Skill Exchange Event Aug 2023 - Grow Track - Take your data analysis to the next Level with these 4 underused features

Transcript

Hi, I’m Dan Mooney, a senior business consultant with Adobe. Hi, I’m Ryan Loggis, principal business consultant at Adobe. Today we’re going to talk you through four underused features in Adobe Analytics that will really help you take your analysis to the next level. So let’s get into it. A little bit more about me. Again, I’m a senior business consultant here with the data and insights practice. I’ve been with Adobe just coming up on one year. I’ve been using the tool for a little over 18 years now across a number of industries and functions, both as an individual contributor and a manager. I’ve spent a lot of my time really helping drive adoption and working across a number of different types of teams, so marketing, UX, BI, design and technology. One of the things I really like doing is trying to uncover the secret powers of Adobe Analytics and help powerful features become more easy to use. I’m really excited to share this with you today. Outside of work, one of the things I really love most is barbecuing. I’m a principal business consultant here at Adobe with the data and insights practice. I’ve been with Adobe for 16 plus years. I’ve loved my time here. I’ve been in the industry for over 18 years, and I love helping clients turn their business objectives into measurement strategies and performing deep dive analysis to really help them optimize their business. I’m based out of Gilbert, Arizona, and when I’m not working, I love to spend time outdoors, surfing, biking, or camping. The first thing I want to cover is one of the really nice features about our conversion funnel or fallout report. Like a lot of the features in Adobe Analytics, if you right click on this, it really unlocks a lot of power. So I want to show you ways to really get more out of your conversion funnels or fallout reports by right clicking. All right, so here we are inside of a traditional fallout report. And in this case, I’m using our geometric sandbox for a demonstration. And I have a fallout report for a traditional e-commerce shopping cart. So here you can see during about 40,000 visits, people initiated a checkout, made it through, and then about 4,400 completed an order. So it gives us a conversion rate of just under 11%, which is very good to know. It’s something that’s really important. One of the things that can really help take your conversion funnel analysis to the next level is simply by right clicking and then doing something I like to do called trend all touch points. So what I’m going to do is just hover over this bottom step of the fallout and right click here. And you’ll notice this menu comes up. What I want to do here is select the trend or touch points percent option. What that’s going to do is it’s going to go in and it’s going to take the daily conversion rate for each of those touch points and trend it. So right now it’s not formatted so good for me to really make out detail. So what I’m going to do is drag this to the left and resize this. So now I have a better view. So instead of just looking at that static or aggregate conversion rate for each of those checkpoints, I could see it trended over time. So what you can see is this trend really helps you understand what’s normal, what’s typical and what’s out of range and where should I dive into to analyze and make things work better. So one of the things that you’ll notice is there’s a pretty static, pretty steady conversion rate for each of the touch points until I get to the bottom right portion of my screen here where you see a big drop off in this last step. And this is where people are going from the order review page to actually completing an order. So this is a pretty big drop off because again, we were looking at just under 11 percent conversion rate. But most of the days during this one week period, it’s half to even a lot less than half of that conversion rate. So that’s going to be an important thing to go in and analyze. You’re going to want to know what caused that and then how can you prevent that. In order to really put a value on that, though, another thing I like to do is at this point, I’ll do something where I’ll pull in a free form table. And what I want to do is during that same time period, I want to trend by day for a dimension and then I’m going to pull in revenue as a metric. So that’ll give me my daily revenue. And then I’ll quickly just right click here again, trend selection, and that’ll give me a daily trend of my revenue. So I can see now I’m just going to reshape this so it has the same vertical alignment. You can see this is that period that we were looking at. So here between April 20th to April 26th, a seven day period, we’re far below what was a normal revenue for that for that week range. So what I did was I pulled this data into Excel and I’m able to see that during this one week period, I made a company made one point six million dollars. Typical during that range was around four point six million. So that’s a loss of sixty five percent or a little over three million dollars. So definitely something that I want to understand and I want to action and I want to I want to make sure that this doesn’t happen again. So another nice feature is is right clicking within the user interface. Again, what I’m going to do is create a segment and then put that into the experience cloud. So I’m simply going to go on this last step again and I’m going to right click and that menu pops up again and I’m going to select create segment from touch point. So this gives me the starting point. I’m going to give this segment a name. I’m just going to call it an abandonment segment. I’ll even spell it correctly. And what I want to do is just change this last last operator from exists does not exist. So this is giving me visits where people took all of these steps but then didn’t complete an order. So another thing that way that you could really use these reports and features to really up your analysis is create segments and put them into the experience cloud. So from here, depending on the skew you’re on or analytics package your organization has you have you’ll have a couple of options. You can select here and push this into the experience cloud and you’ll notice there’s some limitations there and there is a little bit of latency about four to eight hours typically. Or if you have it available to you could also create an audience library. So by doing either of these options what this allows you to do is now customize your experience for these returning visitors and you could use target to personalize their experience so instead of getting a generic experience it helps message to them to get them back into their cart and complete their transaction. Or you could push that segment into campaign and drive more traffic to your site and get that three million dollar audience back onto your site so they could complete their purchases. Another underused feature is really a newer feature but is called data dictionary. Data dictionary just came out this past spring and it’s really has a lot to do around adoption so really helping your power users in your organization to roll out Adobe Analytics to a wider audience. So one of the nice features there is you can customize a lot of the information that you give your internal users about Adobe Analytics. The way you access it is there’s two ways actually to access it. On the left rail you can see this data dictionary icon here but also if you hover over any dimension, metric or segment and you click the eye for information you’ll get this little window with some of the information but as an admin you can also click data dictionary and then view what’s been set up and then update it. So you’ll see here the admins got in and curated the segments and metrics and dimensions that are frequently used with this EVAR, this hotel guests and it’s let them let the users know that it’s similar to hotel date. They even went ahead and tagged it with user attributes so your users can go in and search like hash user attributes and then pull up all of the related segments, metrics and definitions that are related to user attributes. But another thing that you could do here is pull in really important things like EVAR settings and information from your tech documents around how and where the data is collected. This way people who don’t have admin access could see that easily and even your admin users won’t have to go into the admin UI to look things up when they’re doing an analysis or building a segment, really important things to do. So one of the ways you could do it here is you just click this pencil icon and then in the description you could do you can add a little bit of information like one thing I like to do is add in is this recorded through the browser and if so where or is this metadata that’s imported as a classification against an EVAR or an S-prop. I also like to add in the EVAR settings so letting people know is this a first touch that’s your last touch EVAR and what’s the expiration because those can vary and as you know those different settings will have very different impacts on the data that’s returned. So one of the nice features I like here that isn’t too heavily used yet is pulling in that information around where is the data captured, how is it captured and those EVAR settings. So once you have that entered in there you click save, you get your notification up here that it’s complete. Now all of your users when they go in they can not only see those curated components in the tags but also the information around how that data is captured and the EVAR settings. The next topic we want to cover was distinct count. Distinct count gives you the unique number of unique values in a particular variable. So a few examples of this might be if you’re a publishing company and you want to look at the number of articles that individuals are reading if you use the distinct count feature it will let you know how many people are looking at in different categories. So for example you might have business articles and entertainment articles and fashion articles. So if you use the distinct count feature it will allow you to see how many people are reading not just in one specific category but they’re looking into at articles in multiple categories. You could do a similar thing with the number of products that are viewed. So an individual might view the same product five times but if you’re using the distinct count segmentation feature it will allow you to see how many people are looking at different distinct products not just the same product over and over again. You could do this for product categories, site sections or any other dimension that you have. On this slide I’ve pulled up an example of how I like to use this distinct count feature in analysis. When I’m working with clients I love to do kind of analysis around different segments or personas oftentimes called cluster analysis where you try and divide all of your traffic up into distinct groups that behave differently. This example here that I have on my screen is again of a publishing company and oftentimes when you’re creating these cluster analyses you’ll see traffic break out similar to this. You’ll have a group in this case we’re calling them one-timers where it’s the majority of your traffic but they only come to the site one time they look at maybe one article and then they leave and then all the way on the other end of the spectrum you have what we’re calling our content addicts which are individuals who come back they’ve had 15 visits and each of those visits they’re reading several articles. What’s of interest in this group that I’ve created here we have a group for people with narrow interests versus a group who are more category agnostic. The way that I’ve determined this is using that distinct count feature. Here we’ve got unique content categories and for this group while they’re coming to the site a lot and they’re looking at a ton of content they only see a little over one unique category on average whereas this content agnostic group when they come to the site they’re seeing equally seeing a ton of a lot of content but you can see by this unique content category here that it’s spread out across multiple content categories so maybe they’re looking at business and finance and entertainment and the reading articles across a lot of different content categories. Let’s jump into the interface now and I’ll show you how to get this metric. All right now that we’re in analytics I’ve pulled in a metric table here that has all of my top metrics. Anytime I perform an analysis for a client around segments I like to start this way with pulling in all the key metrics for that customer. Again we’re in the GeoMetrics report suite which is a retail client and so maybe I want to answer the question I want a segment on people who have looked at more than three distinct products and so to do that I’m going to come over here to segments and we’ll click add and we’ll just call this greater than three distinct products and then we’ll do a visit level segment here and within one visit I want people who have seen three products so I’ll come up to the product dimension we’ll drag that across and then with the operator here we’ll click on this drop down and then we’ll go all the way to the bottom here and we’ll look at distinct count and we’ll say is greater than or equal to and we’ll set that to three and then just go ahead and hit save and then that should show up here in your segments greater than three distinct products. I’ll just drag this over here. And so now if we’re looking at the insights here we can see that for conversion rate people who are seeing more than three unique products they’re converting at 10% compared to 5% for all visitors and it might be a little bit better comparison if we said compare this to other people who have had product views so let’s pull in a segment for product views exists. Alright so comparing people who’ve seen at least three distinct products compared to just anybody who’s seen a product and we can see that the conversion rate is 10% for people who’ve looked at three distinct products compared to 7% for people who are just looking at who have a product view in general and so you can gain some valuable insights looking at it this way and understanding people who are looking at multiple items in a particular category. Let’s do one more thing with this I’ll turn it into a metric so we can see what percentage of people in a particular group are looking at a certain number of products so let me pull in mobile versus non-mobile visitors. So I already have segments created for visits from non-mobile and visits from mobile so we’ll pull those two segments in here and now based off of the segment that we created previously I’m going to create a calculated metric so we’ll go over to metrics and hit add and then I will find that segment that I just created recently three distinct products and we’ll drop in our visit metric inside of that and we’ll call this percent with greater than three distinct products and we’ll change this to a percentage with two decimal places and go ahead and hit save and then if we come over here to our metrics we’ll pull this in as one of our metric options. All right and now you can see there’s a difference between mobile and non-mobile visitors. Mobile visitors are converting at about eight and a half percent of visits from mobile are looking at at least three products whereas only seven and a half percent from non-mobile devices are looking at three products so there’s a difference in behavior based off of these two groups for the people who are looking at more than one product. All right the next feature that I wanted to talk about is attribution models in segments with EVARS and so in Adobe there have historically been two types of dimensions that you’re tracking your website data in there’s props and then there’s EVARS the difference being that props do not persist if we’re looking at a prop is set on a specific hit or page and then the value that’s set does not follow the visitors throughout their visit whereas an EVAR is persistent and so if somebody comes in for example from the marketing channel of paid search and then they continue on to place an order sometime later that paid search value even though it’s only set on the first page of the visit it’s going to continue following them throughout all of their visits until they place that order and that order is associated back with the fact that they came in from paid search and so historically we’ve had both of these dimensions but more recently Adobe is trying to simplify implementations and move more towards emphasizing using EVARS or one set of variables and so there have been some issues with that in the past where if you tried to segment on an EVAR because the value persists if you’re trying to segment on a hit where a particular thing happened you couldn’t do that very well with an EVAR because again of that those persistent values but now there’s a feature called attribution models when you’re creating segments and so you can set the EVAR to act more like a prop so let’s jump into analytics and look at how that works all right so now we’re in analytics again I’m going to just create a free form table here and we’ll pull in we’ll go with the pages report okay and now I want to maybe I want to create a segment where I only want to see hits where the shopping the page was shopping cart so if I come over to segments let’s go ahead and hit add and we’ll pull in the page dimension make sure this is set to hit hits where page equals and then we’ll come over to shopping cart and after we’ve done that we can click on this gear in the right hand corner here and there’s three options there’s repeating and what repeating is is essentially again it’s going it’s it’s just it’s set on this page but then it’s persisting through other hits for example if I click on a link there’s going to be another call that comes across with a link name and this page name is not explicitly set but it’s still persisting and so those will those link click pages will still show up in our link clicks would still show up in reports but that’s not what I want I want it to just be the hit where this particular page was set and so if I come down here to the next option that’s instances so this is all only going to give me instances where this page was explicitly set and not where it’s the value is persisting and so this is the one that I would choose there’s a third option of non repeating instance and this is just if somebody hit the same page multiple times in a row they refreshed it or something like that this is going to exclude non repeating would exclude those refreshed instances there and so we can hit save on this and then apply that segment that we just created here over the report and it’s going to filter it down to just hits where this particular value is set so we have our shopping cart page here now and if I were to pull in try and break this down by you know the subsequent link click that happened I wouldn’t get any information showing up in the report because because we’re using instance the instance is not persistent it’s excluded anything that might happen after the shopping the hit where shopping cart is explicitly set all right that wraps up our presentation I hope you found it useful thanks for listening we’re looking forward to your Q&A. And we are joined live by Dan Mooney thank you so much for your presentation. Thanks thanks for having me it was a lot of fun to put together and love seeing the picture of the family in the pool that was great. Thanks Dan I loved your tip about using the data dictionary to document admin settings it can be such a pain to always be trying to track down how specific metrics or dimensions are generated so having that right there in the interface seems like it would save a ton of time now let’s see kind of questions we have from the audience. So from Rachel and I’m just going to read these but I love Dan’s suggestion to make the settings of each variable visible to users is there any discussion underway about including this information automatically instead of relying on an admin to paste it in. Yeah great question Rachel it’s not right now automatic but like I think it was addressed in the keynote also this is a new feature just came out this past spring so there’s a lot of development going underway with data dictionary I’d love to see that as an automatic feature it’s something worth putting into maybe a suggestion in experience league and getting that voted up but yeah it’s a new feature it’s definitely something worth leveraging Making those settings out of the admin and copying and putting them in makes things a lot more easier to understand provides a lot of clarity for your end users and it’s things that if you’re not an admin you won’t have access to so I mean really those two things for anyone who’s not an admin it’ll make it a lot easier for them to understand you know how and where data is captured what those settings are but it’ll also free up a lot of time for the admins to do more admin type things instead of just responding to the question so yeah hopefully that could be something that could be added in soon but it’s not current but again new products so lots of new feature enhancements coming out. Awesome thanks Dan. Alright from Sonia how do we determine the source of clicks for example if someone clicks on a link in an email web browser MS Teams app to arrive at a particular page or content piece can we capture the source and analyze based on the type of source trigger? Yeah hey so sure you can you’ll need to think through a good taxonomy for tagging those or putting some sort of tracking parameters in the query strings like you would normally would with any sort of advertising campaign but yeah so you can do that but you’ll need to think through what should that taxonomy look like and make sure that you have something that’s consistent and reliable that you could then classify pull into marketing channels and do exactly what you’re suggesting but again yeah you just need to make sure there’s something on that landing page that will identify it as the different source types. Awesome alright our next question came in and so we’re actually discussing these questions ahead of time so if you submit your questions during the presentation we’ll have a chance to look at them a little bit. This one came in data dictionary is there a character limit and none of us knew so I actually pulled up a data dictionary and tried to reach a character limit wasn’t able to but if there are any any product people on the line I think someone said Eric Matusoff was out there so no pressure but if anyone knows the answer to that we would love to see an answer otherwise we have to take that to experience league. Alright the next one here can I use distinct count to define greater than three product out of a list of products I chose for example greater than three products out of some specific five products. Alright yeah Dan. Love the question someone’s thinking through this pretty deep and yeah that’s that’s something you would be able to do it’s when you’re creating those segments it’s it’s just like any other segment in this case you’re using distinct count as part of the criteria but you would just need to define those so it could be specific products or categories of products or parent categories of products but yeah that’s I think you’re right on the money with this that’s something that you can do you need to you would need to define that it would be a finite list but you could if you know what those products or product categories are you could set that up as part of the criteria in the segment and then start to look at that and there’s a there’s a lot of a lot of insights that you would gain by doing that yeah. Awesome thanks the next one we have can you explain the difference between bounce rate and weighted bounce rate. Sure yeah I mean they’re both going to show you bounce rate but if you think of say if you look at like a page like a content analysis if you’re doing pages and you you’re just looking at bounce rate if you just sort by bounce rate your pages with maybe one entry will have a very high could have a very high bounce rate you might not want to see that so that would be just a standard traditional bounce rate now a weighted bounce rate just applies like a volume or a threshold criteria so in that case you’re only going to look at the bounce rate for in this case pages that have at least x like maybe 500 or a thousand entries this way you’re not seeing any long tail traffic that’s kind of distorting your data and you’re able to get to not only to see what the pages are with the high bounce rate but the pages with a high entry or visit count. Cool all right next from Sherry I noticed the data dictionary there is an approval setting what does that mean and when approval is needed when approval needed is seen sorry. Yeah yeah so the the data dictionaries can be updated and approved by admins in your organization so the approval will be will notify you when something’s been changed or updated or a new entry’s been entered so you can have information in the data dictionary but not yet have it approved for kind of the general public within your org to see so it kind of gives you a pre-production way to can maybe syndicate that with other admins or to think it through or to really make sure that the details in there are accurate and applicable and useful for your organization so that approval is so that you could kind of look at things before it publishes and then once you’re happy with it you click approve and then it’s now available to all of the users in your org not just your admins. Great answer thanks next how do you add page percent load with the new web SDK since the plugin does not support it yet I think you could probably make this question about a few different features. Yeah yeah I think that might be another good question for the experience league something to feel there if there’s anyone else on the on the group that could answer. Yeah I mean I know that there’s like constant work being done to you know increase the parity between historical implementations and the new web SDK they’re always coming out with new stuff so I assume that will also be coming at some point but yeah might need to hear from someone on that on the product team. All right how does low traffic impact distinct counts? I’m not sure is there is there anything more that you could add to the question low traffic I mean if you’re doing low traffic it’s something that’s typically not you know over like a half million threshold so it’s going to be long tail type things that you’re not able to classify I’d imagine that low traffic I’m not sure the context of the question but I’d imagine that low tail traffic might not be included in distinct count if it’s something similar to like a classification issue but I think that might get excluded from distinct count. Yeah that’s a good question I’m not sure how that works. All right with the attribution model tip is it true that it will help make the EVAR function like a prop what does that mean for your data for the data that you see? Yeah so historically EVARs had certain settings that admins would set up and once the data was recorded there would be first touch or last touch and persistence that was recorded and set in the admin console and couldn’t be adjusted but with recent improvements the the attribution model you can now reset those so you can take an EVAR that’s set to persist in the admin console for say 30 days in the now in the UI within workspace you could change that attribution model and make that to whatever you want anywhere from the hit to a visit to a week till after a certain event is is occurs so you’re you’re now able to rewrite how an EVAR is is handled and then change that to if you wanted that to be as a as a as a prop you can make it that way or you could just leave it with the default setting and again that tip on adding the EVAR settings into data dictionary would be really useful here because obviously if you if you aren’t an admin and you don’t have access to that you won’t know you won’t be able to go in and see what that setting is so if you pull that into the data dictionary that’ll really enhance that feature where you can now see okay this is set to persist for 30 days I only want this for visit you can now go into the settings in the workspace UI and change that change that persistence. Yeah I think that’d be incredibly useful. All right when analyzing the technical performance of a page will instances be a better attribution for bucketing page load times? Yeah because you in the instances you’re right so you’re going to want non-repeating instance so the default will be the repeating instance so if you have something like a page name pulled into an EVAR it’ll persist past the the duration of that page but if you pull it into the instance it’s only when that that page view or that that value is being set it’s only going to be on those hits. Awesome all right when creating custom segments that are a list of user IDs product IDs or any other large set of unique identifiers what is the most efficient way of doing this instead of creating for example a rule that uses a hundred different or statements? That’s a good question. Yeah so one one way you could do it is there’s a relatively new feature equals any of I believe is the name of it or contains any of for for specific IDs you’re going to want equals any of because there could be similarities but you can copy and paste those and I believe it’s it’s either a space or a comma delimited field so if you had a long list of those you could pull those into excel and concatenate them then take that and copy it and put that into your field inside of the segment builder using that equals any of and then that way you won’t have like you say a really long list to maintain you’ll still need to to to keep a watchful eye on that because you’ll need to see all of those if you have a hundred of them say in that field it’s going to be hard to view but that would be how you could do that is just pull that into that equals any of fields and use that space delimiter. Yeah yeah I’ve definitely done that before it’s very hard to edit it once you put that in but yeah pulling it either in excel or copy an excel list into word and like find and replace with that I think it is comma delimited for that in that use case and then copy it into the equals any of. Good answer all right from Stephen four visits that enter the site through a product detail page can we see a table that breaks out the product SKU number category that was entered on for context our page names are not unique to each individual product so entry page won’t work. All right yeah I guess is if you’re capturing products those details that you need on that same image request because you’re going to want to see on that on that specific image request if you’re say you have a generic a fairly generic page name if you capture say those those product details in separate evars that that would be one way you could achieve not sure how your implementation is currently set up but you you’ll want to make sure that you’re passing those in that same image request on that on the page load on the load of the page of that entry. Right next does data dictionary apply across different report suites? No it’s going to be report suite specific and your variable assignment is is typically going to be report suite specific if you have different different websites with different business models so you’ll you’ll need to you’ll need to do that currently by by report suite could be an interesting option for maybe a future feature enhancement that’s something they could make it across report suites but currently believe it’s just by individual report suites. Yeah all right does a visitor segment look past the date range of the freeform table if I create a segment where purchase exists on a visitor segment and my freeform table is set to June will it pull visits from visitors in June that purchased in April? It it can yeah it’s something something worth validating too just to make sure that there’s nothing quirky going on with the implementation the way the data is processed but yeah that’s something to validate and data warehouse is useful for that too so you could do a data warehouse extract just to validate that data and then see if that is indeed pulling data from from a time period outside of what your data workspace is set up to. All right and we have time for one more quick question and answer. When I copy one project to create another version with a different filter or segment it changes the dashboard for the original project I created too. Is there a way to treat the copies as their own dashboards without editing the original? Yeah I mean one one way to do that might be save as and then save it as a separate name so this way you’re having a separate instance of that one dashboard instead of just copying it I would recommend trying that so instead of just copying the dashboard see if doing save as and then create a different name like you know v2 or something just something distinct separate in the naming convention create a separate project that’s save as and then see if that helps you accomplish that. Awesome well that is all the time we have with Dan so thank you everyone for your thoughtful questions and thank you Dan for joining us live. You bet thank you for having me.