One of my favorite, pow­er­ful analy­sis tools in Data Work­bench is the Latency visu­al­iza­tion. Many clients, how­ever, are not aware of this visu­al­iza­tion tech­nique or they’re unsure how to inter­pret find­ings from this analy­sis method. In this blog post I hope to help you get more com­fort­able with using the Latency visu­al­iza­tion and give you some spe­cific use cases you can start apply­ing today.

The Latency table is a visu­al­iza­tion in Data Work­bench, one of the fea­tures in Adobe Ana­lyt­ics Pre­mium, allow­ing you to exam­ine aggre­gate vis­i­tor or cus­tomer behav­ior before and after a spe­cific event occurred. The ana­lyst selects the event to cre­ate the visualization.

Here’s an example:

Your dataset in Data Work­bench includes rows of online and in store event data for all cus­tomers who have had inter­ac­tions in both chan­nels. In the days after a cus­tomer buys a TV you want to under­stand what pages they are view­ing or prod­ucts they are buying.

Let’s first look at how latency is cre­ated and then how to inter­pret it at the vis­i­tor level.

Once you set an event within the latency table you are now look­ing at cus­tomers who per­formed that event. In our exam­ple of the cus­tomer buy­ing a TV, you would select the TV from a prod­uct table that sig­ni­fies it was pur­chased, and then select “Set Event” action in the Latency visu­al­iza­tion. You are then see­ing those cus­tomers behav­ior prior to and after the selected event, a TV pur­chase. You can add met­rics for the counts of times when they looked at TV stands, or when they bought HDMI cables for the TV, and see where those events fall in time in rela­tion to when they bought their TV. One thing to call out is that if a cus­tomer has per­formed the event you selected in the latency table mul­ti­ple times across their life­time, each of those dif­fer­ent instances of the event will show up on day zero in the latency table.

Latency tables have par­tic­u­lar appli­ca­tion for track­ing activ­ity related to a cam­paign or to a par­tic­u­lar cus­tomer order in which you are look­ing to asso­ciate with a time correlation.

Below is an exam­ple of a latency table to help illus­trate exactly what is hap­pen­ing when you set an event and how the other data is dis­played within the table.

Here are four dif­fer­ent cus­tomers who pur­chased a TV from your company.


The latency visu­al­iza­tion allows us to nor­mal­ize the data and look at the events which occurred in rela­tion to the day the user pur­chased a TV:


The power of doing analy­sis in latency tables is the abil­ity to see how many days after a cus­tomer pur­chases a TV you should expect to see an order of a TV stand or HDMI cable.  If you have a tar­get­ing tool, like Adobe Tar­get, you would be able to tai­lor your mes­sage in an email or ban­ner to cus­tomers who are close to or past this expected order date based on your data in the latency analysis.


Sam­ple Use Cases for Latency

Let’s build upon the use case that I men­tioned above that is look­ing at cus­tomers who pur­chase TV’s.  In the step by step instruc­tions below we will look at how to set the event in the Latency visu­al­iza­tion, add met­rics to the Latency table and save the Latency table to be able to view it as a graph.  Let’s walk through the above use case here.

Step 1: Select the TV from the prod­ucts table that will iden­tify that the TV has been purchased.


Step 2: With this selec­tion made, set the event in the Latency table (remem­ber you can set mul­ti­ple events through­out an analysis)

Set Event

Step 3: Now you are able to see what is hap­pen­ing with cus­tomers before and after they pur­chased a TV from your company.


Inter­pret­ing the data latency table:

First, remem­ber that +0 days ele­ment in the table is dynamic based on each spe­cific cus­tomer.  If you look at the chart above with the four dif­fer­ent cus­tomers, once you set the event all the dif­fer­ent cam­paign inter­ac­tions go to day zero.

Sec­ond, when you are inter­pret­ing met­rics from the latency table you should read it by say­ing, “2 days after the event hap­pens there are 13 HDMI Cable orders.” We can see from the Data Work­bench visu­al­iza­tion above that +4 days after the event, we have the  sec­ond high­est num­ber of TV Stand views.

To get more detailed analy­sis we rec­om­mend cre­at­ing cus­tom met­rics in your latency table.  For exam­ple, cre­at­ing an order met­ric for a related prod­uct to show how many cus­tomers are buy­ing prod­ucts together.

Another great next step is to save the Latency table as a dimen­sion.  This way you will be able to bet­ter under­stand the time series and how it relates to your event. Here are the steps to save a Latency table as a graph:

Step 1: Right click on the col­umn under the word “latency” in the Latency visualization.


Step 2: Save the dimen­sion in your Data Work­bench user folder.


Step 3: You will then open up the Latency table as a dimen­sion from the table menu struc­ture in Data Workbench.


Step 4: Once you open the Latency dimen­sion you can right click in the col­umn under the dimen­sion name and select Add Visu­al­iza­tion then select Graph.


Step 5: You will now be able to see the Latency table in a graph for­mat.  You can use the same func­tion­al­ity of this graph just like other graphs in Data Workbench.


If we take the exam­ple we have been work­ing with through­out this blog we can see that if you add a series to the graph of dif­fer­ent cus­tomer seg­ments: Online Only, Offline Only, and Mul­ti­chan­nel Cus­tomers, you will see where the real power of mak­ing the Latency dimen­sion into a graph can be.


The key take­away from look­ing at Latency this way would be that  Mul­ti­chan­nel cus­tomers are the group that are pur­chas­ing the most HDMI Cables, while Online Cus­tomers are pur­chas­ing the TV.  At the “+4 days” after the TV pur­chase there is a spike in Mul­ti­chan­nel HDMI Cable orders.

The value you can gain from doing a latency analy­sis is the abil­ity to bet­ter under­stand what your cus­tomers are doing lead­ing up to a cer­tain event as well as after an event.  A lot of our clients find value in ana­lyz­ing dif­fer­ent cam­paigns, shop­ping cart aban­don­ment, plac­ing an order, or account reg­is­tra­tion using latency analysis.

Like I men­tioned in the begin­ning of this post, latency analy­sis is one of the pow­er­ful visu­al­iza­tions in Data Work­bench.  If you have any ques­tions or com­ments on how to do a cer­tain analy­sis please feel free to leave them below or reach out to me on Twit­ter at @adaste.