Martin's Blog

A little while ago I was asked to rate a Power BI dashboard. The person who asked, participated in a Power BI challenge (I’ll call it that because that’s the way they are being marketed) and wanted some feedback on the submission. I agreed on the condition that the feedback would be public and in the form of a blog post.

Wait…What?!

I know what you’re thinking: “What a cruel response! Why would you do that? That’s a <insert expletive here> move!”

Those that know me will know that it’s not my style to publicly shame or embarrass anybody. In fact, I am quite the opposite and would very rarely (if ever) get involved in the public soapbox that is social media. This person had a lot of courage to ask a stranger (me) for feedback, and even more so to agree to receive that in a public manner. So why the chosen response?

I think there are some valuable lessons here, and I see a growing trend that I understand but am concerned about: Young and aspiring data analysts (or BI professionals for that matter) participate in Vendor X‘s challenge, wanting to use it as a means to improve their skills and build their portfolio for potential employers. They really need some feedback to grow and get better, but Vendor X is not really interested in that part (or don’t have the skills to provide that feedback), and the participant is forced to try and ask for it elsewhere.

I get it. To Vendor X it’s an opportunity to market their products and/or services, and aspiring professionals are willing to participate in the hopes that they also get something useful out of it. It showcases everything that’s both beautiful and wrong about the free-market world we find ourselves in, and as someone who has been around this block a few times I feel that the younger generation can do with some good advice in this area.

You’re asking the wrong question

Imagine yourself as the recipient of a generic question like “Please rate my dashboard?”. How would you respond? I generally like to help people but feel that this question leaves me with only the following options:

1. Say “No”.
2. Be empathetic and provide a high score.
3. Be hyper-critical and provide a low score.
4. Spend an incredible amount of time to gather the context and analyze all aspects of the report before providing feedback.

I’d love to have number four as a standard response, but the reality is that I just don’t have enough time for all of it. Add the fact that it’s a fictitious “business problem” with sub-standard data and context, and it becomes really difficult to be constructive. As the person who is asking for this rating, how useful would a rating alone be without the relevant context and discussion? It wouldn’t be fair either way (good or bad), and I think that there are better questions to ask if you really want good feedback.

And therein lies the first piece of advice to my aspiring technology friends who’d like to ask for feedback: Instead of asking a generic question, compile a list of ten questions that are both detailed and explicit. Provide the necessary context around each question, and ask it in such a way that the person who receives the question is able to answer easily and without spending more than 30 minutes on it. Ask about a specific aspect of a report or individual item, as opposed to asking about everything on an entire report.

Ask each one of your questions to a different person, and you will have extremely valuable feedback with multiple different perspectives. That sounds to me like a recipe for success, and much better than a generic rating. If you’re willing to show that you’ve done the work upfront, and trust me it’ll take a lot of work, I promise you that most people will provide generous feedback.

I digress. Back to the dashboard…

Lipstick on the pig

I affectionately refer to the development of visual reports as the “lipstick on the pig”, and for two reasons:

1. If you like pigs then a beautiful color lipstick will certainly make them more attractive.
2. Irrespective of the color of the lipstick, it still remains a pig.

The point I am trying to make is that you cannot fix a bad data model with pretty visualizations, nor can you create a good impression with a poorly designed report or dashboard. Why does this matter?

Most of the Power BI challenges I have seen, incentivize you to focus on the visual aspects of reporting only. And because of that you feel that you should add as many visual bells and whistles as you can, because that is what your submission will be rated on. It’s an inaccurate representation of the reality we deal with when working on real projects and with real customers, and it’s probably not doing as much for your portfolio as you would like. Don’t get me wrong, there certainly is a place for it and it doesn’t hurt to show off your skills by participating in these challenges, but most of the time it’s not just about how well you can put lipstick on the proverbial pig that determines whether you are an employable analyst or developer.

Needless to say I am not a big fan of these third-party vendor “challenges” and think that something like Workout Wednesday (WoW) is more useful, because it lets you focus on building specific skills rather than trying to impress with your lipstick.

That being said, let’s honor the script of this particular challenge and focus on the lipstick only.

The dashboard

Before I provide some feedback on the visual aspects of the dashboard, let me say this: If I wanted to be hyper-critical I could have provided a laundry list of items that are “wrong”, but that would not be helpful to anybody. Instead, I am going to point out a few things that I think most people still stumble over when they are trying to learn and gain experience, which is my (correct or incorrect) assumption of the person who asked for the feedback.

The image below is what the dashboard looks like. Take a few seconds and form an opinion of what you see.

My first impression is that it looks a little messy, but why is that my first impression? The author created different areas on the page, grouped the visualizations logically within each area and even used vertical lines to create a barrier or physical separation between the different graphs. Why with all of these things in place, is that my first “feeling” about this dashboard?

Alignment, spacing & consistency

The smallest of things on a dashboard can make a bigger difference than you think. Let’s explore the alignment of different objects in relation to each other, and how it can perhaps create an impression of “messiness”. We’ll zoom in and draw a few visual boxes across the page to highlight what I am talking about:

In the image above you can see that Sales Pipeline in the middle of the screen looks ever-so-slightly misaligned with Sales Amount on the left…and Real Opportunities (on the right) looks a little lower than both. It’s actually not the alignment that’s off here, although it appears to be…the Sales Pipeline textbox is one pixel smaller than its counterpart on the left, and the Real Opportunities text is two pixels smaller than the others which changes the relative alignment between them.

The two slicers with the grey backgrounds are also not an equal distance from the top of our yellow box, and all of these teeny tiny differences are enough to create the appearance of “messiness” to the naked eye.

Another and perhaps better example of slight misalignment can be found with the text in the middle of the screen (image below), where the texts do not align at the top or even the middle.

Consistent spacing is just as important, and the two equally sized boxes in the image below shows how there’s both different spacing on either sides of the vertical divider, as well as between the two dividers and graph in-between. Yes, we’re playing a game of pixels here and unless you get super nitpicky about your own work, you may just end up creating the wrong impression.

Whitespace & the effective use of boundaries

There’s a lot of talk about using whitespace effectively, but sometimes too much whitespace and the lack of distinct boundaries can have a negative impact on the visual perception of a dashboard as well. These two concepts need to work in unison for a clean and efficient appearance. As an example, let’s take a closer look at the bottom section of the dashboard, which contains a vertical divider or line to separate the two graphs. I’ve drawn a yellow box around the divider to highlight it.

In my opinion at least, there’s probably too much whitespace at the top and bottom of this divider and it’s difficult for the naked eye to register it as a distinct boundary. Instead, you may find your eyes wondering across to the next graph without taking the boundary into consideration.

How do we fix that? One way would be to extend the divider and “cover” the top and bottom of the visualizations. This slight adjustment may just give you the necessary effect, and I would take it one step further by changing the color to match the horizontal line at the top so that it is more distinguishable (image below).

In general, the more variation you have in the shape and size of your visualizations the more distinct your boundaries have to be in order to create some separation. As an example (and even though still far from perfect) the image below shows that by just drawing boxes around the visualizations, the differences in shape, size & alignment between them become less noticeable. I’m not suggesting you draw boxes around everything, but clear and distinct boundaries help your eyes focus on a specific area.

Beware of the obvious

Something I see quite a bit of, and again I blame these challenges for creating the wrong incentives, is unnecessary or duplicated information. In the following section of the dashboard (below), we have a sentence stating that sales peaked in May and that the top product was SAAS.

This information can easily be determined from the graph itself, and we can utilize the interactive capabilities of Power BI to change the Total Sales Amount at the top when we click on a specific product (or month)…which makes the sentence at the bottom unnecessary. Don’t be tempted to add another data point and/or visualization just for the sake of adding it.

Analytics is iterative

While I could carry on and provide feedback on more items in this dashboard and my interpretation of whether they are good or bad, I’m not going to. The reality is that if you were working with a real customer, most things would have been picked up and addressed as part of that feedback loop. My opinion simply doesn’t matter that much, because I am not the one who will be using this dashboard.

Every part of the analytics process is and should be iterative. It is ultimately your customer who should decide what works for them, and how they would like the data to be presented to them. Of course you’ll need to bring your own skills and creativity to the table, but it’s your job as analyst or developer to understand your customer’s needs and preferences first, and go through as many iterations as necessary to get to the end result.

If these challenges are to be of any value to you, it would need several iterations with some feedback at every turn. I would even go one step further and say that if I were a potential employer, I would be more interested in the process than the end result…because the process tells me more about your skills and level of experience than a one or two-page dashboard. If you want to expand your profile in order to attract potential employers, my best advice to you is the following:

• Ask the right questions to get the best feedback. Feedback is a critical part of your growth, but do the work upfront to maximize your results.
• Get real world experience. Find a non-profit and do a project for free…you’ll get a much better return for your investment than participating in third-party vendor competitions alone.
• Blog about your project. Showing potential employers how you are working towards improving your craft will make you more employable.
• Find a mentor. Reach out to somebody in the community and ask them to mentor you during your non-profit project. Be prepared to do all the work…a mentor is not a trainer and you should only expect them to point you in the right direction, not solve all of your technical challenges.
• If you need to accelerate your own learning, pay for training. If you’re not prepared to spend hundreds of hours to perfect your craft or do not have the time to get up to speed, training is your best option to accelerate that process. Invest in yourself, otherwise nobody else will.