Driving analytics adoption by delivering value


I think user adoption is a pretty universal concept in IT and working with Einstein Analytics is not any different. You can spend hours, weeks or months on building great dashboards so your end-users can get some insight, however, if they do not use those dashboards I would argue that it’s been a waste of your time and money. Hence why user adoption becomes such a key (and often overlooked) component for any Einstein Analytics implementation.

My personal opinion and/or observation is that people often consider user adoption when they find out their dashboards are not being used – well after they have been finalized. It would be great if the fix is that easy, but I believe you need to start way before the dashboard is done to have the best chance of getting some meaningful and well-used dashboard. So let’s have a look at how I believe you can influence the Einstein Analytics user adoption.

What can we learn from Experience Economy?

Before I dig into concrete concepts and things to consider in order to increase user adoption, I want to throw some theory on the table. If you have a smartphone you can probably relate to having a lot of apps on your phone. I believe there is a certain lifecycle we all go through which starts when you acquired one of these little tech wonders/monsters. First, we check out the app store to find a lot of apps; some funny and some useful. Second, we start deleting those funny but rather meaningless apps. Third, we only add meaningful apps that are going to help us on our phone. I guess we realize as we consume that it’s a fun distraction, but it quickly gets old and we move on. Only some apps remain on our phone which we wouldn’t consider deleting for different reasons.

Why is this relevant? Well, when I was writing my master thesis back in 2011, I was determined to scope and design a mobile app for a company, but what was best practice in doing this? Back then this technology was still relatively new and I had myself gone through this app lifecycle. I had downloaded funny apps like the beer in a glass app (I honestly don’t remember the actual name) but you could see how the beer glass got empty as you hold your phone upside down. Meaningful and useful? Not the slightest, so that app quickly got deleted from my phone. But some apps did remain on my phone and I used them on a regular basis. So I considered what principles should you use when you are designing an app to avoid that it got deleted, how can you ensure you stay relevant and top of mind? Doing my research I though experience economy was interesting to look at as it’s all about adding value to your product. Traditionally, experience economy talks about making physical goods into an experience; consider for example a cup of coffee from a vending machine versus in a cafe with a view of the ocean. The latter is an experience you value more and stay with you longer though both options keep you caffeinated. I saw similarities between this and mobile apps and I figured I could draw parallels between the two in my thesis. And again I see parallels between a mobile app and Einstein Analytics, as the goal here should be designing analytics app and not just single dashboards.

In my research, I ultimately reached a framework of how things to consider and I will highlight three here. The first step is that your app should be themed and answer the basic questions (who, what, when, where and why), the second step is to make sure the user is invited to interact with the app and finally, the app should work in collaboration with other channels that the user is already leveraging. Obviously, I am not trying to rewrite my master thesis so this blog is extremely simplified, but I still believe these steps are valid when creating your analytics app to make sure they add value and stay active in the long run. Hence in the next part of this blog, I will discuss how you can follow the framework by:

  1. Design with your end-users in mind,
  2. Design with clear intentions,
  3. Put analytics front and center.

Design with your end-users in mind

First another university story. Back in then, I had a class in internal communication and I remember my teacher giving us two questions that I have taken with me and use in any form of communication. The first question is really all about making sure that your message (read dashboard) is relevant for your audience:

What Is In It For Them (WIIIFT)?

And the second question aims to make sure the message (read dashboard) is easy to decode:

Will your grandma understand that?

I think both of these questions are extremely relevant for user adoption because if you consider and apply them you will make sure you 1) understand who you are creating a dashboard for 2) understand how this dashboard can help them in their job and 3) make sure they know how to use the dashboard and apply the insight they are getting. If you want to dig more into the art and mechanisms of communication and why it’s important I previously wrote a piece on that called “Fundamentals of Communication” you might find interesting.

So how will you achieve having successful dashboards? Well, it comes down to understanding your audience. I already touched on this topic in a previous blog of mine about design and interaction principles, which I still stand by. But how do we link audience, message and ease of use? A while back I was introduced to the Einstein Analytics Look Book (see the Dreamforce 2018 version here), which outlines a great approach to figuring out which dashboard are needed for your end-users. In your design workshop, you should (among others) consider four things:

  1. Audience,
  2. The questions the audience have,
  3. The KPI’s and charts that answers the questions,
  4. What actions they want to take.

For the audience, you typically want to focus on a single role. An example could be marketers. Ask yourself how you can describe this persona. Maybe they are:

  • Working from the office using their laptops,
  • Using traditional marketing channels,
  • Finding it difficult to understand their sales impact.

Once you have the persona nailed down you can start putting yourself in their shoes – maybe you have even observed them – and you ask the questions they would ask.

  • How many leads are generated from the website?
  • How many marketing leads get handed over to sales?
  • What is my best performing marketing channel?

With these questions in mind, the next step is to find KPI’s/charts that answer the questions. Each question might not be answered with just one KPI, it could very well be more complex than that. Also, it is not unlikely that you find out that you do not collect data for certain KPI’s, but then it’s up to you to work with the business analyst and Salesforce admin to make sure that you find a way to collect this data.

  • Number of leads per channel,
  • Number of converted leads,
  • Number of converted leads per channel.

Finally, consider if the answers they are getting leads to actions. Maybe your marketer notices a poor performing campaign and wants to check out the details or a good lead has not been picked up by sales and wants to escalate the assignment.

By understanding ‘who’, ‘why’ and ‘what’ you are already one step closer to getting your users to use your dashboards as there is something in it for them if they use your hard work.

Design with clear intentions

Alright, let’s return to your grandma. Of course, I don’t mean you need to create dashboards for your actual grandma, but this is really all about making sure that a person with limited knowledge can view your dashboard and understand what is going on. You do not want your users to spend more than 10 seconds (if even this much) to understand how they can benefit from it. So whatever you do when you are designing your dashboard ask yourself if it will ease the understanding. This might sound very basic, but from my consulting experience, I can promise you, I have spent a long time decoding dashboards in both Einstein Analytics and standard reports and dashboards in Salesforce. So trust me when I say it might sound simple but it’s hard to practice. So let’s have a look at some principles.

  • Unclutter your dashboard,
  • Make sure to structure and group your data,
  • Consider usage and actions.

Uncluttering your dashboard really comes down to looking at every single component and decide if it adds any value for your user. If you can think of a purpose then keep it. But if the element is disturbing then remove it. Unfortunately, this principle is never clear cut, it always depends on the scenario, sometimes a legend is needed in a chart and sometimes it is not. When in doubt I would always ask a peer or maybe the end-user and get their feedback if they understand the meaning (in 10 seconds) you have probably done a good job.

The second principle has two sides to it. First just because you can add everything to a single dashboard doesn’t mean you should. For a marketer, they might want to both see sales influence and campaign engagement, but by splitting this insight in multiple dashboards it becomes that much easier to understand the individual dashboard. Second, within one engagement dashboard, you want to separate email engagement, website engagement and social engagement into different sections. By making sure you apply a structure to your data you guide the data story and make it so much quicker to decode.

Finally, getting insight is one thing, but when you understand the questions your users have you can accommodate for that in the dashboard. Meaning add list widgets, global filters, and toggle options to quickly drill down into the data. I wrote more about this in my blog post on filter and selection options. Once that answer is found, are there specific actions that are obvious? If an assignment escalation is relevant you can make sure that there is a global action for it and enabled it on the dataset.

If you want some good examples of how this is done try to install some of the Templated Apps that are available when creating an app in Einstein Analytics. I promise you the product and design team has worked very hard to come up with some amazing looking and well-targeted dashboards. There is no shame in gathering inspiration from them.

Put analytics front and center

One major mistake I think most of us do is thinking “well, the dashboard is created, now users can use it”. While that is true, what says they are able to find the dashboard? You might send an email informing them of the new dashboard, but that email is quickly forgotten. Instead, put your analytics solutions front and center where your users are. Since we are talking about Einstein Analytics, I would assume your users are in Salesforce, so it makes sense to make sure dashboards are embedded in Salesforce pages. The Einstein Analytics lightning component makes it extremely easy to get your dashboards in front of your users and even in a business context, so they get insight where it is relevant, in other words adding value. An example could be to show engagement details on your campaign overview. Just make sure that if you embed dashboards on record pages, make the layout compact, nobody wants to have a ‘scolling monster’. Oh, and Salesforce classic users can also embed their dashboards.

Another way to get dashboards in front of your users is by leveraging the Einstein Analytics mobile app. If you are by the coffee maker and you get asked about that marketing initiative you started yesterday – well, chances are you have your mobile in your hand or pocket and can quickly get that dashboard loaded and find the answer.

Educate your end-users

Some users regardless of how much you try to accommodate for their needs might still find it hard to use the dashboard. Maybe they do not know that they can click on a chart and explore it further or that they can collaborate on charts using chatter. Hence you should expect to provide some level of training to your end-users especially for those more comprehensive dashboards.

Now training can come in different formats. The typical one is hands-on training, but frankly, that rarely happens at least not for basic users that are meant to just view dashboards. There is often an assumption that they would be able to navigate the dashboard because “that’s very simple”. I’ve learned my lesson implementing Einstein Analytics, it’s not always that simple for all users. So what can you do? Some charts might need a little bit more context to understand, in order for some users to understand what that metric means. Well, leverage text tooltips, it’s a brilliant way for users to get more details by simply hovering over a textbox. Another option you can leverage for a single chart or the whole dashboard is to create a YouTube video and link to it in the “Onboarding” section in the widget formatting. This way users can view a quick video directly from the dashboard giving them instructions on how to explore widgets further, that they can use the mobile app or they can apply actions to records directly from the dashboard. It’s really up to you to decide what training to provide by observing your users and when in doubt don’t be afraid to ask them.

Get executive buy-in

While the above sections give you concrete tasks on how to achieve user adoption through a targeted dashboard and leveraging the features of Einstein Analytics there is one thing you cannot forget, which is equally important and that is executive buy-in. In fact, I would say before even considering touching a dashboard in Einstein Analytics the most important thing to ensure is executive buy-in. It may sound obvious, but it has such a huge impact on the solution if the executives are supporting the work and reinforcing the usage. Executive buy-in can come in many shapes and forms; one natural is they set their signature on the dotted line. However, that doesn’t mean a successful implementation. Rather I would strive for executive involvement throughout the process.

One time I was delivering an Einstein Analytics 3 day training/workshop and during the session, the executive sponsor was available in the room. He did do some of the exercises, but the most interesting thing was he was there when the attendees were playing with their real data. One of his biggest desires was to get insight from some external data, which had often seemed like a black box of information. Armed with excel sheets the team tried to import the files to begin building some of the dashboards that they really needed – only to fail. They discovered that these files were in such a format and lacking details that they would never get what they set out to. The interesting thing for me to observe was not the failure but that the executive sponsor with this experience now knew 1) what would create value for him and his team 2) what effort needed to create a good solution 3) why the data failed to be imported. Instead of dropping the project he knew where his help was needed and how he could help generate value so the time and effort they all spend on the analytics solution would be more valuable. And don’t forget that it’s also the executive sponsor that is able to encourage and reinforce the use of the solution once the hard development has completed.

I hope this gave some food for thought on how to ensure user adoption. Unfortunately, there is not a scientific guaranteed way to ensure your investment and effort is not wasted, but I believe these tips help the process. If you have any additional tips feel free to drop a comment below.

How useful was this post?

Click on a star to rate useful the post is!

Written by

3 thoughts on “Driving analytics adoption by delivering value”

  • 1
    Mark Tossell on October 9, 2019 Reply


  • 2
    TANUSHREE ROY on March 12, 2020 Reply

    Hi Rikke
    I am enable Adoption Dashboard dataflow which is UserActivityReportetldataflow. But due to this it cross my row limit which is 16cr. Now i dont want to delete the dataflow and the dashboard. But i dont want this much data.
    is there any way to stop data which is coming from salesforce objects without deleteing my Adoption dashboards and dataflows.

    Please suggest.

    • 3
      Rikke on March 12, 2020 Reply

      If it’s a dataflow you can add a filter node looking at date.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.