Why your dashboards suck, and 4 things you can do about it

Dashboards suck. Big time. Why? Because the people who build them are not thinking about their purpose.

OH SNAP.

Let’s put together a list of what dashboards are not for:

  • Showing you what’s going on
  • Updating stakeholders on key metrics
  • Marrying data from different sources
  • Offering a “heads-up” view of the business

Yep, none of that. DOUBLE I’M ON MAURY POVICH AND YOU AIN’T THE FATHER SNAP.

Dashboards are 100% absolutely not at all in any way whatsoever about information. That’s what reports and meetings are for. Dashboards are for one thing and one thing only: action.

Let’s break this down into two separate views. First, we’ll think about why this is the case. Then, we’ll explore comparisons in other parts of our life. Got 5 minutes? Let’s do it.

Why dashboards aren’t about “information”

When I used to see comparisons between Google and Facebook’s engagement metrics, I wanted to stab myself in the brain. The thinking was that since people were spending more time and consuming more pageviews on Facebook that Google was being bested.

Looking at this claim in terms of purpose rather than metrics, we can see this means that both of these sites are successful (and if Facebook’s metrics just passed Google’s, it also shows that Google was a hell of a lot more successful at that point in time than Facebook, not the other way around). Why? Because Google’s sole mission (for search) is to send people AWAY. Facebook’s mission, on the other hand, was to get people to stay. So, if Facebook’s visits, session duration and pageviews are just passing Google, that means Facebook (at the time) had a very, very long way to go to best Google.

Regardless of who’s winning, which site breeds more productivity?

Remember the claim up there? People don’t know why they’re designing dashboards. Dashboards that are trying to give people the lay of the land, provide information, context and a wide view are not functioning properly. Dashboards built this way draw the user in, asking them to spend time clicking different views and tabs, taking mental notes, drawing comparisons, and figuring out whether or not everything is okay. The dashboard is pulling when it should be pushing. The measure of a good dashboard is its likeness to Google, not Facebook. Good dashboards put people to work.

Pushing isn’t exactly the right word. Dashboards should be shooting you out like a human cannonball toward the things you should be working on. Dashboards need immediacy, urgency, alerts, all screaming at you like wild banshees about specific parts of your business, site, social presence, etc. LOOK AT THIS RIGHT NOW OR YOU WILL PERISH. That is what dashboards are for. They are tools for tacticians to target their efforts, not for generalists to broaden their understanding.

How this looks in other places

Let’s take a look at dashboards elsewhere. Cars, planes, stuff like that, for example.

Now before you start picturing a 747 cockpit that looks remarkably like your corporate dashboards, let’s think about a car and build from there. In my car, my dashboard tells me some pretty basic things about my engine (tach, temperature), speed, and fuel. There may be some niceties like outside temperature, etc. I may have a GPS that is telling me where to go, too. And my dashboard has a bunch of unlit indicators about other things like tire pressure, oil levels, windshield fluid, airbag status, etc. etc. etc. Keyword: unlit.

The purpose of all of this is twofold: one is sort of like realtime reporting: I can indeed see some non-critical basics (that could become critical, like speed) when I’m not having to pay attention to a traffic situation or a more pressing matter. But the second purpose (the important one) is what I normally cannot see in the unlit area. Things that will come alive when and only when I need to know about them. My GPS tells me when to turn and shuts the hell up in-between. Sure, I can look at it, but that’s again in my downtime.

Can you imagine what driving would be like if all of these indicators were on and I had to check in on each one constantly to understand status and what I should do in response?

Graduating to the 747, we have a lot more going on, but again, almost everything is unlit or not intentionally drawing attention. If the pilot wants to get a swath of information about the plane’s status, he certainly can by looking around (just as we can by leaving our dashboard and getting into our analytics tools and other reports/meetings). But she’s looking around in his downtime. She’s not looking at 300 dials when taking off. She’s not looking at 300 dials when the engine is on fire, because when the engine is on fire, there is a HUGE amount of focus drawn to the exact dials and controls the pilot will need at that moment.

The plane is a lot like your business. This data is available. It’s sitting around. It’s being produced in reports. But mushing all of it together into a dashboard is not the right thing to do. Mushing all of these dials into a cockpit isn’t the right thing to do, either (it was necessary before today’s technology), which is precisely why most cockpits are starting to look more and more like our cars (a lot of the dials are unused or redundant and have been moved to an as-needed LCD in front of the pilots).

So, great dashboards are…

QUIET when things are not whacky. You look at them, don’t see anything special, and know you have some time to focus on a project you’ve been meaning to pay attention to, you can schedule a meeting to think through new ideas, or you go dig into your data more deeply to find opportunities. A quiet dashboard is a signal that you have the freedom to work on new priorities. All dashboards need a quiet, unlit capability.

LOUD AS HELL when things are whacky. When you log in or view them, the dashboards will tell you immediately that you need to cancel your meetings, hold your calls, order lunch in, and focus on the buzzing alarms and dials out of their tolerable ranges.

TAILORED to the people who are going to look at them. Have a report or meeting if you want/need/have the time to talk about other peoples’ lives and wider status. Your dashboard should be about your plane, not all air traffic over the pacific.

CONTEXTUAL to purpose. If you have a table that looks like the one below, your dashboard is a complete and utter failure:

Why? Because this is out of context, lumped together, and not actionable. Is the purpose of your paid search homogenously to attract new visitors? Is every keyword indicative of a user ready to convert? Should the conversion rate of social be compared to your SEO efforts? Everything on the dashboard should live in the context of purpose. You should have a page about audience attraction, a page about high-propensity converters, a page about amplification and content acceleration, a page about different audience segments and expected results. Then, when you look at the dashboard, you know specifically where you are deficient and maybe even what you need to do about it, not just what the conversion rate of your display campaign is (which is completely useless to anyone who can do anything about it, when presented at this level of granularity).

So, the next logical step is to share some good dashboards with you. If you’re interested in seeing the types of dashboards that actually cause action and keep information in context, leave a comment or contact me on twitter (@evanlapointe).

 


No Trackbacks

8 Comments

  1. One thing that sticks out about any “dashboards suck” post I read (including this one) is there’s a bigger underlying issue with company culture and training. As you state, there *should be* a difference between a report and a dashboard. I’d argue that to anyone you are distributing these things to, they are the same.

    So, how do we overcome this? It’s not a training issue, IMO. Every executive I’ve ever worked with always wants to see the same report, either due to comfort or laziness. When the “dashboard” always has different metrics on it, at best they think you did “analysis” and at worst, they think you are trying to hide useful information hoping they wouldn’t notice.

    Do we go ahead and force people to get comfortable with the concept of actually using the dashboard concept you are advocating? Is it we send out “reports” with an executive summary written in plain English as a means to highlight what’s important? Something else?

    Posted October 23, 2012 at 12:02 pm | Permalink
  2. The “dashboard vs. report” distinction Randy brings up is a good one. I spent a few years saying reports should only get regularly disseminated if they have an operational purpose, and dashboards actually were the only thing that should be scheduled (and argued that a scheduled dashboard…combined with triggered alerts…made more sense than a “real-time dashboard”).

    I read the first part of this post and disagreed with the “what they’re NOT for” list. I’ve always felt that they’re certainly not the be-all/end-all comprehensive repository of information, but they certainly are the “performance measurement check.” The more I read the post, the more I think I wholeheartedly agree.

    At the risk of faux paux-ing the blogosphere, I actually mocked up an “ideal” dashboard that gets to the pants-on-fire-you-better-act concept: http://www.gilliganondata.com/index.php/2009/11/09/the-perfect-dashboard-three-pieces-of-information/

    I think that aligns with your point. I know I spend a lot of time up front with clients guiding them to “less is more” for any recurring delivery of information. Usually…with success.

    Posted October 23, 2012 at 12:24 pm | Permalink
  3. Alyson Murphy (@AlysonMurphy)

    I would say there are two very distinct purposes for creating a dashboard. The first being the one you have outlined. The one that tells you what is varying from normal deviation. These are definitely great for people who own KPIs.

    However, if you subscribe to the idea that a company should have goals, resulting business objectives and KPIs that say whether that business objective has been met, then I would say that there is worth in a dashboard that illustrates the change you are trying to produce. I may have 1 project or 25 that are trying to change a KPI and I will measure their effectiveness through testing. But a good leader is going to want something that says, “See! What you are doing matters. All 20 of us are changing this metric and making the business better.” Creating and maintaining a data-driven organization requires people to be excited about watching their efforts move the needle. Nothing does that better than a dashboard that highlights the KPIs of the company’s goals. That however, would disagree kind of with your two first bullets of what a dashboard is not.

    Posted October 23, 2012 at 2:06 pm | Permalink
  4. Great stuff! Let me see if I can put some coherent thoughts together:

    @Randy:
    You make a hugely important point. Here’s the way I would [mis]handle it:
    - If there is already a dashboard in place, I’d examine it to see if it meets some criteria around causing action and not causing distraction. How, on a person by person or department by department level does the dashboard specifically address and reformat information in a way that causes action when action is obviously needed and stays quiet and passive when none is. If it can’t pass that test, it doesn’t work well enough.
    - If there isn’t a dashboard yet, during the requirements gathering phase, set up the dashboard the way I outlined in the post. Don’t get bogged down in KPIs as much as expected results or expected interactions at a segment or use-case level. The dashboard and conversations around it should be about what the business expects in order to build a system that highlights unexpected results that require action.

    @Tim
    You’re dead on. It’s the up-front stuff. This isn’t a post about dashboards as much as purpose. Yes, today dashboards do a lot of this stuff in my opening list. But those things do not drive action, at least not in the same way another approach would. In my mind, the purpose of action is a lot better than the purpose of information, at least according to the structure and economics of a complex organization. People aren’t being paid to be informed (most people, at least); they’re being paid to block out the bad and usher in the good, proactively.

    @Alyson
    Super great points here. What I’d ponder on this one is what good are my “progress against goals” metrics if I don’t understand the components? I look at these types of dashboards like a canned food drive at a high school where we put the big thermometer up with how much progress we’ve made. Yes, it shows progress. No, it does not show anything that suggests where we are leaking opportunity or experiencing better-than-expected results in a window we can take advantage of. I’d classify this as a “report,” rather than a dashboard. It’s a piece of information I can receive. The dashboard is much more sophisticated, often backed by models or tolerances that direct me toward the right information or context to jump immediately out of the dashboard and act. All that the KPI dashboards accomplish is the creation of open-ended questions that the dashboards I’m talking about are already answering, thus they are simply status reports. Dashboards also are fairly bad at accomplishing the goal you outline in terms of communicating success and allowing the team to see, enjoy, and communicate things worth getting excited about. On the other hand, a true analysis, presentation, or graphic is far better for this goal, scheduled weekly, monthly or quarterly to ensure that small hills and dips don’t distract from the overall sense of accomplishment (and what can be learned from a bigger-picture perspective). Too often, in dashboard format, the bigger message is lost in the day-to-day minutia and volatility. This certainly doesn’t invalidate what you’re saying; I’d just approach this type of asset in a different way, keeping the dashboard itself highly tactical and actionable to suit the everyday needs and status rather than high enough where it’s disadvantaged at either the tactical or strategic levels (not good at either).

    Posted October 23, 2012 at 3:13 pm | Permalink
  5. Alyson Murphy (@AlysonMurphy)

    @evanlapointe:
    First, I am really not trying to argue that the type of dashboard you are describing is not something you should have. I believe strongly in that style of dashboard. I have created many of them. I am also not advocating that this should in some way substitute tracking the value of the individual changes and doing testing at the project level.
    That being said, I still think there is worth in the “canned food drive—woohoo we are working together to reach a goal dashboard.” I’m not saying that it is for every team and every leadership style. However if you have hundreds of people working on your websites and proposing ideas to make them better, it’s pretty useful to focus that group. One way is to set big over-arching goals and track them. I am not advocating that this dashboard is reviewed in isolation or by every single member that touches the websites, but rather whenever you go through the status of the projects that are being worked on. “Look, we just wrapped up Project 855. This project was meant to do this and change this metric, and you know what? It did. Good job to Janie’s team on that one. You know what else? Remember our company goal to be up in metric ‘x’ by this much? Check out this amazing visual graph that illustrates our metric ‘x’ for last year, the projected value of what ‘x’ would be this year, and our actual results for this year. Check that shit out! We are way up! Part of that is going to be things that Janie’s team worked on, part of that is going to be successes from other projects earlier in the year. If we keep this up, we will reach our goal. That stuff is great. Go team.” Does it make you jump out of your chair and do something specific? No. Does it recognize that business is actually done by people you need encouragement and reinforcement? Yes. Does it help keep the projects focused by forcing someone to say “will this project help us achieve that specific goal?” Yes. A couple of years ago, I would have said that this is kind of dumb. You aren’t specifically doing something with this piece of data. As I’ve watched our organization start to get people from every department to think about data and think about the specific goals of their projects, I have come to realize the need for unified goals and unified momentum towards those goals and the data that backs that up. If you are going to create goals that are founded in data, then you need to show people some data. You can simplify it, and tailor to the people you are talking to, but you still need to show it to them just the same.
    We may be having a bit of semantic disagreement though. You said you would call this a report. To me if you look at something over time, it’s a dashboard.

    Posted October 25, 2012 at 11:24 am | Permalink
  6. Aah, yes. Gotcha. I think we are more or less in agreement about the value here. You are 100% right that there is value here.

    Here’s what I have cooked up in my tiny brain:
    Dashboard: moment-by-moment view designed for reaction. Often reflects highly granular timeframes, even realtime. Corollary = dashboards in other things like cockpits, cars, etc. GPS may tell you progress, but overwhelmingly purpose is to inform the pilot of things they need to react to.

    Report: backwards-looking view where granularity is appropriate to the message. Can be backwards looking from this very moment, but is designed to smooth out volatility to ensure the core message, KPIs, etc. are communicated clearly. Reports can be generated frequently and looked at often. They may even live on a screen that is constantly up in a hallway. Differentiating factor is that the granularity and volatility cannot be interrupted mid-cycle, which would introduce the volatility and signal noise we’re trying to avoid in the first place. Reports are intentional in this way. Reports is a big universe of these types of things and many others (all the stuff we call reports today)

    Analyses: you know better than anyone what these are.

    I find that clearly differentiating between reactive dashboards and other types of assets by name keeps things a little clearer, but that is probably me being asinine. What you’re suggesting is 100% absolutely needed, valuable, etc., it’s just a matter of what we’re calling it and the context within which that information is presented to ensure the core message is clear. 99% of the unanticipated work I have experienced as an analyst was because a message in a report was sidelined by an unintended other message. Things like:

    - That’s great, but I’m looking at the CTR on our display campaign…why did that fall (commence meaningless work)
    - I’m glad we’ve made so much headway on this project, but in the last 3 days, it looks like the line is suddenly lower… (usually explained qualitatively or attributable to regular volatility)

    It’s these risks I try to avoid by compensating with language. Asinine? Yes. Helpful in my past? Also yes. The point of the post was really to say if all your dashboards do is status, you’ve stopped short; not to say that if you have these dashboards or reports, that they are bad in any way.

    Posted October 25, 2012 at 11:40 am | Permalink
  7. Andy Gibson

    @Evan – I’m interested in seeing some good examples of actionable dashboards. Our clients have a tendency to ask for dashboards that are pretty and full of data, because they think that is what they need. No decisions actually come from these dashboards, they just like having them (and I would suspect very few actually check them routinely).

    Posted April 4, 2013 at 9:42 am | Permalink
  8. Andy,

    I’ll work on a post with some examples. Help me out, though: what types of audiences are you dealing with?

    Posted April 5, 2013 at 9:26 am | Permalink