5 addictive things about Tableau 9.0 (fresh off the beta testing labs)

Preface: This profile is an add-on to my multi-part series on self-serve BI tools. Read my other posts here (https://www.dhirubhai.net/today/author/23200039) and here (www.ovum.com/authors/surya-mukherjee)

Hi, my name is Surya and if you’re reading this, you probably know that I am an addict.....of self-serve BI solutions. I follow many self-service BI vendors very closely in the course of my job and influence enterprises in their technology buying decisions. Fair disclosure - this might be one of the reasons that I am among the first to get a beta-license. However, no money has changed hands for this post – this is purely my unadulterated opinion.

Why am I writing this?

I get numerous queries from enterprise clients and vendors around applicability and comparisons between BI tools (self-service included). In most of these queries, I find a line of business person trying to understand how this technology is relevant to their careers, and how to make a good business case for it. This piece is meant for business users only and will hopefully start some of us on that journey.

So, what happened?

Tableau, a self-serve data exploration/visualization and business intelligence tool, launched version 9.0, which is in beta at this moment. Over the weekend, I signed on for the platform and tried my hand at data manipulation and analysis. This post is an attempt to summarize my experiences with Tableau desktop.

First impressions are last impressions! The first 5 minutes.

The first five minutes with any software tells you a lot about its makers.
When companies get too big, sometimes they forget that the user experience starts at landing page/download button. Don’t believe me? To test this hypothesis, try to download business software (or even white papers) from the websites of yesteryear incumbent mega-vendors and you will get your own version of the truth.

Some questions which haunt me when I look at new software are: How many MBs will I have to download, and will it be worth it? How much of my personal data will I have to divulge and how many surveys will I have to fill before I actually hit something useful? These are huge concerns with almost everything we download now-a-days. Seriously, even flashlight apps are way too greedy for information.

I could go on. Happily, downloading Tableau did not feel like an invasive medical procedure. No weird questions, no survey, nada. Also, in the size department, Tableau won handsomely. It felt classy and elegantly coded – how could I not like that? Telling a business user that your software has a million lines of code is probably the biggest mistake vendors can make in this day and age.

Tableau did not try to win a lines-of-code war with verbose coding. Thank you.

#1: The sheer intuitive brilliance

Among all the solutions I have seen and tested, Tableau is a hands-down winner for how simple and intuitive its interface is – without sacrificing a single level of granularity. The idea that any dimension or measure can be used in multiple ways (as a filter, as a parameter, etc at the same time) is extremely powerful and continues to be the number one draw IMO. In experience terms, Tableau feels like the love child of Android and iOS – it is a giant leap forward from a dumbphone OS, can be used by a novice, but at the same time allows a power user to go look at data in almost every imaginable fashion. You can make one Tableau chart and refine the bejesus out of it on your data set without needing a PhD in data science. You may actually never need to code a single line. In terms of new features, LOD is my favourite in V9.0.

Intuition is hardest to explain, but for a hint of what I am getting take a look at my pillars to intuitive interfaces. IMO Tableau 9.0 checks most of these boxes.

#2: The new and improved data management capabilities

The world has really evolved since the early 90s, but one thing that hasn’t changed is the 80-20 ratio. In my conversations with numerous business users and organizations, I find that still, 80% or more time is spent cleaning and wrangling the data. Data cleansing is hard, messy, and ad-hoc for the most part. Also, data cleansing and blending requires one to think in an inverted fashion – meaning you need to know what questions you intend to ask to properly format data. This puts most people off the analytics adoption curve; it is just not how our minds are designed to think.

Thankfully, vendors have started realizing that the value in bringing self-service principles to data management. A lot of Silicon Valley dollars are now going to companies that can make the data massaging process as enjoyable as getting an actual massage (xplain.io, Trifacta, Collibra, et al).

Tableau is not the first company to pay some attention to this area, but I was happy to see profiling and basic data management automation tasks applied right at the point of data load. The software has added a Data Interpreter which basically understands your data – at the backend there are some pretty sophisticated algorithms doing matching, profiling, and so on. One of the frequent complaints with Tableau (and to be fair, many of its competitors) was that one had to transform the data a lot and many times to get it to an analysis ready format (like crosstab normalization) . The Data Interpreter sets you up nicely on that journey; making adding new data rows/splitting existing ones/pivoting data as easy as Excel. Nicely done.

Also loved the new data connectors - an enviable list compared to peers.

#3: The parallel everything – faster results for a easily bored audience

V9.0 introduces a lot of parallelism into Tableau. Parallelism is one of those things IT gets much more excited about than business at the onset. If you’re like me, you could probably stare at download speed graphs on your home internet connection for an entire afternoon and wonder if splitting devices on the 2.4GHz channel versus the 5GHz channel bumps up the numbers in any fashion. But make no mistake – business ultimately cares about results and the speed at which you get to them. If you just bought a nice laptop or machine with 4 cores and decent RAM, Tableau 9.0 will suddenly seem a lot faster to you than 8.0.

Parallel processing and querying is slowly becoming table-stakes for all analytics vendors, especially because NewSQL+NoSQL databases and analytics products are slowly combining forces to form homogeneous solutions (and not disparate products). Tableau is no longer the weakest (read slowest) link in the analytics journey. If your database can run queries in parallel, Tableau can support it. Tableau can also aggregate in parallel using all the juice that your laptop’s 4 or more cores can provide it. Also, Tableau V9.0 has invested in more intelligent caching, which allows it to intelligently tier data for getting you the best throughput.

In business user terms, V9.0 will help you get the maximum performance you can out of your existing hardware and applications. Since Moore’s law still holds, your software will automatically benefit from incremental improvements in any part of your analytics environment. Also, do not be surprised if you find that the software is a lot faster in doing/loading things you look at regularly – it is indeed trying to, somewhat heuristically, understand your usage patterns.

#4: The speed of thought analytics and some prioritization

We’re all used to Google predicting our queries as we are typing them. Sometimes that leads to hilarious bordering on creepy suggestions. For a map of “Why is <country> so X” where Google fills up the X, take a look at this map.

But I digress.

Predicting your queries is good, especially if you are trying to help people who don’t know how to code, code. Tableau has added this feature to its interfaces now along with some new search features. It also has cleaned up and reordered a lot of things in the analytics pane, so that users can add or remove calculations/metrics at run-time. Some analytic measures that were a bit hidden have also come to the forefront.

#5: Tableau public, Tabelau help, Tableau community

Last, but perhaps most important, is the sheer size of Tableau’s dedicated fan base. Tableau’s user base in many ways mirrors popular open-source communities, but without their hang-ups . The line you’re least likely to hear at a Tableau group is “that’s a stupid question, you noob”.

Tableau users are genuinely underserved business process owners who live in the p2p world – you help me, I help you – and we pass it forward. It is truly bewildering to see a whole new category of users help each other out by sharing knowledge, although no one is being paid to do anything. I am not aware of many other companies that get paying users to practically run half of their break-fix department. But then, such is the cult of Tableau. One quick way at seeing how popular a BI vendor is to go to LinkedIn Groups. Here’s what I found out:

Lastly, I absolutely love Tableau Public. It is the perfect bridge between the consumer and business worlds and showcases how well the company truly believes in data activism and journalism.

#6: Now for the brickbats

As far as first impressions go, this is what I have. There are improvement areas as well. Things I would like to see Tableau focus on are:

More data management/wrangling functionality

Tableau must invest more in developing data management tools. Apart from providing a visual approach to analytics, the key to unlocking self-service BI is to first help end users access a range of clean and consistent data sources; and, second, integrate them painlessly. This, of course, has profound implications for what, how, when, and where data is sourced. Self-service BI implementations that focus heavily on solving the complexity of data access and integration while incorporating governance and compliance are therefore more likely to gain sustainable enterprise adoption. Visual data profiling is starting to become a must-have. Competitors such as IBM Watson and Oracle’s soon to be released offering may prove to be serious challengers to Tableau in the future. There is no shortcut to data quality – the vendor has to enforce a basic data management workflow even for desktop ad-hoc users.

A less watertight distinction between worksheet, dashboard, and story

The market has already discarded the notion of disparate reporting and analytics functions. I believe that the concept of worksheets/spreadsheets and dashboards are ‘so web 1.0’. The modern business user likely needs one workspace and doesn't care about what you call it. Also, I’d like to see a more intuitive way to link charts (this does exist but takes some reading up to achieve) that use the same data so that one click on a chart changes everything on the other. Qlik Sense appears to do this quite well.

The continued addition of more and sexier graphics

Self-explanatory, extremely important.

An ability to hum the IT tune

As Tableau grows and grows, it will need to work with the very people it was designed to evade. Sure, IT imposes red-tape on many things, but they are also your last line of defence in a world which is increasingly throwing up #sonyhacks. In the future, Tableau will have to work better with IT to ensure that it gains their approval at the time the business case is being presented. It is critical to remember that without IT involvement, self-service cannot transcend becoming a siloed desktop analysis tool. On the other hand, it is also equally important for IT to realize that shadow IT systems exist and flourish in even the most locked-down of organizations.

These are still early days to call out a winner in the self-service analytics market. As of now, the user is king, the market is awash with choices, and it is all warm and sunny. Fellow addicts, enjoy the happy hours while they last. Cheers!