TIBCO Spotfire Promotes an Insidious Myth
The number of viable visual data exploration and analysis tools can be counted on the fingers of one hand. TIBCO Spotfire is among them. The merits of this product are undermined, however, by the irresponsible ways that TIBCO is currently promoting it. A new marketing campaign by TIBCO illustrates what happens when marketing professionals who either don’t understand analytics or care little for the truth are allowed free rein.
Here are a few lines from TIBCO Spotfire’s new “Finally…Answers Made Easy†campaign (emphasis mine), supposedly written by the company’s CTO, Matt Quinn:
At TIBCO, we believe just visualizing data isn’t enough. Embedded deep in the brains of data scientists lies a knowledge set that can truly benefit any one of us who has ever struggled with the dilemma of which graph to choose for a given data set. How many times have you highlighted a data set in Excel, selected Insert Chart and ended up with nonsense? You try a different chart, play with the axes, change the numerous options – before you know it, you’ve wasted an hour and haven’t made any progress. You certainly haven’t gotten anywhere near insight or understanding. Imagine if your software knew what you needed to see, even if you didn’t?
We have mined the data in the data scientists’ brains and shared what they know about visualizations: all the arcane rules about using measures on density plots, when to use aggregations and how to use time series correctly. Spotfire will automatically examine your data and recommend the best visualizations for it. Allowing our algorithm to choose the correct visualization will let you focus on what you know best – your business.
When you took your driving test, they didn’t ask you to explain the principles of the internal combustion engine – you just trust it works. Whereas your grandparents may have been a dab hand with a spanner and an oil can, life has moved on. So it will be for the future of analytics – it will work smarter, so you don’t have to.
Similar to Tableau, Spotfire attempts to determine an appropriate chart based on the data that you’ve selected. This is a useful time saver when it’s done well, but it can’t peek into your mind to determine what you want to see, so its guesses are frequently wrong. This feature can also serve as a useful guide for data analysis novices, but in this potential also lies the problem: you can’t let software do your thinking for you. The big lie that’s being told here appears in the last few words: “It will work smarter, so you won’t have to.†This is not only a lie—it’s a dangerous lie that keeps organizations trapped in ignorance, wasting their time, unable to tap into the value of their data.
Well-designed software can indeed help you “work smarter,†but not “so you won’t have to†work smart yourself. Data exploration and analysis software, no matter how good it is, cannot provide a workaround for your lack of analytical skill. Software vendors hurt you and ultimately hurt themselves when they claim that their products can be used effectively without the requisite analytical skills. They hurt themselves because, when customers learn that they were sold a lie and can’t actually use the software effectively, they become disgruntled and eventually move on to another product. Sadly, they rarely make a better choice the next time around, and the doomed process begins anew. No one wants to believe that a product that they spent a great deal of money to buy won’t solve their problems.
This marketing lie is in line with the “self-service BI†lie that’s been told for ages. The notion that BI software can auto-magically enable people without analytical skills to make sense of data is ludicrous, yet it’s an appealing lie. We want something for nothing, but the world doesn’t work this way. Analytical tools can’t help us do better and faster what we don’t already know how to do ourselves. It can only augment our intelligence—extend our reach and help us work around limitations—never replace our need for intelligence and skill.
TIBCO is certainly not alone in its willingness to spread misinformation in its attempts to sell its products. Every one of the viable visual data exploration and analysis software vendors have played fast and loose with the truth and mislead potential buyers to varying degrees. Most of the wannabe (i.e., not viable) vendors in the space are even worse.
I suspect that the first vendor in the analytics space that’s willing to tell the truth about its product and what’s required to use it will eventually lead the market, assuming that its product is good, even though they’ll lose many sales in the process. A vendor could differentiate itself from the pack by being truthful. The people who spend their days trying to make sense of data tend to respect truth. They’d find it refreshing to witness honesty coming from a software vendor. This vendor could honestly say, “Here’s the good news. The skills needed to analyze data can be learned by any reasonably intelligent person, given the right resources and enough practice.†This is indeed good news, but it’s not as sexy as the claim that a software product can replace the need for skill.
Damn, damn, damn…getting value from data requires skill and effort. After all of these years of trying and failing to get value from data without paying our dues, why are we still so willing to believe otherwise? There are no shortcuts to enlightenment.
Take care,
14 Comments on “TIBCO Spotfire Promotes an Insidious Myth”
People who spend time actually making sense of data value honesty in a vendor, and often possess (or are open to learn) the analytical skills to use the product well. On the other hand, people who decide which BI product their company will use (which are frequently not the same people) may be a different story.
The vendors know exactly to whom they’re selling, I fear.
Steve,
I completely agree with you. Unfortunately, it’s human nature to want something for nothing or for minimal effort. Think of how many Ab-Masters or Thigh-Masters, or fill-in-the-blank Masters have been sold over the years. The commercials show a pair of super fit athletes using the device. What the manufacturer encloses with the device is the “Oh, yeah. Here’s the hard part. The diet that you need to follow to lose the weight so you can actually see you abs.” So it is with “VI Masters.” On the other hand, it is worth remembering that those of us who practice in this field have a responsibility to share and teach what we know about data visualization and analytics. It’s easy to forget that what becomes second nature for an analyst may be a completely new experience for the BI self-service recruit.
Hey Stephen, Brad Hopper here. I’m a Spotfire/TIBCO veteran and analytics practitioner. Naturally I’m also a long time follower of yours and proponent of the best practices you promote for data analysis. Now I can add to that my appreciation of your views on florid marketing language and quote attribution!
Hyperbole notwithstanding, it is a serious and studied goal of TIBCO Spotfire to make analysis easier and faster by, paraphrasing you above, augmenting, not replacing intelligence. Spotfire 7’s Recommendations feature is pretty powerful and different from what others are doing – I wanted to do some “truth telling” here and, if you are interested, invite you to see for yourself if Spotfire Recommendations presents solid options in-line with what you yourself might suggest.
We agree that while professionals may be skilled in their business they don’t necessarily know best practices for data representation and analysis. You’ve spent a career helping to promote these practices, what can Spotfire contribute? Well, after a user selects attributes to investigate, the Recommendations feature shows a gallery of fully rendered plots (not icons) from which one or more can be chosen to be part of an analysis. It’s the user’s job to decide, by looking at real data, what’s interesting and what’s not. All of the plots are not guaranteed to contain insights, but they will be appropriate to the data and metadata of the attributes chosen, and it will be faster to choose a ready configured a plot than to make one from scratch.
It’s rare that a feature is useful both to users with a little and a lot of experience. For casual users, Spotfire Recommendations reduces the mechanical requirements of configuring plots and instead focuses attention on evaluating properly presented data. For power users who would use these best practices anyway, the number of clicks to get to a desired outcome can be dramatically reduced. One is tempted to call it a short cut, but there’s nothing skimpy about Spotfire Recommendations because the data is shown from many angles at once, some that a seasoned analyst might not have initially considered or might have built then discarded. Recommendations is actually one of my favorite capabilities to come along in quite a while, we would love for you to check it out.
I certainly appreciate Brad’s response and Stephen’s original post. I remain open minded about the ability of tools to intelligently assist people wanting to analyze their data. From my own experience, however, I have trouble believing that these fully rendered graphs can achieve that. I constantly discover new insights by looking at data in different ways than my initial attempt. I even gain insights when adjusting axes so that the visualization enlightens. Sometimes, you need to zoom in; other times you need to take a more distant view. Many times, it is only in the relationship to other data that the best visualization reveals itself.
So, I guess I am skeptical. Not that it can ever be done, but that we have reached a point where this kind of “assistance” is productive. It is the very act of interacting with my data when I am constructing visualizations from which I learn the most. Anything that deprives me of that learning does not sound like assistance to me.
It was just yesterday at work that I suggested something along these lines. I thought it would be helpful to create a general list of things to make sure you try for different types of data for some of the new analysts.
I completely agree that knowing what to look for or merely having curiosity are necessary traits of good analysts. However, something needs to help jog ideas for novices at times. For that reason, I do not mind the suggestions given by the software. That said, they do drive you down one particular path and many users may never consider the other options available (which often would have been superior).
I would prefer a list of options based on the data source. Each option could lay out the visual and explain what each (one by one) is best for displaying to users (i.e. showing a line graph and stating that the data appears to present a trend…and why a line is best for this). This would not only complete the grunt work of making whichever choice is selected, but educate the user as well.
Good day Steve, long time. Without a doubt, tools help, but the fact remains that—a fool with a tool is still a fool.
The suggestions from tools like, Tableau, Excel 2013, and SpotFire help. It will still take a skillful designer to come up with a useful visualization.
Steve,
Thank you for the article. I am very new to data analysis and searching for insights. I was interested in this section:
“Here’s the good news. The skills needed to analyze data can be learned by any reasonably intelligent person, given the right resources and enough practice.”
Would you please consider writing an article with recommendations on which resources to use and what to practice to improve for people new to making sense of data?
Mark,
You probably won’t be surprised to hear me suggest that you begin by reading my books. Specifically for data analysis skills, I recommend “Now You See It” and my new book “Signal,” which will be available in about a month.
Thanks Steve. I’ll get started with “Now You See It” and I am looking forward to getting “Signal”.
Mark,
Also being new to data visualization, I took a continuing education class on data vizualization theory at a local university. Thankfully, its textbook was Steve’s “Now You See It”. It’s my to-go reference. I choose to remain data viz tool agnostic and marginally I am attentive to their viz choice suggestions. Instead, I rely on the common sense techniques in Steve’s book that align with the tenets of basic human visualization. I’m dismayed at the “dumbing down” of these tools. I’ve worked in IT supporting information workers for decades. No amount of dumbing down will foster greater acceptance. People must start first with the basic aptitude for this kind of analysis and then study and improve their skills like any other craftsperson.
I’ve read all your books and I strugle daily to follow your principles! Thanks for all that sharing!
On the current topic, I’ve been waging this war against the hype too. I’ve even blogged about the so-called “Data Discovery” ruinous marketing effort done by some vendors to dismiss all the hard work with “self-service BI” and the hype like that. Reading this post was exhilarating and liberating! Thank you very much!!!
Steven,
Ran into this recently with that had a similar topic. Alteryx has a big headline that says “Data Scientist Not Required”
http://www.alteryx.com/press-releases/data-scientists-not-required-new-alteryx-release-puts-predictive-and-customer
I dislike the headlines saying something is not required. They are always required, but perhaps software makes the task easier.
The purchaser of the product is often an executive that buys based on feature lists. SAP & Oracle have made huge strides in the ERP world by selling lists of features, and delivering absolutely awful software that “sort of” works once it’s been customized enough.
The BI vendors are starting to take a page out of the ERP vendors’ book, sadly. Why deliver a tool that needs people to be trained on it? Just throw together a few magical algorithms and now “you too!!” can do big data analysis on your paltry 10 gig invoicing system database. Oh, and guess what, you’re now a “data scientist”…
Laziness, like sex, sells – a real shame, but it taps into human nature. I’m frequently faced by clients requesting:
* Insight from data which doesn’t relate to the question(s) they are asking
* Precise predictions from noisy, unreliable data
* The famous “fix my business” button
It’s the duty of every professional analyst to present a polite and friendly “No” to such requests, together with a concise and coherent explanation of why “No” is the right answer. Data visualisation offers an extremely effective means of explaining what the data can tell us, as well as the limits of those data and why they can’t be twisted and stretched past their natural elasticity. Data Science should – by definition – embrace scientific method; forming a falsifiable hypothesis, collecting evidence, analysing evidence in the light of the hypothesis, drawing conclusions and developing new hypotheses… and round we go again. If it doesn’t then it is – at best – ‘informed guesswork’, and – at worst – ill-informed opinion.
I don’t claim to live by this creed – I fail to live up to these standards consistently – but it’s something I try to drum into my colleagues and my customers; we can achieve impressive results, in remarkably little time, but there are real and fixed limits to everything we do… and every analytical task requires detailed understanding, careful thought, rigorous testing and extremely careful communication.