I’ll have the tall raspberry slice in the back, thank you. Looks delicious!

Considering the market-speak at the top and the beautifully-ineffective graphics on this slide, I’m guessing it was put together by a marketing person who probably doesn’t know the first thing about data visualization.

Has Perceptual Edge targeted marketing folks in its workshops? If not, maybe you should, since they seem to have a surprising degree of influence over what decision-makers expect to see in BI solutions.

Delightful! Our company is on the path to an SAP implementation. Given the propensity of people to choose the glamorous (defaults?)over the informative, I suppose that we’ll see lots (more) of this. Painful!

It’s a marketing slide, the only thing it is trying to communicate is that you want to buy the software. The software itself is perfectly capable of making a gradient-free 2 dimensional visualization. This slide is about potential, not about showing actual numbers that are relevant to the audiences business. Perhaps you should buy a copy and see if you can use the tool to make a chart you’d like to see before bashing it.

I’m with Jaime on this one, have you ever implemented BOBJ? Its easy to throw mud, but can u quote actual business user community members NOT liking this? It seems to me that your dealing with a “tomato vs tomatoe” issue here!

Gabe,

Just before starting Perceptual Edge I managed a large business intelligence team that installed, used, and supported Business Objects’ full suite of products. The decision to buy Business Objects was made just before I took the job. It was during this time, in part due to my painful experience with Business Objects, that I decided to start Perceptual Edge in an effort to improve data visualization within the business intelligence industry. That was 9 years ago.

Since then I have worked with thousands of people who use BOBJ. Not once has any of them expressed satisfaction with BOBJ’s data visualization capabilities. On the contrary, I have observed their pain and frustration firsthand. It is the empathy that I feel for them and those who use other poor so-called data visualization products that drives me to do what I do.

If you’re familiar with my work, you know that I don’t just “throw mud”, but also write thoughtful critiques that are based, not on opinion, but on a large body of scientific research.

Peter,

I would love to see evidence that BusinessObjects 4.0 has improved in its data visualization capabilities. I’d appreciate it if you would post examples that demonstrate this in my discussion forum. I’ll gladly showcase any convincing evidence in this blog.

Hi Stephen,
I work in the “hot-dog stand” marketing group in charge of SAP BusinessObjects BI. I think highly of your books on data visualizations and I do agree with your comment about the 3d graphic chart represented in this slide. I would like to take this opportunity to add a few comments to the ones that were made earlier by you and some of your readers.

1- This slide used here is about 2 years old. It was initially used to show some of the 4.0 things we were working at the time. It’s using a composite graphic showing a number of 4.0 graphic options available to users, a query panel and a ribbon bar. The comment you initially made on the 3d pie chart is bang on. SAP employees and customers made similar comments in the past. The bottom line is that marketing folks are no longer using it. Even a hot dog seller (I find this quite funny) knows when to drop something that doesn’t work on a particular street corner.
2- Although the supporting points of this particular slide are not as clear as they could have been, I think the headline “Engaging new user experience” is at the core of the work done in 4.0 and we stand by it. The 4.0 BI suite offers a more unified and engaging user experience across our BI solutions/tools with identical graphic options (same shared charting engine) and similar workflows based on a ribbon toolbar across our BI tools/solutions. UI design is not an exact science, but I can tell you that our product group has not worked in a vacuum. They have work in collaboration with UI specialists and have run numerous workshops with customers and BI users in order to get their feedback and understand how they work.
3. PowerPoint Slides. Ah the devil’s application everybody hates but all use… My (personal) take on this is that ppt slides don’t tell the story by themselves, they support it. The story told around this slide is incomplete because it lacks what the presenter SAID and it lacks the full context (what was said and shown BEFORE and AFTER this particular slide. I suggest the following document for reading more on What’s New in BI 4.0. This is a better and more useful reference point: http://help.sap.com/businessobject/product_guides/boexir4/en/xi4_whats_new_en.pdf

One more thing: I personally believe that business people by far prefer to show more information than less. That goes for charts and visualizations and presentations as well. I will not speculate as to why people do that, but I will say that: it is a reflection of our general lack of skills identifying and using the right prop (graphic, chart, slide, etc) when it comes to telling a business story. All is not lost, I see a number of initiatives and attempts at introducing tools for helping people. An example for business user: SAP BusinessObjects Explorer (try it at microfinance.sap.com). The tool “suggests” the best possible graphic for the data selected. Another example is Kuler.com for helping designers with color palette and theme choices. Maybe not perfect but certainly steps in the right direction.

Best regards,
P. Leroux

Hello Pierre,

Thanks for the thoughtful response. I’ve read the marketing literature about BusinessObjects 4.0’s new features, as you suggested. I don’t consider promotional literature a reliable source of truth about a product, but I did use it to identify possible improvements to your data visualization and analysis capabilities. Unfortunately, I found little of relevance. Here’s all that I found in the document:

Advanced Analysis (formerly Voyager)

1) A workspace sheet can now “contain up to four crosstabs and charts, and allow you to group related analyses” by using multiple sheets that are accessed via tabs.
2) “You can now control the positioning of parent and result members.”
3) “Aggregate calculation, which summarize a selection of members, can now be changed at runtime.”
4) “Each measure member can have its own aggregation.”
5) “You can now asymmetrically expand different parts of a hierarchy when multiple hierarchies are nested on an axis.”
6) “You can swap the axes of any sub-analysis independently from the main analysis.”
7) “You can reorder members to make comparisons easier.”
8) “You can sort by member names and values.”
9) “Conditional formatting can be applied to “columns, rows, or cells to highlight important results.”
10) Data from up to four sources can now be combined in a single sheet.
11) Calculated fields are now supported by richer formulas.

These are all capabilities that have existed in other data visualization and analysis products for many years.

Interactive Analysis (formerly Web Intelligence)

1) The following new graphs have been added: pie charts with depth (I assume that this refers to the absurd example that was featured on the PowerPoint slide), scatterplots, bubble plots, box plots, heat maps, and tree maps.
2) A few new functional additions to graphs, all basic in nature, such as the ability to easily change between bar, line, and area graphs.
3) A few additions to graph formatting capabilities, once again all basic in nature, such as the ability the adjust the height and width of graphs, including a few that are definitely heading in the wrong the direction, such as the addition of light and shadow effects.

It is an embarrassing admission that scatterplots and bubble plots have not existed in this product until now. The same can be said for the functional and formatting improvements that have finally been introduced. I’ll be interested in seeing your new graph types (ancient to most data analysts) to assess their effectiveness.

It appears that no improvements to data visualization and analysis have been made to Dashboard Design (formerly Xcelsius), Explorer, or Crystal Reports.

Regarding the silly 3-D stacked pie chart that appears on your marketing slide, you stated: “The bottom line is that marketing folks are no longer using it.” Even if few members of your marketing team are still using this slide, this doesn’t change the fact that you’ve introduced at least one absurd, completely useless graph into your products. This is shameful. I don’t need to see the other slides in this marketing presentation or hear the words spoken during the presentation to know that 3-D mounds of jello don’t belong in a business intelligence product.

Regarding an attempt to support best practices, you wrote: “I see a number of initiatives and attempts at introducing tools for helping people. An example for business user: SAP BusinessObjects Explorer. The tool ‘suggests’ the best possible graphic for the data selected.” Unfortunately, some of the suggestions that Explorer makes are misguided – definitely not best practices. You’ve tried to emulate the “Show Me” functionality of Tableau without understanding best practices in graph selection. In fact, overall, Explorer is an attempt to emulate functionality that exists in other vendors’ tools that are eating away at your market share without first taking the time to understand them and their appeal beyond a superficial level. For a more thorough assessment of Explorer, read my blog entry titled “Big BI Is Stuck”. Compared to a decent visual analysis tool, Explorer is extremely limited, buggy, and a poster child for bad practices. I appreciate the fact that Explorer is an attempt by Business Objects to better support data exploration and sensemaking, but unless you’re competing in the Special Olympics, you don’t get points for the attempt alone. When you’re as far behind as you are, tiny steps forward won’t keep you in the race. Fundamental change is necessary, starting by looking at the world with fresh eyes to recognize where you stand in relation to what’s needed. Going back to the “hotdog stand” analogy, by only listening to the folks who flock around your stand, you’ve lost touch with the culinary interests of the larger world out there. Fast food won’t nourish future generations.

Hi,

I found an interesting paper on Dashboard Design from SAP published on April, 11th 2011. Perhaps a better base for further discussions than this one slide, which is really annoying.

You can download it here: http://bit.ly/fhCmKB

Perhaps not the end of the road, but a step in the right direction, I think!

the best, Lars

Lars,

If it’s true that plagiarism is the sincerest form of flattery, I suppose I should be thrilled by the new publication “SAP User Interface Guidelines for Dashboard Design” (author unknown), for most of its contents were lifted from my book Information Dashboard Design. I say “lifted” instead of “derived,” for this paper fails to acknowledge my work as a source, despite the fact that its first sentence is a direct quote from my book (without the required quotation marks). This is reminiscent of Microsoft’s attempt to patent sparklines without acknowledging Edward Tufte as the inventor. Should I be concerned that SAP, which recently incorporated my bullet graph into its dashboard software, will attempt to patent it as well?

I’m thrilled that SAP is finally embracing my ideas, but doing so without proper attribution is unprofessional, disrespectful, and an infringement of copyright law. I would request that SAP correct this by adding the appropriate citations to my work, but I don’t want my name associated with this publication. Here’s why: In many places where the author attempts to paraphrase my words rather than copying them exactly, and where the author departs from my ideas, this paper fails terribly, resulting in confusion.

The third paragraph of the paper’s three-paragraph introduction provides an early example: “A monitor is a detailed look at a specific area or even just one aspect in a specific area. A dashboard is not a monitor, although a monitor may also display graphical information but rather an overview on a whole area or even across several areas (and then at a fairly high level).” Say what? What could this possibly mean and how could it possibly matter?

The third section of the paper, “Layout Possibilities,” departs from my work into the realm of bad advice. It explains that a dashboard is made up of panels, each of which contains “visual information,” which can be laid out in any of the following ways:

You definitely don’t want to approach the arrangement of information on a dashboard by starting with a grid of panels. You must arrange information based on several factors, including relative importance, logical scanning sequence, relatedness, the nature of the data, and so on. You then use position, size, white space, alignment, and other attributes of layout to display the information in a meaningful, functional, and eye-pleasing manner. This cannot be achieved by beginning with a layout grid. Vendors encourage you to use layout grids for dashboard design because this is how their software works, not because it’s the best method. Not only does this notion of a rigid grid of panels conflict with the layout flexibility that’s needed, if you gaze at this collection of sample layouts you will experience a sickening visual effect. Notice the illusory circles that appear between the corners of the rectangles. The author explained dashboard layout using an awful layout of examples. Before writing about visual design, one should always take some time to learn about visual perception.

When the author attempted to rephrase my explanation of when bar charts should be used, he must have been indulging in strong drink. He begins by saying that bar graphs should be used “if single values are to be compared across static categories.” He substituted the term “static” for “discrete,” thereby rendering the recommendation incorrect. Another description of when bar graphs should be used is so off-base, I have no idea what the author is trying to say: “If values are considered to be longer or shorter.” So there you have it: whenever you consider the values that you’re displaying longer or shorter, use a bar graph.

In addition to changing the meaning of my words through sloppy rephrasing and including recommendations that are entirely off-base or unintelligible, the author also contradicts himself. Pie charts only work correctly when they’re used to display part-to-whole information (in other words, values that add up to 100%). Even when used for part-to-whole information, bar graphs still work better. The paper contains a section titled “Use Bar Charts Instead of Pie Charts” which explains the reasons why bar graphs work better than pie charts. However, later in the paper the author goes on to recommend a pie chart over a standard bar graph: “For comparing proportions, i.e. the relative distribution of a given amount over the categories, use a stacked bar chart or pie chart, if the fact that they add up to 100% or to another KPI is important.” Should bar graphs or pie charts be used to display part-to-whole data? Based on scientific research, the answer is almost always bar graphs, but if you only read this paper, you wouldn’t know, because the author has recommended both of them as the superior choice.

I could go on, but you get the point. SAP produced an official publication that attempts to recommend the dashboard design principles that I teach without acknowledging me as the source – bad form, but what makes it worse is the fact that they screwed it up. The author of this paper clearly doesn’t understand the material. This poignantly illustrates how little SAP Business Objects still understands about data visualization. They’re so far out in left field, they don’t even realize that this publication should be a source of utter embarrassment.

What’s most significant about this publication is something that I have not yet mentioned: it admits that SAP Business Objects’ dashboard product (formerly called “Xcelsius”) is in many respects poorly designed. Again and again this paper tells its readers to avoid aspects of the product that it predominantly features – the very aspects that they’ve promoted for years – including 3-D charts, circular gauges, bright colors, lots of filters, and multi-screen layouts. Rather than telling their customers to avoid these bad practices, which are fundamental to the product, why not remove them?

Hi John,

“Mistakes were made”?

Admitting a shortcoming need not involve shame. It is a healthy response to mistakes, which vendors would experience as relief, and those they serve would welcome as a sign of integrity. Progress begins with an honest assessment of oneself. This is true of organizations as well, though much harder to muster.

The author of that paper seems to be: @FrodoJansen
according to this web site: http://www.pieterhendrikx.com/sap-user-interface-guidelines-for-dashboard-design/