There was a time when C-Level executives roamed the halls and pounded tables demanding more dashboards to give them insights into KPIs and day-to-day operations. Best-in-class tools like Tableau have done an amazing job of turning these demands into a reality. In this new reality, executives and mid-level managers typically have dozens of dashboards custom-tailored just for them. Problem solved, right? Well, not quite.
Studies show that very few people ever actually look at dashboards. Instead, they now roam the halls and pound tables asking for answers and action, not dashboards. The good news is that the answers executives seek are typically locked inside their dashboards – they just don’t have the time and skills to monitor them 24/7, mining them for important insights and changes.
The solution? Customizable Tableau alerts built right into existing dashboards. Tableau superuser Anthony Chamberas thought Tableau’s built-in alerts would provide the solutions that he and other analysts needed. However, Chamberas quickly ran into the limitations in Tableau’s built-in alerting around:
- What you can monitor
- How often you can monitor it
- Who can be notified
- What channels you can use to notify them
- How well you can customize them
- How much context you can provide
Tableau’s alerting tool also requires business end users to set up alerts directly in the dashboard, requiring them to take on the role of an analyst. Chamberas found that people who need alerts don’t want to visit a dashboard to get them, and they also had no interest in becoming their own analyst.
Make no mistake, Chamberas is a huge Tableau fan. In addition to serving as co-facilitator of the 4,000+ member Boston Tableau Users Group for the last seven years, he is also a Tableau Ambassador and the leader of the new Tableau Developers User Group. With that background Chamberas decided to build a tool with all the functionality his fellow Tableau analysts and business users were asking for: a tool that could easily relay intelligence directly from dashboards into the email boxes and Slack channels of those who can act on them – without them having to decipher a dashboard. I had a chance to interview Chamberas about what drove him to build a more customizable Tableau alerting tool for executives, managers, and analysts.
Jason Bailey: Analysts have put all of this effort and energy into distilling raw data into visualizations so that people can just look at dashboards and get all the answers that they want. Why take the information back out of dashboards and put it into emails, Slack, or tools like that?
Anthony Chamberas: I think you just answered part of it, which is that all of this energy has been put into dashboards, and we’re hearing over and over and over again either from the people who have built them or the people who are supposed to use them that people aren’t using them..
We were talking to an executive at Pitney Bowes yesterday and she said, “Yeah, I don’t want to go to a dashboard. Just tell me what I need to know.” We literally hear that in every conversation we have with executives and business users.
The vision for Winnow Analytics is that it can just sit on top of your data and be constantly looking at it and notify the right people when something happens. And when I say “data,” I mean data generically. That’s the vision – it’s such an intelligent system that it just works. And I do see real value there.
JB: Got it. It’s more about getting the insights you need when you need them without having to go and stare at the dashboard.
So for a long time, Tableau has been very successful without any alerting, and recently they added their own alerting solution. Why do you think that they went as long as they did without alerting, and what is it about the new alerting tool that they’ve added that you think requires additional functionality?
AC: I think they’ve been so successful because it’s such a fun tool to use. It’s solicits this energy when you use it, and people love it. I think they’ve been able to ride that for quite some time. However, if you talk candidly to people internally, specifically in sales, even they saw a gap with how they want to use Tableau. They don’t want to have to be going in and looking at the dashboards, they want to be notified. They were the ones that actually pushed for alerting, not only because they were hearing it from their constituency, but also because they wanted it for themselves.
Tableau alerts caused this tension with the Tableau purists who believe that dashboards provide better visualization, interactivity, and overall customer experience. They didn’t want to boil all that down to something simple, static, and one-dimensional. So there is that internal tension, and I think that the compromise was “we’ll do some kind of lightweight alerting,” and what came out the other end was a little bit half-baked. At Winnow Analytics, we’re trying to pounce upon that and take advantage of it to really do it right, because we are hearing from a lot of people that what’s in there now just isn’t enough.
JB: Specifically, what would you say are some of the limitations of the current Tableau dashboard alerting solution that you’re trying to address?
AC: So number one is our premise that people who need the insights don’t want to go visit the dashboard to get them. In order to turn on the Tableau’s current alerts, users need to go visit the dashboard. Then they need to set it up for themselves. It’s a pretty simple process, but they still need to do it. They need to take on this role of analyst, even though they might be a business user.
From that point, it is somewhat limiting in terms of what you can monitor, how often you can monitor it, and who can be notified and the channels that they can receive it in. So when you set up an alert, it’s just for you. I think you can maybe specify others to receive it, but they’re going to receive the same thing that you receive, so there’s no notion of customizing. They need to be Tableau users, and there’s no way to put context around it.
So I could subscribe you to an alert and you’ll get an image. You don’t know why you’re getting it, what it’s about, what you’re supposed to do, and there’s no way to add anything to that. Then if I have a bunch of alerts set up, there is this notion of being able to manage them. They kind of go workbook by workbook and dashboard by dashboard to see which dashboards I have alerts set up for, and it’s not a smooth process. So again, we’re trying to centralize a lot of that management, not only from the administration side of it, but also from the end user perspective, so that they can subscribe and manage their own alerts.
JB: Got it. And when you say they “get an image,” is the alert just sending them the whole dashboard? Is there no context at all?
AC: No. There is a subject line, so you could put a context in the subject line, but obviously that’s limited. But that’s it. There is no personalization or ability to add prescriptive actions for the person receiving the alert to take.
JB: So the alert is essentially saying, “Here’s the dashboard again that you weren’t looking at and you don’t necessarily know what to do with.”
AC: It might repeat the metric that was selected for monitoring with the threshold value, but it’s a one-sentence thing, like, “Sales is over 1000.” Okay. So what? Why? Why is that important?
JB: Right. And it doesn’t not tell you any actions to take?
JB: So you already hinted at it, but specifically what you can do with Winnow Analytics today that you can’t do with the existing Tableau alerts tool?
AC: I’d say multiple communication channels is one – Slack, email, and text, as opposed to the single channel with Tableau alerts. The personalization by putting the person’s name in the message, but then also getting more specific about filtering. Providing their own specific thresholds and doing all of it with a single template rather than having to go off and set it up separately for everyone. So it’s templatizing it.
So the context … multi-channel, prescriptive, at scale. I think those are the advantages of using Winnow Analytics’s Tableau alerts, in a nutshell. Also, using machine learning means we can be more intelligent about how the events are detected and how the alerts and prescriptions are sent.
JB: Got it. There is typically a long and difficult data-mining process, which is partially why we have analysts. Are we at a spot where AI or machine learning can realistically replace that skill set? Is that the intention, or is it really designed to augment their skills? Or are there just some things that can be set to alerts and other things that are more nuanced that the analysts should handle? Obviously when folks are faced with dashboards, part of the problem is that there is some overhead in terms of being able interpret them and that requires a special skill set. How are you thinking about that?
AC: We’re still need the analyst to:
- Clean the data
- Shape the data
- Build the predictive model that’s called for
- Print it into a dashboard
- Design a meaningful layout
- Represent it graphically and visually
It’s the last step where machine learning can help close that gap between the product that’s generated by the analyst and what needs to be consumed by the business owner. There’s still a gap there. It’s close, but it’s just not enough.
JB: Can you give me an existing use case? It can be anonymized, but maybe an example where Winnow Analytics is solving a clear pain point for one of your customers? An example where maybe dashboards just weren’t cutting it and now Winnow Analytics is in place and it’s solving the issue?
AC: I’ll talk about one of our banking customers. They have a dashboard that is tracking the performance of the branch and three different product lines, and it wasn’t being used by the branch managers. So they put Winnow Analytics’s Tableau alerts in place to deliver that information to the branch manager.
The bank’s approach is to be positive as opposed to negative when it comes to notifications and alerts. So their alerts let them know when they’ve exceeded their target as opposed to just falling short of it. So that’s just been their decision to do it that way. But it’s now bringing that information to the branch manager rather than leaving them in the dark for the month.
JB: That’s great. So it brings up a topic around timeliness, right? I think my personal experience has been that often dashboards are almost a punitive tool. Managers look at them after things have gone wrong to figure out who to blame. Are you seeing that alerts more proactively help people get in front of being able to apply data and analytics to problem solving rather then using it retroactively to look back on what went wrong?
AC: Yeah. And it doesn’t have to be forward and backwards looking, necessarily. One of our advisors splits it out in a slightly different way. She worked with Tableau, and she breaks it out as monitoring versus enablement. So there’s monitoring, and that’s more punitive. You’re just looking at a process and working through what’s wrong, and penalizing or notifying when something is wrong. She says that that’s soul-sucking work. It’s not going to inspire anyone.
On the flip side, she prefers enablement more when it comes to analysis. Alerting someone to an opportunity or something in the data that might go missed, as opposed to pointing out something bad that already happened. It’s providing a very clear path to action. And in some cases that can be predictive, and in some cases that can be more historical in nature or real time.
JB: That’s helpful. And then obviously we’ve all have a million tools in place, and the last thing that a lot of folks want is to have to learn or install another tool into their daily use. How big of a footprint does Winnow Analytics have? What does it take to onboard and what does integration into people’s teams look like today?
AC: So my personal overarching philosophy when it comes to analytics is that it works best when you don’t know it’s there. So that’s my goal, branding, visibility, company reputation aside, my preference would be that people don’t need to know it’s there. It’s just invisible in there, it’s working well when you need it to work, and otherwise it’s in the background.
Now to get to your question about footprint and difficulty in implementing: We’ve extended that philosophy so that it should be very easy for all of the different personas involved, whether it’s the executive or the business manager or that analyst from the IT program.
We try to address ease of use or ease of implementation across those personas. That’s why we made it cloud based, to make it easier for IT. We made a simple web interface for the analyst so they’re not needing to go through a whole lot of complex logic. We designed very basic screens to manage subscriptions for the business owner, and then some for the executive who just wants to receive the notice. It’s as easy as receiving an email.
JB: Do you think that it’s a data problem that you’re solving or a communication problem that you’re solving?
AC: It’s a communication problem. We’ve been pretty intentional about not trying to solve the data problem. But it’s communication and execution and operationalizing. It’s getting the information at scale for people. That’s what we’re solving.
JB: Getting it into their preferred channels in a format that they can understand and act on, right?
AC: Yeah. Personalized. I mean, you think about what a lot of the BI tools are doing – they’re processing massive amounts of data, but not in a personal way. We want to bring that personalized use to it.