Using Flow and Fallout Visualizations like a Rockstar in Adobe Analytics

It’s no secret: I love Analysis Workspace. In fact, I think it is the main advantage Adobe Analytics has over Google Analytics. That is because Workspace allows for seamless collaboration between analysts, marketeers, product owners, and other business stakeholders. With enough enablement, there is no difference in which tools different groups of analytics users would use: It’s always the best one!

Workspace is the perfect combination of sophisticated functionality and an appealing user interface. But because of this user-friendly interface, not every advanced function or use case is immediately apparent to every user. This can lead to funny situations, where experienced analysts never really use certain parts of Workspace that could save them a lot of work.

In today’s post we will take a close look at two of the most undervalued features: The Flow and Fallout visualizations. While they seem quite similar in functionality and trivial to understand on the surface, they work quite differently under the hood. If understood and used correctly, they can be the entry key to a world where complex sequential segments can be built within seconds by every type of user, and sophisticated customer journey analysis becomes trivial day-to-day business. We are going to take a close look at how those visualizations work, since they are not made of some secret functionality, but simply combine segments, metrics, and dimensions in a clever way. So let’s put our detective hats on and start investigating!

Going with the Flow (Visualization)

We first start our journey by taking a close look at the option an empty Flow viz gives us:

Here we can take an interesting first observation: Workspace only lets us drop dimensions and dimension items into any of the three drop zones, while segments or metrics get straightout rejected. More interestingly, even pathing props (like Entry Page) are entirely not allowed. And even dimension items (like a certain page) are only allowed in the middle slot. The only allowed thing to drop in the middle slot are “normal” dimensions, be it builtin ones or custom dimensions. To get things started, I went ahead and dropped the page dimension into the “Entry”-slot and selected a “Visit” Container. To compare it to other reports, I’ve added a Page report with the Entries metric and an Entry Page report with the Visits metric:

We can see that with this setting, the total Path views (1) are equal to both the number of total Entries (2) and total Visits (3). Also, the Path views for the Home Page (4) are equal to Entries (5) and Entry Page Visits (6) for that same site. That leads to an important question: Which metric is actually behind this Path views number? To get to the bottom of this, we can use some Workspace magic: Right-clicking the top node let’s me trend the value for this node:

That’s gives us something to work with, since this will produce a new Line Chart (1) in our panel. The best thing: By going to the settings of that chart (2), we can show (3) the underlying Freeform Table (4):

And there we have it! When looking at Flow charts, we are looking at Occurences. This is a very important detail, because the Flow settings let us choose between Visit and Visitor as container scope, without actually ever using the Visits or Visitors metric. So while they can be the same (as in our example above, because we used the “Entry”-slot) they don’t have to be!

Occurences might be a strange metric to use, but it actually makes a lot of sense. With Occurences, we can not only get the number of Visits where a sequence happend, but the exact number of times a sequence took place. This is important to remember, since we are not only looking at how many people or sessions showed the sequence of events. We would have to use the Fallout viz if we were looking for that number!

So what will happen, if we change the Flow settings from “Visit” to “Visitor”? In my case, the numbers only slightly change. But what do they mean now? According to the Visit-logic, we correctly deduce that we now see what the first Page for a given Visitor has been. To verify this, all we have to do is compare it to the Entry Page of our First Time Visits, right?

We guessed wrong! But why are those numbers different? Shouldn’t the number of Entries in the First Visit be the same as the first Page for our users? Not necessarily! To explain, let’s look at another Flow Viz, but this time with the Visit Number dimension:

Here we can see what the culprit of our confusion is: Our reporting window does not cover all of the first Visits for all of our users. Some users came to our site before the selected date range, which makes “Visit Number 4” the second most common Entry Visit Number. That is another very important detail: The Flow Viz will only give you the first thing your users have done in the reporting window. This is not necessarily the first thing they have ever done on your page.

If you want to make sure that a given Entry in your Flow viz is the actual first thing a Visitor has ever done, add a Segment to your Panel with a Visitor scope and a definition of “Visit Number = 1”. That way you are only looking at users who had their first session in the timeframe you are looking at, giving you a valid number. I love to do this kind of analysis with dimensions like Marketing Channel or Referrer Type, to see how users move between different channels along their user journey. And if you are lucky enough to have access to Customer Journey Analytics (sigh…) make sure you throw your Device Type dimension in there!

Coming back to the right-click menu where we found the “Trend” option before, there is another awesome option waiting for us. By clicking “Create segment for this path”, we can just create some very sophisticated segments on the fly! Depending on the container we selected in the Flow settings, we will get either a Visit- or Visitor-scoped segment containing all of the touchpoints we have in our Flow viz including the right-clicked node. That makes it super simple to track users who, for example, came first to our site via SEO but later came back through SEA, showing cannibalization of natural search.

But there is even one more thing that the Flow visualization can do for us. We already explored the way to get to the Freeform table containing our numbers by trending certain nodes. This can be taken even one step further by modifying the segment that Analytics creates for us, switching out the fixed elements for only the dimension itself. That way we can for example create previous- and next-page reports in Workspace! I like this feature so much that I dedicated a whole post to how to create those reports.

That’s a lot of information to digest. I didn’t cover any of the basics, like Label Wrapping, Interdimensional Pathing (nice term, Adobe), or breaking down nodes (highly recommend to try that one!), since there is a good chunk of information about those out there already. I suggest to take some time to dive into how Workspace is creating the underlying segments to fully understand what is going on. Now onto our next subject!

Complex Flows made easy: Fallout Visualization

While Flows are awesome, there are some limitations to what we can do with them. For example we can only create nodes using a single dimension at once. Also, we can’t select single dimension items before dropping them onto the canvas, making it hard to find Flows where we exactly know what the touchpoints should be. Luckily we have another viz that can do exactly that: The Fallout visualization! It looks quite similar to the Flow viz, but with less drop zones:

Again, we can choose between a Visitor and Visit container, restricting if the selected fallout should happen for a given Visitor or all within one session. Depending on what we choose, Workspace gives us a default first touchpoint for all Visitors or Visits with the option to remove it. For our first deep dive, let’s drop the Visits metric two times into both of our Fallouts:

The first touchpoint is quite easy to understand: Since all user activity in Analytics happens within a Visit, we have 100% of both Visitors and Visits who get to that touchpoint. What about the second touchpoints? On the Visitor level, we might be intrigued to think we are looking at user with two or more Visits. But what is going on with the Visit Fallout? How can two Visits exist within one Visit?! To get to the bottom of this, we need to look at the underlying Segments and Metrics again. Luckily, we already know how to get there: Right-clicking (1) a touchpoint lets us trend it (2). From the created line chart, we can go the settings (3) and show the Freeform Table (4):

First thing to notice: We are indeed looking at the Visits metric when using a Fallout viz with a Visit container (or Unique Visitors for the Visitor container). This is already different from the Flow viz, where we were looking at Occurences and single hits! Now let’s take a look at the used segment:

That’s interesting. Are we indeed segmenting for Visits, where one Visit existed after another? Not quite! The explanation is something specific to Adobe Analytics: Technically, every Hit occurs within a Visit, so a Visit exists for every Hit. What we have found here is a handy workaround for a limitation of the Fallout viz (and segment builder), since we can not drop Occurences in them. But with this segment, we are really segmenting for Visits (or Visitors) with two or more Occurences. Neat!

Now that we know how those segments are built, let’s get to the best feature of Fallout visualizations: We can drop more than one touchpoint criteria per touchpoint! Also, we can mix dimensions, dimension items, segments, and metrics! For example, consider this complex Fallout:

Now we are analyzing, how many of our Visitors came first via Mobile and had a Page View, and how many came back later. All I had to do is drop all the criteria I want on the touchpoints. Let’s again take a look at the segment Analytics creates for us here. Notice that all of the criteria must be true within a single hit (even with a Visitor container), so all must be valid at the same time:

Now that’s quite a segment! For me as someone who trains people to use workspace, I would rather have them build those segments in the Fallout viz than in the segment builder, since that visualization immediately shows the impact for the rest of the funnel and keeps the technical complexity away from them. If they then want to monitor specific user journeys, all they have to do is build the Fallout and then trend either the absolute value or the percentage of users who reached the touchpoint. I also like to use the “Trend all touchpoints (%)” option, but it requires some more explanation to the business users (since early touchpoints include the users of later touchpoints) to prevent accidents.

A more advanced use case would be to use the segment from a Flow viz as a touchpoint in the Fallout viz. Since I especially like the Entry and Exit Flows, hijacking a segment from the trended Flow to use in the touchpoints of a Fallout would be a real Rockstar move!

While I’m very happy with the Fallout viz today, there is one more thing I wish we had. Since sequential segments like the ones above are hard to build by nature, some more flexibility in this awesome viz would be great. Let’s look at the available restrictions between touchpoints:

That is quite limiting. Compared to the settings we have in the segment builder (esp. within/after constraints, and time or dimension restrictions) there is not much for us to use here. The same is true for the criteria within a touchpoint, which always creates Hit level containers. I think that’s a real limitation. I would love if both my users and me would be able to play around with those constraints and immediately see the impact for the whole funnel! That would make it really easy to create a report like “People who came first via SEO and returned via SEA within a week”, showing cannibalization both in a funnel and over time.

Let me conclude this part about Fallouts with a warning: Don’t use the breakdown feature in Fallout visualizations. I would love to recommend it (since it is working great in the Flow viz) but look at the kind of segment it creates:

For some reason we can’t choose the dimension we want to use for breakdowns in Fallouts. That makes Analytics default to the Page dimension, creating a complex segment like this, specifically including the Page dimension in the segment definition. I’ve seen business user just happily switch out the Page dimension in the Freeform table, not knowing they would have to change the segment in multiple places as well. That is dangerous when looking at non-pageview events! The solution would be simple: Just let us choose the dimension like with the Flow viz! But for now, just better don’t use breakdowns in Fallouts.

Conclusion

I hope I could show you some cool things about my two favorite visualizations in Analysis Workspace. They can make building complex sequential segments by hand almost unnecessary, making them usable by a much broader audience. At the same time, it is important to understand how they work under the hood and when to use which. I’ve created a small FAQ at the end of this post with my personal recommendations.

As always, let me know if you found this post useful and have a great day!

Frequently asked questions

Frederik Werner

German Analyst and Data Scientist working in and writing about (Web) Analytics and Online Marketing Tech.

2021&2022 Adobe Analytics Champion

EMEA Adobe Analytics User Group Lead

www.frederikwerner.de/