Adobe Analytics Champion Tips and Tricks Follow-up Q and A
We gathered five Adobe Analytics Champions from across the globe to share the tips and tricks that helped them master Adobe Analytics. During the event, over 90 questions were asked by Adobe Analytics users around the world. Since our Adobe Analytics experts weren’t able to answer all of those question during the live Q&A, we got the group back together and hosted a follow-up Q&A session where the Champions answered YOUR questions and shared even more expert tips & tricks.
Transcript
Hello everyone, welcome to the Adobe Analytics Champions Tips and Tricks webinar follow-up Q&A. We had so many fantastic questions during our webinar that instead of just answering these questions on the CHAMP forum, we decided to get the group back together for a Q&A. So without further ado, let’s go ahead and jump into that first question. This first question is for Kea, and we’re going to start off with a tricky one, Kea. So it has a story first. It says, the story part is, we have an invented metric called engaged visits. Visits minus bounces. Our site UX leaves a lot to be desired, which makes engagement an even more difficult read than it is to begin with. It seems like this invented engaged visits is an vanity metric that can be misleading. People are hung up on it, and they think it proves something. All right, that’s the story. Hope you followed. Here’s the question. Are there better, more straightforward, not invented Adobe Analytics components slash combinations of components we can use besides the scenario described above? And what are some best practices ways to gauge engagement on a website? That’s a great question, and I like story time too. That was great. So measuring engagement really depends on what your organization wants your visitors to complete and the content on your website or platform, mobile app, whatever. It sounds like generating leads is part of your goal, so maybe instead of looking at visits minus bounces, I would look at the ideal user journey and see how far into the journey or the workflow, the user path the visitors get before they drop off or complete the transaction process. It’s really depending on what your organization focuses on. So, for example, my organization is a news organization. We focus on content consumption. So we started to look at visits minus bounces like you did, but we eventually realized that people who landed on the content they want will leave once they consumed it, but they’ve already achieved their goal. So instead we shifted our measure for engagement to different metrics depending on the content type. So, for example, articles, we look at the average time on page, scroll map to see if the visitors spent enough time reading the article, have they gone deep enough if it’s a long-form article, and for multimedia content we look at consumption rates, for example, how much of the video or audio did they watch or listen to. That’s it. That’s all I got, Justin. That’s more than enough. Fantastic answer. Just making sure. Great. Thanks, Jaya. All right. Next question. This is for Benj. It says, do we need to download a CSV file to analyze the cohorts? Can’t this be done in the analytics tool itself? So it’s up to you on what you’re most comfortable with, and it depends on the complexity of data and other segments that you want to include in your report. So you can download. For me, I prefer to download it so that I can play around it and see some other trends so that I don’t mess up my dashboard. So it’s really up to you. Excellent. All right. Mandy, we have a question for you here. This is going back to the 404 filter you spoke about. So it says, how did you create the page equals error 404 filter? Can you walk us through that? Yeah, of course. So for our organization, we actually have the page name set as error 404 when they hit a 404 page. So what I did to create the filter was I literally just brought in page name equals error 404 because that’s what it’s called in our implementation. So what you would probably want to do is check your implementation. So go to your site, hit a 404 page, and check the analytics call for it and see what the page name is coming in as, see what other props or EVARs you have coming in, and then look at that information, find what’s unique for your 404 page, and then set up your segment to use that unique information. So for us, it’s the page name. For you, you might have a different EVAR identifying a 404 page depending on how your implementation is set up. That’s great. I’m just curious. I’m going to throw this out there. Are any of you other champions doing anything with 404 errors as well? Not a lot of people really track errors is something that I found. A lot of people in their analyses tend to focus on where they’re doing well, but looking at where we’re falling is, in my opinion, one of the biggest things to do to try and improve. Yeah. Well, I love it, Mandi, just sharing some fantastic knowledge. Thank you. All right. This next one is for Gibson, and this one was, I believe, when you were talking about free forms. They were asking for some more clarification with the deduplication and not deduplication in free form. Can you elaborate on that? Yes, thank you, Justin. I think on this tip, I actually confused a lot of our audiences. After listening to that recording, I also noticed that there was a bit of a confusion. I’m going to try and explain this as best I could by giving a story. So what I’ve noticed is that in our organization, you have people that create different reports using the free form table. And what they do is that they try to filter out data. So one of the ways that you can filter out data within the free form table is by directly selecting the dimension item or value from the left side. Or you can either select the filter option from the free form table. So if you’re using any of those methods, you should bear in mind that your subtotal that you get from your matrix will not be duplicated, such as unique visitors. So if you have two unique visitors that visited your page, you would not see them as two. You might see them as 10 visitors, whereby if you use segments or you use a filter option, then they will be duplicated. So you will see that it was two unique visitors. So what I was trying to show is that when you are filtering out your content by directly selecting your dimension value from the sidebar, you should take note that your subtotal from the free form table will not be duplicated. So that’s all I was trying to show, but I’m happy that someone asked that question for me to clarify. Thank you. Thanks for clarifying. Appreciate it. All right. We’re moving through this. Katie, this is for you. It says, has anyone used the segment comparison for A-B testing? Sure. So I haven’t personally used the segment comparison for A-B testing. I’ve only used A-B testing with Target, and it’s got a great look in itself with its kind of templated reporting. But I do think it’s something that could be utilized, and I would be interested to see if anyone else has. I think it’s definitely something that could be of use, but nothing’s on my end. So I think it’s definitely something that could be used. Very interesting idea. Maybe if anyone has done that, let us know in the comments below in the thread on this Chant Forum or the community post. Awesome. Thanks, Katie. All right. Another one for Kaya. So it says, if I have a feature releasing on November 1st, 22, how can I build a segment slash time period that compares the previous end days to the subsequent end days that works regardless of when I open the workspace and run the report? Got it. That’s a great question. Unfortunately, dynamic date range columns are not available yet in workspace. There is a thread, I believe, in Experience League for ideas. So if you really want that feature, upvote that idea. I believe there was a prototype to test that feature in labs last year. So I hope it’ll become available soon. But yeah, that feature needs some love. So definitely upvote that idea. Just search for it in Experience League and hit like. And hopefully, the developers will pay attention. Perfect. Thank you. Yeah. If you have ideas, please submit it there. Great call out. All right. Benj, we have another one for you. It says, for your marketing channels like email and social media, do you use tracking code or last-touch channel detail? I use both. If you want to be specific on which creative or which ad version brought your traffic, then you need to have the external tracking code. But if you just want to see the overall channel performance and you don’t want very specific to the ad, then just use the last-touch channel. Fantastic. Great. All right. Mandy, you’re up. This one is asking, do you have any advanced-level alert monitoring? Or is there, sorry, is there any advanced-level alert monitoring in Adobe? So when you go to build alerts, in addition to adding the dimension, you can add different segments as well. So you can sort of make the alerts more advanced yourself by using different combinations of segments and metrics. It’s all basically the same alert builder, but by adding different conditions, you can make it a bit more advanced and a bit more specific to what your needs are. Perfect. All right, Katie. I got one for you here. It says, when dragging segments into freeform tables, I’ve never seen all visits as an option. Is it available in the left menu as a segment? I neither see it in the metric builder. Yeah, so I think it may depend on your implementation, but every implementation I’ve had, it has been a default metric, or excuse me, a default segment. If it’s not, you can definitely check in with your implementation team. And I definitely love to chat and kind of talk through that, but it is a default in my implementation. Katie, I have something to add on here, because I think the user who asked it may be confused, because it has all visits. For my side, I only see an R implementing as visits, so not unique. It’s all, but it doesn’t have the word all. I think that’s one of the things that the one who asked the question can check in their dashboard. It doesn’t have to be all visits, but visits will do. It only means it’s not unique. Thanks, Benj. Awesome. Teamwork. I love it. All right. Okay, here we go. When we use the date range comparison, we get a percentage change. Is there a feature that is being worked on where I can see the raw number change instead of the percentage change? So I don’t think that that’s something out of the box. I do create, so I’m going to share my screen. There is a shortcut that I use. You can use calculated metrics to create this column. So the quickest way I’ve done this is just follow the steps, do create a compared time period, and then in percent change, I just go down to edit component. It’s going to load. And then just so I don’t confuse myself, I change the name. It works completely. If I can spell correctly. And then just delete the compared date range so it just gives me the value. I also need to just be mindful of the format. And then also be mindful of the show upward trend. I think this should be good. The other thing to note here is that out of the box, this is only available in the project where this was created. If this is something that you want to apply to other projects and add it to component list and those projects, you’ll need to check this box. But for now, I’m going to hit apply. And it should give you that value change. So it’s something that’s not necessarily outside the box, but definitely something that you can create and make available in your workspace project. Stop sharing. That’s all I got. That’s a fantastic tip. Thanks for sharing that. Sure. All right, Benj, this is for you. Often when I’m looking at a top traffic driver using the marketing channel detail, but some of the child references are cryptic. Example, Google colon colon colon empty. Do you know what that means? And if there is a way to rename it? Yes, definitely. So there’s a way to rename it. Just add a segment. And what I do is like, okay, the hit and then you add either referring domain that contains, let’s say, google.com or external tracking code that contains TW. Sorry, I keep on using Twitter. Li for LinkedIn, FB for Facebook or IG. So you can customize it because what happens on my dashboard also, what I notice is that banners, it’s very generic. It will capture those from YouTube ads. It will also capture from DB 360, CM 360. And sometimes some of like Google paid search also are like, so it’s very messy at times. So if it’s a significant amount of data, and I really want to make analysis and correlations, then I will, I will further break down like the banners, the last such channel is banners. And then I have my segments. And even for social media, I have also my segments on each specific social media channel. Is it? Hope it answers your question. Perfect. Thank you. All right, this one is for Mandy says, if we use anomalies, or if we use anomaly exists with the percentage threshold, but there was some issue with the tagging historically, is there another setting we can use to not include the date range with issue to compute with the percentage threshold on the anomaly? So this one’s a little bit trickier, and it goes down into more about your implementation. So if you do have issues with the tracking historically, that will impact the percent change. What I would recommend is there are different options to change by a certain percent or anomaly exists. What I usually do when I set up an alert is I sort of go through those different options and put in different percentages. And it’ll always show you how many times that alert will fire in the last 30 days in the upper right hand corner. So in that preview area. So I usually go through and have a look at how many times it will fire. So if you do have errors in your implementation historically, if it’s greater than 30 days out, it looks like it’s going to fire. But you can look at those different settings, look at how often it’ll fire in the last 30 days and find a level that works for you. And also, you can always go in and update this as well. So once your tracking has been fixed, and it’s been fixed for a few days, you can go in and update it. And also, once your tracking has been fixed, and it’s been fixed for a few weeks, you can always go in and update those thresholds as well if they need to be changed. Great. Thanks, Mandy. All right, Gibson, this one is, it seems like they’re asking, it’s not necessarily a question, but almost they have a comment that I’m assuming they want you to respond to maybe give your expert knowledge on. So it says one of the issue, one of the issues in workspace is low traffic where metrics with very low volumes are not displayed. Do you have some some knowledge you can share on this or? Yes, Justin, I like this technical, you know, comments. So I wouldn’t really classify low traffic as an issue. So what happens is that low traffic is an Adobe Analytics functionality that ensures that the most important values appear in your report. So unique values collected up after approximately 500,000 existing values are listed under or bucketed in a dimension labeled as low traffic. So for example, if you have more than 500 clients, or let’s say a million as, as for example, and they all have IDs that you’re tracking, right. So in a month, if they all log in, some of your clients IDs will be classified under that variable. So it’s not really an issue. That’s the functionality of Adobe Analytics. Great. Thank you. Thanks for sharing. All right, Katie says, does segment comparison require a certain level of Adobe Analytics? Yeah, so I’m not sure I’ve been able to use it in every implementation. I don’t know if anyone else has has never not been able to use it. But I would be interested to see if anyone else you know, definitely comment if you if you can’t use it in time to talk to your Adobe rep and get some more information on it for sure. Sounds like this maybe. Go ahead. I think there is a level of package or SKUs that you have in Adobe Analytics that doesn’t, that doesn’t include calculated metrics. So if you include calculated metrics into your segment, you wouldn’t be able to use those Adobe Analytics. But I think everything else after that should be fine. So potentially, is where we’re at. Yes. Excellent. Thank you. All right. Okay, going up. Let me see back to Kaya. So this one says on the dates to confirm the date comparison cannot use the panel date. So the target dates are the dates that you want to know right now are based on the setting in the panel. Anything else, adding the time period column comparing the time period column date range segments should override panel setting in workspace. So you can start off with the base in the date range setting, and then work off of there based on what date you want to know. And then the date range functionalities you’re using. Great. All right, Benj. This one says the dash of days when the user opens the email is from a sample. I’m assuming it’s one of your samples that you shared. Sorry. It’s, it’s possible to see what the data from. So it’s possible to see what the data from predictions with AI. So I think they’re asking, is it possible to see data predictions with AI? I think it’s like every like AI is a big word now. So some of the champions can add in also I invite you guys. But so first, yes, it’s from a sample. And second, like the the email tracking on that Adobe Analytics can track is not the open. So just want to clarify that it’s the when the user click on a button in the email, and it will go to the website. So that’s what we interact that click not be open. And I’m thinking, like, if you want to use AI as predictions, I think it’s possible. Yeah, what do you think, guys? I think, like, a simple example for me, like, just simple prediction, not really AI AI. But what I do is that now that now that have the benchmark, like, let’s say, for the five days of the week that you have, like, excluding the weekends. So let’s say you have your 50% during Thursday, then you can just have like, okay, how many are your email recipients? Then you can, you can have a wrap a rough estimation on, okay, how many roughly like how many people will go to the website during our work will click from email and go to the website on Thursday and throughout the week. And then when should you send again your, your marketing emails and like based on your cohort analysis, also the return frequency. So you can create that formula. So it’s also like predictive, but I so far I haven’t used any AI in my, in my workspace. Yeah, like you said, AI is kind of a hot topic right now. So if anyone has anyone listening has done anything with AI, please comment in the thread below and share your knowledge because I’d be curious to see as well. Thanks, Benj. Alright, Mandy, this is similar to Gibson’s where it’s not necessarily question, but it’s a comment that we’d love to get your thoughts on. It says it would be useful if alerts and then in quotations changes by percentage could have below or above separation. Yeah, that’s one of the things that I thought as well is for most of the options in there, you can pick if it’s above or below, but the change by percent looks at both. This is something that, you know, we can we can post this on experience league and bring it up to the product teams. And, you know, hopefully this can become an advancement to the feature in the future. Because I agree, I think that’s something that would be helpful as well. Thanks, Mandy. Alright, Gibson. How do you track internal searches within Adobe Analytics? What keywords or phrases are users searching for on your site? Another technical question. Yeah, so with Adobe Analytics, so it makes it easier for us to track some of the important metrics, right? Like unique visitors visits pays views. So those are all out of the box metrics that you can get. But with things like search queries and search phrases, those are the things that don’t come as part of the search. So tracking internal searches requires some development, the key ways and phrases that users are searching would need to be manually tracked using custom variables through the help of a developer. So in my organization, what we’re doing is that we have a data layer that we ask a developer to populate with the keyword and the phrase that a user searches. And then we also have a search for the keyword. And then we have a data layer that we ask a developer to populate with the keyword and the phrase that a user searches. And then they trigger an event that we have created within launch. And then within our repository, we have dedicated evas that we use for tracking the keywords and phrases. And also we do count how many searches a user has done using a metric. So that is something that doesn’t come out of the box, it needs to be implemented. Great, thank you. And Gibson, while you’re up, I actually have another one for you right here. Can you please share a few tips on identifying bot traffic? Everyone hates bots and would love to get your knowledge on this. Yeah, exactly. So they are good bots like Google crawling on your site and then indexing it. And also that those malicious bots that you know, we hate because they come to our site for a malicious intent, right? So there are many ways to identify bots and I’m just going to cover the most popular methods that we use. So as a first line of defense, Adobe Analytics provides an option to filter our bots using a function called IAB, which needs to be enabled for each report suite. And if you know the IP addresses or user agent of targeted bots, you can create bots or processing rules to segment and filter out bot traffic. Another method is bots often used custom agents, which are classified as unknown within our dimensions. So you can look out for things like not specified unknown, when you look at your traffic, then you would know that that is coming from your bots. And sometimes organizations also do take security measures to make sure that they block bots traffic, you know, from accessing your site. So basically, there are many ways to identify because also the people that are creating these bots, they are becoming more smarter and smarter because they know that we are able to identify them. So I think one has to always look out in terms of whether the new bots out there and how to filter out for those. Thank you. Thanks, Gibson. All right, bench. So this one says, Have you looked at if there was a specific day during these times, it gives you the best. Let me see what I’m wondering if that’s referring back to one of your specific email examples. That probably doesn’t give you a lot to go on bench. Yeah, definitely can do that. So if you want to find out, you just want to see, let’s say Monday, like so Adobe workspace is just drag and drop, right, and very powerful, you can layer okay, day of the week, and then hour of the day. And even if like, you can also even go further, like, which day of the month, so that it even even has a feature. So you can be as granular as you want, you can be as crazy as you want in investigating or in doing deep dives. So you can do that, like I say, maybe you’ll find out like, okay, Thursday, at 12pm is the best time. So go ahead. Thanks. So we only have about six questions left. So we’re flying through this. Mandy, this one is for you. It says, in the analysis workspace, how do you highlight anomalies in line charts? I’ve tried it in the graph settings, but it seems to not be working. So other than like, in the graph settings, there is also the overall user preferences, you can try checking in there and make sure that you’ve got anomalies turned on. If you go into I think it’s user preferences under settings, and it’s in the same area where it’s like show anomaly, show sparklines. And I forget what the third option is in that section. But you can check in the overall user preferences, make sure it’s turned on in there. Other than that, it should be available to show those anomalies in in the graph. They should also be showing up in the table as well. So I would check in your freeform table that you’re using to build the graph and make sure that they’re showing up in there. If not, it could be that Adobe isn’t detecting it as an anomaly. So it sort of uses a calculation to figure out if the number is, you know, so high or so low that it’s considered a significant anomaly, it might be that the change isn’t big enough to trigger that anomaly. And in a case like that, where it isn’t, but you do want it to be marked as an anomaly, that’s actually where you can use the new feature annotations and add those to your graphs as well if something isn’t being marked as an anomaly, but you want to call it out in your data. So just a shout out for that new feature because I don’t think a lot of people have really adopted that yet. But it’s a really cool feature. Thanks for sharing, Mandy. All right. Gibson says, What are the must have extensions to add to your launch environment? Not include Adobe Analytics and Core. And do you recommend using Experience Cloud ID service? Okay, the way I see it, there are no must have extensions for an implementation. So it all depends on your requirements. The required extensions for a basic Adobe Analytics implementation is Adobe Analytics and Core and Experience Cloud ID services. But the extension catalog has many extensions that you can choose from to make your launch implementation much easier. So what I do as a rule of thumb, before I even implement any code or script, I go to the extension catalog to check if there’s no extension that exists for that because we’ve got some smart developers out there that are constantly creating these extensions to make our lives easier. So I wouldn’t say that there is the must have, but it depends on what you actually trying to do and just check if it exists within the extensions catalog. Great, great answer. Thanks. All right, Mandy, back to you. This is about alerts. Have you had to get creative around somehow letting Adobe know that certain events slash dates are expected to have spikes, so they do not send an alert, perhaps via segments or FTP? I haven’t had to do that personally. With our website traffic, we haven’t really encountered anything that we know is going to be much higher or lower. If you do know what it is, you could always use segments. I’ve never really used time-based segments, so I’m not quite sure how that would work with the alert. I basically try and set them up so that way they’re at a sensitivity level where it won’t fire unless it’s actually a significant error. So that is also where you could look at changing those percentages, maybe moving it to a higher level if you know that you’re going to see a certain amount of change on a regular basis. Great. Thank you. All right, Gibson. So we actually have just two more, one for Gibson, one for Mandy. So we’ll go back to you, Gibson. Says, what is the best way to calculate unique customers from my organization ID? I think the user that asked this question, they probably like the out of the box unique visitors metric that comes with workspace. So basically, with an implementation, if you are tracking an organization ID or a user ID, what you can actually do is that normally we store it in an EVA, right? And if you’re using the unique visitors metric, it would not give you the uniqueness of the organization ID. So what you would need to do is to create a calculated metric. And there is a function within the calculated metric space. It’s called approximate count distinct. So that would allow you to count all the unique organization ID and give you the total that you’re looking for. Great. All right, Mandy. Last questions for you. It’s also about alerts. It says, is there a best practice for establishing thresholds? Ooh, you saved the best to last. I did that on purpose. Yeah. In terms of a best practice, it’s really going to come down to sort of studying your website traffic and looking at the changes that happen. So depending on what type of website you have, if it’s a retail shop, you probably see differences between weekday and weekend traffic. You might see changes between morning and afternoon. So depending on what granularity you’re setting your alert at, so is it an hourly alert or is it a daily alert? You really want to make sure that you’re familiar with the type of traffic that your website receives. That way you can set it at a level where it won’t fire too frequently. For instances that you don’t need to be worried about, but also sensitive enough that it will pick up actual issues. So a lot of what I’ve done is really trial and error. So I’ve set an alert and looked at that 30-day history, how often it’ll fire, and then let it run for a week and sort of review how often it fired, compare it to my website data, and then go back in and adjust it. So really, it can be trial and error. It can sometimes take a few weeks to get it set up to the right specificity. But just making sure that you’re staying on top of it at the beginning, making sure you’re aware of what type of traffic you expect to see, and then being able to set it up based on that. So it’s really going to come down to your specific organization and what type of traffic you get. Perfect. Well, thank you. I want to just thank everyone for all your wonderful questions and for joining us for this Q&A today. And thank you to our brilliant champs for taking the time to answer these. This was fantastic. Hope everyone has a wonderful rest of their day. Thank you all. . .
recommendation-more-help
76e7e5df-e11d-4879-957b-210b1656662e