Time is on Your Side

Transcript

Hi, and welcome to my session, Time is on Your Side, How to Use Custom Date and Time Ranges to Your Advantage. My name is Jen Dungan and I work for Torstar here in Toronto, Ontario. I’ve worked with the company for about 15 years. I started as a developer, moved into QA and finally took over the web analytics implementation. Throughout that entire time, I’ve worked with Adobe Analytics all the way back to when it was called Omniture for those of you who remember those days. In addition to being passionate about analytics, I enjoy reading, crochet, playing video games and DYI projects around the house. In addition to LinkedIn, you can usually find me on the Experience League helping other users like yourself with questions and support.

Or you can actually find me at the Nom East User Group, which you can actually join or join another user group session in your area. As you can see, this session will focus all around how to use date and time to your full advantage. One of my inspirations of today’s topic came from the Experience League, where I’ve noticed a lot of questions recently around rolling dates and date segments. If there are other topics that you need help with, this is a great place to start and maybe it will inspire another session in the future. Hopefully these tips will allow you to go back to work later today or tomorrow with some fresh ideas and spark your imagination to create new and interesting ways to use some of what I’ve showed you with creative twists or your own flair. During the session, feel free to participate in the chat, share your favorite time quotes or post questions to the Q&A for the session that will follow the presentation. So let’s make the most of our time and dive in. A staple of the workspace interface is the date selector, which most people know to be the fixed calendar selector and the predefined rolling date ranges. Typically, the calendar is used to select fixed date ranges, though there is a quick rolling dates option that we’ll get to a little bit later, and the preselected date ranges, which are all based on a standard January to December, Sunday to Saturday calendar definition. What you may not realize is that Adobe actually allows you to customize your calendar definition. In the report suite settings, you can configure your calendar away from the standard Gregorian calendar. There are a few options that you can choose, but modified Gregorian will be the easiest, allowing you to choose the first month of the year and what day of the week you want the definitions to be based on.

This will affect your week and quarter definitions for your entire suite. But be warned, you can only have one configuration per suite, so make sure that you have properly investigated your needs before making any changes. Now, what if these options aren’t enough? The predefined segments may not meet all your needs.

Changing the calendar for your suite may not suit all your users. Some users may need the standard calendar and others may need a fiscal calendar. You may have special content that is released on a specific day of the week, and you need to create a report that follows a customized weekly breakdown. Or you may just need to do some reports based on custom date ranges as determined by your business needs. So the first custom date range option, one that I’m sure many of you use today, is to use the date picker on your workspace, selecting a date range with the calendar control, then simply check use rolling dates.

It’s quick and it’s easy and will automatically create a rolling date range based on the number of days selected, and it will always make it relative to today. Meaning if you have 22 days selected and today is included in your selection, tomorrow will still include today and 21 days back. If you choose a date range that goes up to four days ago, every day when the calendar rolls, the range will continue to show up to four days ago. Taking this a step further, many people don’t realize that you can click on rolling daily to rolling daily and customize the options. Within these advanced options, you can customize start and end options by start of, end of, and fixed day. Start of means that you’re looking at 120000 AM even in your rule definitions. When used on the start, this means you will include the specified day in the range, and when using start of on the end rule, it will not include the specified day. End of is the opposite. It looks at 1159.59 PM of the specified day. So when used on the start rule, this will not include the specified day, and when used on the end rule, it will include the specified day. It may take a little practice to get used to the rules, but Adobe will update the calendar control and date preview to show you exactly what’s happening.

Fixed allows you to lock one of the dates to a fixed point in time. Like maybe your report is about a specific campaign that started on a certain day. You can lock the start of the date range specifically to that day, but make your end date roll to either yesterday, so you can look at the full day data, or to today, so you can see the current data as it rolls into the system. The next customizable option is the reference point. By default, this is set to current day, but there are many different options that can give you better control. In most cases, you will probably use day, week, and month, but all of these have valuable uses. It’s up to you to see what works best. Next is the math that you want to apply to the reference point. Basically, which direction are you going? Minus will go backwards. In this example, you can see minus 21 days. If the current day is August 14th, then the start will be July 24th, as this is 21 days before the reference day. Plus would go forward. Now this probably doesn’t make sense if you’re using current day as your reference, but if you’re using current week or current month, then you can bring the date forward from those points. You can also set it to use no math modifiers at all by selecting none. And finally, you can change the unit of the math to calculate your ranges. You can mix the scope of your reference point and your math unit. You will also have noticed in some of my screenshots earlier that in the advanced settings, there are also start and end times. Note that these do not actually limit your reports to things like your core business hours if you use those start and end times as such. The start time attaches to the start date and the end time attaches to the end date. So they are no longer using the 12 o’clock or 11 59 59. If you want to limit your reports to a specific time ranges every day, like your core business hours, or your off business hours, or even specific times during a flash sale, you should do that through the use of segments, which I’ll show a little bit later. You can use the same logic, not just in your workspace date ranges, but you can create reusable date ranges for your users. This means that you can use your advanced knowledge to pre-create some important ranges for your business. If you remember a few slides ago, when I was showing how to configure your calendar for your suite, I also mentioned that some users may need traditional calendar breakdowns, while others might need fiscal ones.

This is where you can create those variations. Now, I won’t argue that this will be a little time consuming upfront, but it is a viable option and because they’re reusable, once they’re built, you don’t have to do this each time.

In my suite, I made the decision to keep the standard calendar definition Sunday to Saturday. So for the reports that I need to look at Monday to Sunday, I created a series of custom date ranges. Once I figured out the pattern of the math, I created a one week ago, a two week ago, a three week ago, etc. And I built as many ranges that I needed for my reports. Not all of those are shown here. Maybe you also need to make a fiscal Q1. Here I’m assuming that the fiscal year starts in March and therefore the quarter will run from March 1st to May 31st. Or maybe you need to create a last fiscal quarter that is relative to now. So in this case, you can actually build it from the standard calendar quarter definitions. Be creative, play with the options and come up with ranges that suit you. One thing that my company has is a series of content pieces that publish each week on Wednesdays and another series that publishes on Fridays. I can use custom definitions to watch how the content is performing week over week based on the publishing day for the content rather than on a calendar week. There is one last housekeeping setting that you will want to apply to all your date ranges and that is to tag them. You can tag them right from the workspace by selecting all the ranges, clicking on the tag icon and adding the tags from the dialog that appears. Or you can manage them from the date ranges component area. Tags make finding your custom date ranges much easier. My recommendation is that all date ranges use a standard custom date range tag or something similar, something that fits into your naming structure so that you can find all such items. Then use more specific tags for each set of dates, whether it be custom weekly ranges or fiscal ranges or whatever series of ranges you’re building. The visualizations are a bit more restricted since line graphs are still locked to standard date granularities. But you can still make all your freeform visualizations as normal and use bar charts as a proxy for the line graph. This is just a sample of what you can do. You’re all creative people and I’m sure you will find unique solutions to fit your needs. Another thing that people don’t realize is the order in which your visit level segments are processed in the context of your workspace. Segment definitions are processed first, then the date range of your report is applied. So if you create a segment that is users that hit page X, for instance, your segment will actually find all users from all time that ever hit page X. There is no time constraint on the data currently. Then in your report, when you apply a date range, the visitors that are shown will be limited to the full list of users that the segment returned and who visited your site within the specified date range.

These users did not necessarily hit page X within the timeframe of the report. They merely hit page X sometime in their user lifetime. And if you aren’t aware of this little quirk, this could be having a big impact on your All hope is not lost, however. You may have noticed in the segment builder that date ranges are available in the left hand component menu. You can add date ranges, predefined or your own custom date ranges that you’ve built right into the segment. I highly recommend that you clearly indicate the date locked nature of the segment in the name and description. When you use this segment in your workspace, it’s highly recommended that you also make a note in the relevant places to indicate that the segment itself is overriding the selected date range. Sadly, at this time, there is no dynamic date range that will apply the selected workspace range into your segment. And as you can see here from my example, my date range is clearly August, but my return data is from July because my date range segment is locked to last month. Doing a side by side comparison, you can see the impact of not binding your visitor level segments by a date range. The report date range is selected to last month, the exact same as the time locked date range, but there’s a huge difference in data returned by each segment. And in this case, it’s 1.2 million page views, which is quite significant. Credit for this next tip goes back to a presentation I saw at the 2017 Toronto Insiders Tour. You can create a peak time heat map by day of the week with a few simple steps. First, pull in the metric you want to use.

Next, replace the default breakdown with hour of day. Make sure this is hour of day and not hour so that the data will roll up properly. Now, find the standard definition day of week. And before you get ahead of yourself, if you just drag this over as is to filter your column, only the top five will be brought over. But since there are seven days in a week, we want to expand the dimension to see all the values that are available to us. Now, select all the values Sunday through Saturday and drag them all over at once and align them underneath your metric. Make sure filter by appears and not replace. Now, you will notice at this time that all the data for your date range is rolled up by hour of day. But because the individual values right now are less important, since we really want to make a visual heat map, we’re going to adjust the settings on the metric. We want to deselect totals, show sparkline, number, and percent. Basically, clearing out all of the data. Now, we want to actually change it from bar graph visualization to conditional formatting.

And while you can adjust the limits for your conditional formatting if you wish, the auto-generated values should be just fine for this purpose. Now, when we look at the final result, you can see what hours of the day and days of the week average out to the most or least traffic for the range of our report, giving You can even extend this to look at weekdays or versus weekends instead of individual days or build out time bounding segment to look only at your core hours. Have fun with it. The last piece of time-based control that I would like to share with you today is how to create segments to bind your data to specific hours of the day as hinted at a few times throughout this presentation and to share a great tip that came from the experience league from Andrew Wathen. I’ve added a QR code linking to that blog post for those who want to read more. It’s an excellent read. Now, if you recall a bit earlier, I alluded to how to create core hour segments and by extension of that, non-core hours. So when you create your segment, you simply use your hour of day dimension along with less than or equal to and greater than or equal to and create bounding ranges. These can be used in your workspace like any other segment. Keep in mind when you’re using less than or equal to 6 a.m., this will include all of the segments between 60000 and 6 5999 since these are all part of the hour. Now, here’s where things get interesting. You can use your non-core hour segment or create your own custom segment specifically for do not disturb and in combination with a calculated metric to ensure that your alerts won’t disturb you in the middle of the night when your low thresholds are more likely to trigger. In your custom metric builder, use the IF function. Now, apply hour bound segment and the most inclusive metric as your logical test. Occurrences is your best option here. And if the value is true, it’s an example, if you have any data in an off hour, which in most cases you will have something, you will force a value which will be above your alert threshold. In my case, I’m using 1001 and I’m using a static number to set that.

Otherwise, for your on hours, you’ll use the real value of the metric. You can see the results here. And while we’re never going to use this in a workspace as this shown, this is really just for demonstration purposes, you can see the hours of the day where the fake data is shown instead of real data and then during the daytime hours where real data is returned. And what this will allow us to do when you make your alert, instead of using your standard metric, you can use your new custom metric. And this ensures that during off or do not disturb hours of the day, you won’t get any alerts because as far as your alert rule knows, these hours are now always above that threshold. So from basic to advanced date ranges, there are many ways to customize your reports to your needs. Don’t forget about configuring your calendar and your suite settings. Now for most of us, the default calendar is going to be what we want, but some businesses are very focused on fiscal calendars. So analyze the needs and make those changes when it’s appropriate. And while a lot of us don’t like those day to day housekeeping tasks, I can’t stress enough how much spending that little extra time to tag your components will save you in the long run. Using time to your advantage isn’t just about looking at your data, it’s about optimizing your own time spent during the workday. Don’t forget visitor level segments may not be doing what you think they are in the context of your report. Always think about scope. And lastly, make sure you and your colleagues are getting a good night’s sleep by not pestering them with overnight alerts. Unless of course your business is international and you have business at all hours of the day, but be creative. Think about setting up alerts by geographical region and restricting each group differently. Thank you for spending a little bit of your day with me today. I hope you enjoyed the session and are living here a little inspired. So with that, let’s move on to QA.

It’s time for us to welcome Jen to the stage. Jen, oh my gosh, you gave us so many timely tips. I think the top question on everybody’s mind is what is your favorite time quote? I’m going to have to show my geekness here, but it’s going to come from Doctor Who. Time is like a big ball of wibbly wobbly, timey wimey stuff. Oh my gosh. Awesome. All right, well, and kind of funny thing in here is we’re going to start off a question that you were just recently asked at work and it’s can you do percent change with your custom date range? Absolutely, although unlike what I showed in my presentation where I was doing the custom date ranges as the breakdown, what you would actually do is you’d put those custom date ranges underneath your columns and then calculate your percent change across the columns to see. So one week ago to two weeks ago or one week ago to three weeks ago or whatever percent change you’re looking at. So it’s a little bit more manual, but it’s still totally doable. Awesome. Very cool. All right. Now, we had one that like struck with me, too, because when you were talking about your visitor segment, that was really great. Well, your whole presentation was great, but that one really struck. And I think it struck with this person, too, because they say, to be clear, if you create a segment without attaching a specific date range to it, it will pull data incorrectly? Well, like anything, not necessarily.

Unfortunately, it’s like anything analytics.

So if you think about it, your visitor level segments are going to be obviously the most because you’re looking at all time visitor data. Now, your visit level metrics or your visit level segments and your hit level segments obviously won’t have as much crossover or inflation that we’re talking about. But even at the visitor level, you have to take into consideration that what you’re looking for may actually be all time data. If I want to create a segment that’s looking at visitors that have ever subscribed or visitors that have ever purchased or visitors that have ever done some action, of course, you’re going to be pulling out the right data. But if you’re looking for visitors that performed an action within your date range, your report range, that’s when you’re going to run into the problem. So just like everything else, you just have to be mindful about how you’re building the content and building your segments to make sure you’re actually returning what you think you’re returning. Yeah, yeah. Sometimes you do have to kind of like triple check your segment, like pull data a couple different ways to say, hey, this is really doing what I want with my segment. That’s really good advice.

And this one’s kind of along the same lines.

You know, I don’t know if that was really a gotcha with the visitor one, more of like a reminder, but someone’s asking, are there any other kind of gotchas that they should kind of think about when using these different times? Well, again, there’s probably always gotchas that no one can anticipate, unfortunately.

But those are the biggest gotchas that I’ve noticed. Again, when it comes to visit level or hit level segments, you’re not going to have as much flux. If you’re using a visitor level segment, but you have subcontainers which are at the hit level or subcontainers that are at the visit level. So I guess really the biggest gotcha is just confirm that the data you’re returning is actually looking like what you expect it to see. Try to do as much correlation when you’re testing out your segments to make sure they’re actually doing what you think they are doing. Sorry about that.

Yeah, and that’s the case with everything. You don’t just throw stuff up and go, OK, it’s good, because if you do that and you’ve missed something, then everyone’s making decisions based on bad data. So do your due diligence, check, double check and triple check. And if you’re like me, check like 20 times to make sure.

You check time after time. Yes, that’s exactly it. Too fun. All right. Well, one of the things you did talk about using time for is those alerts. And so we have a question asking if you can explain some more about the alerts and how people can use them to help them out. Right. So I’m not sure if you’re looking at just alerts in general or if you’re looking at alerts that are time based so that you’re not getting interrupted in the middle of the night. So I’ll touch on both of those just in case. For anyone who hasn’t used alerts, alerts are a really good way to send out an email notification to you or to your stakeholders that something doesn’t look right. It can be over a certain amount, like a hard number. It can be based on a percentage. So you could actually look at rates to say, oh, if the bounce rate goes over X or the conversion rate drops below Y, you can actually send out alerts to your stakeholders because something might be wrong. Now, again, always look at the scope of what your alerts are doing because like me, I got a global suite. I set up my alerts for the website, alerts for the app, alerts for AMP, because I don’t want those to be crossed over because if something’s going wrong, I want to know exactly which aspect of my site is causing the problem.

Now, when it comes to time-based, again, that’s fully a choice. You may want to monitor all night long and don’t mind having alerts coming in at all hours of the day. But in a lot of cases, if you’re doing it, especially at a raw number value, when it goes overnight, especially if your business is based in a certain time zone, 2 o’clock in the morning, you may not have a lot of sales or 3 o’clock in the morning. You may not have a lot of whatever. And if you drop below those thresholds, you’re going to be getting an alert every hour going, you didn’t make this quota and you didn’t make this quota. And you come in in the morning and you’ve got like a bunch of emails saying that something’s wrong and you’re like, no, no, no, everything’s fine. But so you just have to play it a little bit by ear about what exactly your alert’s doing, who’s on the alert notification and what you want it to do.

Now I’m with you. I think some of the alerts I’ve made, I first make them for myself to make sure that they don’t go to haywire. I know you kind of have that little box at the top, right? That could say like how many times it would have gone off, but it’s still nice to try it out on myself for a little bit. So yeah, good. Another tip that I do with alerts is rather than sending them to individual emails, if any of you guys are using Slack, Slack channels can actually, you can have an email attached to a Slack channel. So I actually send my alerts out to a Slack channel that anybody can join and anybody can watch the alerts coming through. So it’s not just going to me and then I have to alert the team. The team can see it immediately.

That’s a good one. I like that. I think the only other ones I’ve done is other than to an email is to a phone number. And that was like when there was no sales, you know, when you really need that text message right away to say, hey, it’s during the middle of the day and there’s zero sales, there’s a problem.

Oh yeah, that’s not a good email to get.

Oh, let’s see. Well, talking about your segments, really, I think have struck a chord with people. And so we’ve had a question saying, can you explain a little bit deeper the difference between the hit visit and visitor segments? Absolutely.

So similar to how it sounds and similar to a lot of the vernacular that Adobe uses, hit view or click or the actual action that’s performing. So you’re looking for a very specific combination of, is it this page and this value and this all on the same actual hit. So that’s going to be your most compact type of segment. The next one up is visit. So essentially you’re looking for users within their visit who have met certain criteria. And when you use that visit level segment, just blank into your page, every single page view and every single action within that visit will be counted, not just the ones that you’ve built in. It’s any visit that has X then comes through and all the traffic for that visit shows up your workspace.

And then visitor is any visitor that has met X criteria from all time that happened to be within your report range. So again, that’s why I talk about timeboxing those ones, because in certain cases, I don’t want to know visitors who purchased a year ago. I want to know visitors who purchased this month or this week, depending on the scope of my report. And I hope that helps. No, that’s perfect. And it actually leads into something that another person dropped in talking about segments and time constraints. It says, might I suggest adding an additional option under the hit visitor visitor under that selection in the segment to maybe have visitor within time frame to constrain the results so that you don’t have to add those additional dates. Oh, God. I love that hint hint to any Adobe people on this call. That’s actually I am literally going to post that idea onto Experience Lead. I just haven’t gotten to it yet. But as I was recording this, I was thinking, oh, that would be so lovely. So I’m totally on the same wavelength with whoever asked that question, because that is definitely a need that we are missing today. Oh, all right. Well, it doesn’t make me smile because you like your presentation. And now just like this answer right now, you’ve talked about how you’re helping people out or you’re active in Experience League. Would you tell us a little bit more what you’re doing there? Sure. For anyone who’s not aware, Adobe has this great forum that’s called the Experience League, and it’s a place where you can post questions, ideas and ideas that you can vote on them. And hopefully Adobe likes them and will implement them into future product releases. You can post blogs and you can even post in discussions.

So usually ideas and questions are the most active, but any user can come and post a question if they’re having problems with, you know, how do I pull this or how do I build this type of segment or something’s going wrong. There’s a whole community of people and employees actually who monitor this and will help answer questions the best we can by posting screenshots or say, hey, try this or try that. And so anybody can actually become a community advisor. You just basically have to be active on the community, have been providing a lot of good information to fellow users, and they may just contact you and go, hey, we’d like you to be a community advisor.

Oh, well, that’s cool. Now you’re always so helpful, so that’s cool. I’m a perfect fit for you. Well, okay, I’m going to ask a personal question here. So I really loved your peak time heat map. There you go.

Can you tell us a little bit more about it? Like, how did it come about? How are you guys using it? It was pretty cool. Yeah, well, as I actually said, I saw that originally during a rock star event for the insiders tour. So I didn’t actually come up with the idea, but as they were kind of like leading up to where they’re going with this, and that’s really, really cool.

But essentially how we tend to use it, we actually use it in a couple of different ways. One being a newspaper company, we want to know when the peak times or the best times to deploy content because you want to actually get your audience and try and capture them. Now, obviously, news, if there’s a breaking news story, you’re not going to hold it back and say, oh, our peak time is not until five o’clock tonight. Let’s not say anything.

But if you have, like for us, we have investigative series coming out. So knowing when to publish those or what times to kind of capture the most attention when people are coming into the site, we can use that like a long span to look at the last month or the last two months to try and get an average of how our users are using the site and when they’re coming in. But on top of that, for our development team, we can actually use that to help figure out when the best time to deploy code is. Oh, look, this time of day is usually kind of low. It doesn’t get a lot of traffic. So that’s probably the best time to deploy so that people aren’t getting the cash refresh issues. Oh, yeah. I love how you use it both ways, like both from running the business, the authors and editors to the technical side. That is really cool.

And kind of to piggyback on that, because those heat map kind of things, they use that conditional formatting. Do you have any suggestions? Because I know it’s kind of, I always have to kind of hit and miss with it, but I’d love to hear your suggestions on it.

Sorry, can you repeat that last one? Sorry, on the conditional formatting, do you have any tips on the best way to deal with that? That is always a tricky one, because if your data is very volatile, if you try and lock it into place, you’re going to start seeing issues in most cases. So for the most part, I just leave it on automatic.

But when I’m trying to do, say, a percent change, and maybe I want at the 0% and then above 0%, like when it goes up or when it goes down, I’ll actually adjust that and I’ll actually put in my time binding to middle of the road is zero, and upper is 0.001, and lower is negative 0.001. And I’ll just kind of lock it in as close to zero as possible so that I know that above zero is in the greens, and a below zero is in the reds, or vice versa, depending on if your negative is good and you’re, you know. Good point, good point. Oh, I love it. And I know there’s been a couple more questions and, you know, segments and hits and visits and visitors are always, it always takes a couple times to get it. So I think, what’s the best way to get a hold of you on Experience Leaks? I think some people could ask you some questions there. Right. So technically speaking, you can just, you know, file a question and any of the Experience League people will join in. Or you can search through and you can find me and send me a private message if you really want.

That’s awesome. Yeah. And I’m sure there are lots of blog posts about, you know, hits and visits and visitors to help clear things up, or maybe they just add more confusion. But I’ll do my best to help support and get people to understand because it is a little tricky when you’re first getting into it. I will admit that. Well, that’s awesome. This has been a great time, Jen. Thank you so much. Thank you for having me.