If your sales dashboard tells you that sales are down, but not why, you don’t have a dashboard
I tweeted the following a while back:
If your sales dashboard tells you that sales are down, but not why, you don’t have a dashboard
It’s worth expanding on this a little more than the 140 characters allowed on Twitter. What do I mean by this? Simply, if your dashboard shows key performance indicators, or a list of metrics, but doesn’t provide provide the accompanying information that helps you determine the cause of variations, then you don’t have a dashboard.
If you search for sales dashboard, kpi dashboard, or business dashboard on Google Images, you get similar looking array of dashboards. These typically have highly colored charts taking up most of the prime screen real-estate, and most fail at helping you understand any variations in the metrics since you last viewed the dashboard. Would you consider this a dashboard?
It’s one of the first results in Google Images search and apparently is an example of the typical dashboard that this company delivers to their customers.
The design is as bad as most – so much emphasis on color and effect. Look at the first cylinder bar chart – is the sales goal $120,000 or $$150,000. As stated underneath, it’s $142k – what extra information is the bar chart telling us? Absolutely nothing, it’s just there, taking up space and rarely, if ever, changing.
The 3D column chart is equally as useless. If those are the sales values for April, my next questions will be: how does that compare to March values or to the forecast revenue? This dashboard does not empower the user with this information.
This dashboard is nothing more than a display of all the chart types and 3D effects that the software was capable of producing. One could argue that this is only a mock-up, but follow the links to the Google searches above – virtually all the examples are like this, whether produced by vendors of dashboard products, vendors who create a dashboard for their users, or, most disappointingly, by companies that are consulting in the dashboard/data analytics space.
A dashboard must show the key performance metrics, display comparisons to prior time periods, to expected values, and then, right underneath, show the information that explains the variations in these numbers – no clicking to drill down or selecting another tab – as soon as you make your user work to get the information, you can be sure that many just won’t bother.
What should a dashboard look like? The key performance indicators should be in text up the top left. But wait, text is uncool, we’re trying to get away from rows of tabular data surely? Yes, rows and rows of information are not helpful, but when the manager goes to their sales dashboard, the primary number they want is “how much money did we make this month?”. They don’t want to have extract this information from a poorly designed cylindrical bar chart. Then answer the next questions with text values – “How much should we have sold?”, “What’s the difference?”
See the charts? Charts are for comparisons, but not a 3D chart that takes up a quarter of the screen, instead a sparkline that quickly allows us to compare sales over the last year, or where we are compared to forecast. A few more rows may break the KPI into different time periods, and present other KPIs.
Then move onto the supporting information – performance of sales people, number of customers, average sale size, timeline of sales, expenses, discounts, products, regions, and so on – anything that is helpful in understanding variation, nothing that is not. If your dashboard shows the local weather, and you don’t sell umbrellas, your dashboard is not fulfilling its potential.
Detractors will complain that there’s too much information – people like eye candy and revving gauges, dashboards should be simple. Yes, dashboards must be easy to read – thoughtful design and good use of color is critical. But combine this with information that actually solves the business problem – how do I improve sales – and your users will embrace the dashboard. There may be a bit of a learning curve, but if you are deploying an initiative that could fundamentally change the way your business performs, would you expect your employees to only take a few minutes to understand that initiative?
Are your dashboards fulfilling their potential? Data Driven creates dashboards that solve business problems and improve your company’s performance. Based in Boston, Data Driven serves customers in New England and the whole of the US through data analytics and dashboard design.
gopalakrishnarao1
Excellent . How can I implement the same in my organisation?. Dash boards details how to create and how to present step by step guidance is required.
Alex Kerin
There are lots of excellent resources out there which will help you on your way:
Chandoo’s site: http://chandoo.org/wp/excel-dashboards/
My guide on how I created an Excel sales dashboard: http://www.datadrivenconsulting.com/2010/01/data-visualization-challenge-results-part-one-addressing-the-data/
http://www.datadrivenconsulting.com/2010/01/data-visualization-challenge-the-objective/
http://www.datadrivenconsulting.com/2010/02/data-visualization-challenge-my-dashboard-design/
Jon Peltier’s site on advanced excel charting: http://peltiertech.com/
And lots of others.
The key to a successful dashboard project is knowing what business problem you are trying to solve, and presenting the information in a way that allows the end-users to solve the business problem. It doesn’t matter what tool you use – it just happens that Excel is flexible and ubiquitous, so is a great tool to use.
If you need more help, feel free to contact me.
Daniel
Perhaps I’m arguing semantics, but isn’t there a difference between a “dashboard” and, as Stephen Few calls them, “Fractal Analytical Displays” that incorporate more data analysis capabilities for understanding the nuance behind the KPI? You say “If your sales dashboard tells you that sales are down, but not why, you don’t have a dashboard” but by analogy, my car’s dashboard tells me to check my engine without telling me that the temperature sensor is broken. Aren’t we talking about two different displays for different purposes or audiences?
Alex Kerin
I would argue that a dashboard does not need drill-down capabilities, but to be useful and be adopted as part of the company culture, it does need the supporting information right there on the screen.
Your example of the car dashboard is a good one to discuss, and perhaps begins to explain why many current business dashboards look like they do, and that perhaps a different choice of word would have served us better.
A dashboard must serve the end-user to enhance the task they are set. In a car, our job is to drive the car safely – we are provided the basic information needed to do that (speedo), plus alerts that something may need our attention that affect this safety (check engine light – don’t want to break down in fast traffic, or low pressure in a tire). Our primary use for the car dashboard has been fulfilled – telling me that the temperature sensor is broken is useful later down the line, but during this task (driving the car safely) it’s information that just detracts.
Compare that to a dashboard provided to a sales manager. Their dashboard task is to improve sales. Just telling them what the revenue from last week is does not allow them to complete that task – they need to know who sold what (or didn’t) and for how much and for…..
Designing a dashboard is all about the end-user, what their business task is, and providing them with the information that they can influence. If they have no influence over the information on a dashboard, then the dashboard is not helping them solve their set task.
Alex Kerin
And as an addendum, I would argue that a business dashboard shouldn’t have drill-down capability (at least to any great extent) – if the information is important enough, you mustn’t make the user search for it – it should be right there. Certainly drill-down analytics have a critical role, but I don’t think they should be in the hands of the sales manager. Providing drill-down is an easy way out of actually considering exactly what the end-user needs
Stephen Few
Alex,
Your comments clearly set you apart from most of the folks who promote dashboards. Thanks for standing up for dashboards that actually inform.
Daniel–My term for screens that simultaneously display data from multiple perspectives in multiple charts for analytical purposes is “Faceted Analytical Display” rather than “Fractal Analytical Display.” They key is that they allow people to see data from many perspectives–that is, from many faces or facets. I’m not sure what a Fractal Analytical Display would do.
Steve
Daniel
Stephen – Thanks for the clarification – my typing fingers got away from me. At least I accurately remembered that there was a difference between a dashboard and a deeper analytical tool!
Alex – Your extension of the dashboard analogy works for me, and I agree that not everyone wants or should have access to that richer analysis capability. Do you find that sales figures and other “dashboardable” KPIs are easily explained by one or two other data points? In other words, can the sales manager’s questions be answered in a satisfying way with just one or two more spark lines?
I don’t have enough experience in business analytics to know whether causes and effects are cleanly related in things like sales or revenue. In my own work, changes often have multiple small causes, or are caused by something we’re not measuring, and thus can’t appear on the dash.
Alex Kerin
“In other words, can the sales manager’s questions be answered in a satisfying way with just one or two more spark lines?”
I would answer no, the sparklines allow a greater understanding of that KPI – how has it changed, where are we (graphically) to forecast, but there needs to be more information as well – other metrics, breakdowns of the metrics by sales person, products sold, and so on. There’s even room for text based statements – e.g. Sales Person X sold 45 products more than the average, An unusually low number of widget Y were sold), but this information comes underneath the KPI display – I didn’t show it in this post, but this post should give you the an idea of what I mean: http://www.datadrivenconsulting.com/2010/02/data-visualization-challenge-my-dashboard-design/
That dashboard isn’t really a sales manager dashboard, so many elements would be different in a real scenario, but it begins to explain what I mean.
Tom Quist
First off, this is a wonderful discussion. I appreciate you bringing this up, Alex (you linked me to this discussion from Chandoo’s site). I’ve thought about this a good deal and I tend to agree with Daniel’s point mentioned above, that dashboards can’t possibly include enough information to truly explain why any given trend is occurring in the complicated environment in which most of us work. To me, the dashboard, and performance metrics in general, are more of a tool to identify where and when a problem might exist. The variables that could be causing the problem are often too many and too complicated to communicate in data form. For instance, John Doe’s sales might be down this month for all products, but does your data tell you that he is having marital problems at home which are affecting his work? I think the manager’s job is to be aware of problem areas and then investigate them the old fashion way – by talking to the responsible department or employee. Of course, any data to verify claims can be useful – I just don’t think that it’s always available.
I agree with your point (Alex) that too much interactivity requirements can be burdensome for a manager. When I recently designed a dashboard for our local government, I tried (and I’m still trying) to find a balance where users can gain information “at a glance” while allowing them the flexibility to drill down should something catch their eye. That gives the power of determining what is important to the manager rather than the designer of the dashboard.
Again, thanks to everyone for sharing their thoughts – this is a great topic for discussion!
Alex Kerin
Tom, thanks for your comments. You’re right of course that you couldn’t include every possible reason why a KPI is varying, but I suppose I’m really rallying against the dashboard where you have 6 highly colored charts covering six KPIs, but absolutely no further information. Add the trends, show recent trends for regions or sales people….
If I can at least see that Bob’s sales have dropped off in the last week, and that he reached his quarterly goal a week ago, I can pick up the phone and see what I can do to incentivize him to continue selling this quarter…. It speeds up my ability to understand and solve problems. That is of course until we require all employees to be implanted with a chip that does tell me about marital issues…
Tom Quist
I definitely understand your logic and I do agree with it. I think that’s why I like the drill down approach. The manager can start with basic information and if she’s satisfied with what she sees (sales growth) she can move on. If there is some concern (declining sales) she can dig a little deeper into the components of sales.
One of the other reasons that I prefer the drill down approach is because of the limited “real estate” allowed on a one page dashboard. I’m a big fan of Sparklines and miniaturized charts, but I think there are many scenarios where a full size chart (complete with axes and possibly data labels) is more appropriate. You can only fit so many of these on one page of a dashboard (hence the advantage of the Sparkline).
I like the chip implant idea. If we all had chips in our head, would we need dashboards though?
Thanks again for your thoughts!
Tom
Dave Sherman
I love the post and the follow up because it cuts to the chase on the real issue with the effectiveness of BI and any kind of data driven decisions. Get the data in context, visualize the data, think and then act on the data.
The natural evolution of this kind of data driven mindset is to bring it forward to mobile devices.
There are many challenges in doing this next step. What are your thoughts?
My list is here: http://bit.ly/cGFDQc
Steve Fernandez
Interesting topic. In my experience the hardest part of developing a dashboard or score card is the balance between what is really a KPI and when the report, dashboard, score card, etc. becomes a tool – meaning a front end to a ton of data. Your topic fits in with making it clear on what’s important and why. And, that’s usually context; either goal based or trend.
I tend to use the following terminology.
Dashboard – something you glance at to make sure everything in order
Report – more detail to understand how things are working (perhaps metrics by functional department)
Tool – when you need to dive in-and-out of the data to start asking specific questions
How you add that context and “why sales are down” to the Dashboard is the real skill.