we can drag in the visualization panel, you’ll have something called text where you can actually create insights and write down your insights and findings just to summarize if you’re sharing that with any other users. But if that’s one way of summarizing the insights, but if you’re talking about annotations, right? So we can create annotations in workspace, we can go into components, create annotation, but while you are analyzing your report, while you’re analyzing it in your free from table audit train, if you right click on it, you will see, you can add create annotation and put a description against it. Let’s say if there was a campaign that you ran, right, a Black Friday sale or something like that, if any event particularly happened on that particular day, you can add those annotations so that people can see that this spike it was or this drop happened because of this or this was the event that happened at that particular point of time. Perfect. So we can help our, well, help ourselves remember and help our other business users to kind of understand the context of what else was going on at that time. Yeah. Next question is about the session end or the session timeout. So can we change the session timeout criteria to something else based on business needs? I think it was a 30-minute example that was provided earlier. Yeah. So by default, it’s a 30-minute timeout that comes, that’s a by default setting. But when you’re creating your virtual report suite, that’s when, and I think in the next, the second step or the third step, you get an option to change that timeout setting to whatever you need it to be. It could be 15 minutes or it could be based on whatever your business needs are. So yes, you can do that when you’re creating that virtual report suite. Excellent. I’ve got a question from Divya, which kind of talks back a little bit more to the segment question that we had earlier. Divya’s asked if we could have a couple more minutes just explanation around the segment creation and maybe a couple of different scenarios and maybe just dig in a little bit more to those scenarios that you explained earlier. Yeah, sure. So when we are creating a segment, right, so you can create a segment, the segment panel opens, we can put it in a based off like a visit container or hit container or a visitor container. So if we have to look into examples where a visitor did something, then that’s when we’ll use that visitor scenario. Visit container would be when you want to understand that entire web session. So customer comes to the website, there is an entire session, he browses through different pages and leaves the website, that’s an entire session. It would be just based on that one particular entrance, clicks, page link request, etc, etc. So those are the three different scenarios. But then while creating a segment, it can be you can drag and drop your dimensions, it can be an and or a then condition like, hey, the customer did this and then this, or a customer could be like, like an example, the one that we just had was USA versus USA and mobile visitors, or it could be, hey, people who came from this particular campaign, and visit clicked on this button or this page or something like that, if you want to go very deeper into those particular segments. So yeah, those are some of those questions, those ways to create segments. If we have to use, let’s say tracking parameters, right, so you will have your URL parameters and understand where are the customers coming from and particular query parameter. So you can use that you can drag that and say contains whatever that parameter is. So that way you can you’re creating a segment based on customers who are coming from that particular tracking URL, which contained a particular keyword, let’s say. Thank you. We’ve got actually another question on segments. And this came in right at the same time. So I’m not sure if you have answered this already. But let me just confirm. So the question is from Shubham. And they’ve asked, can you please explain about sequential segments and how best to utilize those? So sequential segments would be just I think, also, again, in that segment builder, in that definition, you when you define your container, so something has happened. And instead of and where we had that and section, you can have an or, or if there’s another option which says then. So it could be things like, hey, added, came from this campaign and then added to cart or something like that, whatever they did afterwards. Yeah, got it. I think you probably had covered that, but just wanted to double check. Okay, another question from Kapil. Can we use events on the page, for example, video play, add to cart as dimensions? So can we use those events that are coming up on the pages as dimensions? It depends on your implementation. Those are more like metrics at the moment. But while just referencing back to the customer journey analytics that we were talking about in the first place, where you can bring in your different report suite, and that’s our newer omnichannel analytics, that’s where you can start using dimensions as metrics as well. So but yeah, in terms of can we capture those metrics? Yes. So you can capture video plays using streaming analytics, we can capture add to cart. But again, I think depending on your implementation, you can understand how do you want to deploy it. Or if you want to create a segment, I’m just trying to understand what the use case is. So perhaps you want to understand people who added to cart but did not purchase, right? So you can create a segment and then in your analysis, you drag that segment. So that would be more like a dimension for you. So you’re just analyzing those particular group of people who have added to cart. Got it. You touched a little bit on sort of dimensions metrics before. I have a question here around invalid data. So sometimes someone’s given the example of dragging over a metric and it says invalid data. Does that mean that there is no data? You would typically get invalid data when you do metric on top of metrics. So sometimes it’s just because it’s drag and drop capability, sometimes instead of dimension, if you’re dropping a metric on top of metric, that’s when you get like an invalid data. So it’s not about not having that data. It’s just the metric got dropped on top of that metric essentially. Got it. And in this sort of similar vein then of dropping, do we have to drop components into the table in a certain order or in a certain spot? Not necessarily. It can be however and whatever you choose to be. But sometimes if it’s talking about, hey, page visits, then add to cart and then conversion, you would want it to be in a sequence perhaps just to make much more sense out of it and for your users to be, make it easier for the users to read it. So, yeah, it doesn’t depend in which order you drop those metrics. It’s totally up to you and the business cases and the users. Right. I think we’re going to touch on this a little bit in the next section, which is more about how do you visualize based on like choosing the right kind of visualization that’s going to tell the story that you need to tell. So maybe a little bit more on that will come in the next section. We have a question in from Nural from Malaysia. Thanks for sharing where you’re from as well. Are there any tips to insert bulk dimension details rather than dragging one by one to create a segment? I’m not 100% sure. You can do that via API if you’re doing a separate API analysis. To be honest, I’m not 100% sure if there’s a way to do like a bulk dimension in your segments. Sure. We’ll have to look into it. Yeah, we’ll look into it. We’ll look into it and see if we can come back with a link for you. I think we’re kind of going to need to wrap up shortly. I have got a question in around CJA, which you mentioned before, customer journey analytics and a little bit about the differences between Adobe Analytics and customer journey analytics. So the question is from Adnan. Is CJA considered a step above or more capable than Adobe Analytics? It’s not about more capable. It’s more about these are tackling two different use cases. Adobe Analytics was purely from the digital analytics side of things where you’re analyzing all your digital behaviors. And then as in how our product team started talking to other people over the years, a lot of omnichannel analysis, omnichannel experience came into picture. So that’s when the Adobe experience platform was built just from ground up and customer journey analytics just lets you connect those dots from your digital to other channels as well. So it doesn’t have to just stop in there because we started getting a lot of questions around, hey, what about customers who have done this on the website but then went in store to make a purchase or where did they drop off to make a call center or just in terms of like attribution as well? Like how do we start doing that, getting that holistic omnichannel view? So that’s when customer journey analytics comes into picture. It’s more like you can bring in your analytics data with the connector or how so, but you can also start bringing in your in-store data and other sort of data to start seeing that full picture of the customer journey. So all of that remains the same. It still has the same analysis workspace look and feel. It’s just much more data in there and from different channels as well. So not necessarily one or the other. It depends on the use case and what you’re really trying to get out. Yeah. Makes sense. That makes sense. We have time for one more and it’s a follow-up from Esther who asked our first question about the hits. So she’s asked, will hits take into account the bounce frequency and could you guide us on where to get the definitional descriptions of those terms as Adobe terms them? Would hit take into consideration the bounce frequency? It shouldn’t. So typically hit containers will include values based on single page breakdowns like products, list props, list EVAs and any contextual events per se. It should not. I’m not 100% sure with like the bounces, but yes, we can look into it and get back to you perhaps if you have time in the next session Q&A one. Perfect. That sounds good. Just the second part of your question there, Esther, about guiding on where to get those definitions and descriptions. I think Experience League is going to be probably your best bet there. That’s where we house all of our documentation and lots of other tutorials. So it’s experienceleague.adobe.com and I’ll come back onto it with a little bit more detail after our next section. Yeah, just to add to that, we can ask those questions on the Experience League community as well where you can get answers like people jump in, we have Adobe experts and other users as well who answer those questions. Yes, that’s exactly right. And again, I’ll share some more info on that, but that’s going to be a great place for some added learning or added support. So we do have to move on to our next section though. So thank you, Servi, for your time again and we’re going to see you back here for one last time after our next section. Sure, see you soon. Keep the questions coming.