Most users love Excel, non-users hate it. When it comes to data visualization, Excel is generally dispised, except by those that have to make dozens of charts every single day. I call this the Excel Stockolm Syndrome. These are the forsaken data visualization users that keep making 3D pies when they should know better by now. Tired and overwhelmed. Not in the mood the learn yet another tool just to make those elusive “effective charts”. If you link good visualization to a tool they have no access to, you can be sure that the whole message is lost.
Becoming a Data Visualization Anarchist
I think things can be changed from the inside, improving the way people use Excel. I write for Excel users because I’m one of them. That’s not going to change soon. But I love data visualization, not the tools that make it happen. I specially like interaction, multiple charts and making them available on the web. And I need to manage more data (not big data, just more data). Some things can’t be done in Excel or require too much effort.
The Portuguese poet Fernando Pessoa has a funny short story called The Anarchist Banker. The idea is that only a banker can be a true anarchist, because only a banker can be free from “social fictions”, specially money… In data visualization, this means getting the tools out of the way, by learning them or avoiding them.
I chose the learning path and I’m learning R now. I always wanted to make those scatterplot matrices. And I want to play with the ggplot2 package. A programming language is covered but R is not going to pay the bills.
Enter Tableau and Qlikview
I will not complicate matters by discussing how I chose Tableau and Qlikview and not Spotfire, for example. I just want to choose one. Qlikview vs Tableau. Comparison articles like this and this are very helpful, but a man is a man with his circumstances. Each starting point is different from everyone else. Let me tell you what I think I know about these tools in this early stage.
Tableau
I like Tableau, I have to admit it. I like the fact that you don’t have to fight stupid defaults in design and formatting, because I share the same data visualization principles.
I like its enthusiastic and knowledgeable community. Let me give you two simple examples. I spent a lot of time making this horizon chart in Excel, and Joe Mako quickly came up with a better version in Tableau.
Then I tried to be creative with the bamboo charts and Joe Mako strikes again, with a better implementation. I’m starting to get nervous… (kidding)
I know and respect many Tableau users, not because of Tableau but because I share the same views regarding data visualization.
Apparently, maps in Tableau are good enough, so that’s a good point.
And as a blogger, I want to make my work available to the web, and Tableau Public is a nice option (my population pyramid).
The Guardian often publishes Tableau visualizations. I’d like to try that with the local newspapers here.
On the other hand, we know how stubborn some datavis experts are. Is Tableau that stubborn? Can clear vision and the right principles become a straitjacket? I really hate straitjackets (“the idea of”, never actually tried one…)
In my country, Tableau is virtually unknown and I am not sure if I want to sell shoes in Africa.
Qlikview
I know even less about Qlikview. The first chart I see in its video is the pie chart above. Not exactly a shiny example of good data visualization principles.
Apparently there is a very active Qlikview community on Linkedin but not so much on Twitter. Probably this is meaningful.
I keep reading that Qlikview is better than Tableau when it comes to making dashboards, while you should explore the data with Tableau. It’s a good point in favor of Qlikview (that’s what I need now). Extensions and the market seem to be interesting too.
Qlikview has several business partners here. Actually, I was invited to work in some Qlikview projects in 2013 (obviously I have to learn the basics until then). They can pay a lot of bills.
Its not always about features
Not everything is black and white, not everything can be decided based on feature-by-feature comparison. Not everything is heart, not everything is reason. If I choose Tableau, my data visualization skills will improve a lot. Qlikview is harder to predict. I’m sure there are many users that dislike the pie above. If not, Qlikview can be more, hummm, challenging.
What I’m going to do
I mentioned those Qlikview projects, but I’ll try to remove them from the equation, at least for now.
I have a simple dashboard in Excel and I’d like to create Tableau and Qlikview versions. That’s probably one best ways to evaluate a tool, using my own work.
So, can you help me?
I’d love to learn from you. Can you answer questions like:
- How do they compare regarding maps? Is it simple to add your own maps?
- Is it true that it’s easier to make a a centrally designed dashboard in Qlikview, while Tableau has a more exploratory nature?
- How can I share a Qlikview chart in my blog?
As you know my job is to sell and help people with Tableau. So I am biased. But you also know I was a customer and user of Tableau for four years prior to joining them. I joined the company because it is the best piece of software I ever used. I too used to be an excel junkie and loved creating effective excel visualisations. I loved the fact that I had the knowledge to tinker and produce magic in excel.
But I look back on that time and wonder why I enjoyed wasting all that time. Saved time is the power, the essence, the secret of Tableau. Here are practical examples:
You can be connected to your data in a matter of seconds. You don’t write script. You point and click. And you are done.
You want to switch dimensions around in a chart? Drag. Drop. Done. No wizards, no property screens, no parameters to fix. Vizql, the underlying technology enables this instant visual exploration.
So tableau chose a bar chart and you wanted lines, or pies, or circles? Two clicks to switch the bar type.
You want to explore trends over time? Tableau handles dates automatically. You don’t need to script anything and you can expand, split, separate dates and times to your hearts content.
Geospatial content? Maps with a double click? Maps do not require an extension in Tableau. Or if you have a non standard geographic field, just import some custom geocode and you are good to go.
Tableau is founded on best practices of data visualisation. This is core to your blog. Tableau does not do gradient colour schemes or 3d charts or donuts because they are not effective. We have a research team who feed the DNA of tableau (jock mackinaw, pat hanrahan, robert kosara, etc). Also, consider how much tableau is investing in research and development. $200m in the next few years; I forget the exact % but it is well over 15% revenue.. Someone else can confirm qliks investment but I believe it is around 2%?
Can we do beautiful dashboards? I believe so, just check out the tableau public gallery.
Is it all perfect? Heck no. Check Christ Gerrards Tableau friction blog for some of the annoyances.
I am sure this will turn into a great debate….
Nearly any application can create a decent chart, all with varying degrees of effort. So in a lot of respects I believe it is a personal choice for what tools meshes with your perspective of data.
For myself, Tableau makes the most sense because it is built around a grammar for graphics, when you place pills on shelves in Tableau, you are writing code in the visual programming language VizQL. This is the number one reason why I choose to use Tableau every day. VizQL makes Tableau a joy to use. See http://vadl.cc.gatech.edu/documents/5_Hanrahan_CNVAC-PatHanrahan.pdf some background on VizQL.
I attempted to use Qlikview, but it is too dialog box heavy for me, I prefer Tableau approach that enables exploration through immediate results. This enables me to fail faster in Tableau, to try more things in less time, producing better results.
I do not see how data exploration limits dashboard capabilities, or that more capability in one means less in another, or maybe I misunderstand your question.
Tableau has made some UI choices with their quick filters, so it may be possible that Qlikview has different built in capabilities. Tableau’s taste in quick filter UI is something I have been putting up with, so there is room for improvement, but this does not outweigh the value Tableau provides, see http://37signals.com/svn/posts/3263-some-of-my-most-beloved-products-are-those
From what I can tell about Qlikview, one of their big features is filtering based on selection, this is something that takes more effort to reproduce in Tableau, but in most cases on a dashboard it is two clicks away to use a worksheet view as a filter. I do like Qlikview’s concept of white/gray/green things.
You mentioned an interest in mapping, and while Tableau is not GIS software, you can get custom map backgrounds and layers with a WMS server, draw multiple mark types on a map, and there are a few ways to get custom filed maps. What is an example of a map you would like to make?
Although very dependent on data structure, Tableau enables a chart to be produced nearly as fast as the question can be thought. This is not marketing speak, this is my regular life with Tableau, and it may be a function of the amount of experience I have with the tool.
A key consideration if you choose to go with Tableau is that is not a stand-alone full BI solution. You will want a data integration setup for performing ETL prior to Tableau, because with the ability to transform data prior to Tableau, what you can produce in Tableau is nearly limitless. There is no “straitjacket” in Tableau if you can transform your data; Tableau is only bounded by the data structure and contents you connect it to.
Granted there are many things that Tableau is not an ideal solution for, but if what you want can be drawn with dots, lines, and polygons, it can be made in Tableau. For the things Tableau is not good at, I would recommend supplementing with Processing and/or D3.js. What would you like to see made in Tableau?
I chose Tableau because it fits well with my workflow of rapid development with immediacy, and the approach of VizQL just makes sense to me.
In my experience, any tool and be used to create poor visualizations, just look at the history of http://www.tableausoftware.com/public/community/viz-of-the-day to see examples. It is more the about the user than the tool, that finding a tool that aligns well with your perspective and way of thinking about data is the most important.
You are always welcome to contact me if you have any other questions about Tableau.
Jorge, have you tested Excel 2013 yet? It comes with a plug-in called “Power View” built by the SQL Server Reporting Services team. It allows you to use maps, to have interactive charts, link them, etc. It is not out there yet (well, it is, but you need SharePoint and SQL Server 2012) but soon you’ll have the ability to do some pretty cool stuff straight in Excel. I’d say (but I’m biased) that Excel with PowerPivot and Power View trumps QlikView and Tableau in price, performance and ease of use. Have a look here: http://blogs.msdn.com/b/seanboon/archive/2012/07/30/visualizing-the-olympics-with-power-view-in-excel-2013-day-1.aspx for the first post in a series by Sean Boon on using Power View.
You will make brilliant, clever visualizations regardless of the tool Jorge. I follow your blog for inspiration, not just technique.
We tried both Tableau and Qlikview at work. Tableau won hands down due to the ease of use for every skill level. I found the multitude of dialogue boxes too similar to Excel or other developer tools. Not instant visualization – I need to SEE the effect while it’s happening – before Tableau I didn’t even know that I did, but now I do and now I have very little patience for anything less than instant. None of us got to the dashboarding stage, we were too put off.
I don’t know if you are able to embed a Qlikview dash in a blog. I’ve never seen one. Hmmmm… That’s another plus for Tableau, the ability to pull down other people’s work to look under the hood and see how it’s done.
The mapping capability at present is easy enough. It’s not GIS, but it does the trick and they are always making improvements. That’s another cool thing about Tableau, you can put forward recommendations for improvement and vote on them. They actually do make the changes.
As for the $$$ thing, when you become the Tableau Jedi in your country the work will follow. If you build it, they will come.
Nice post. I will keep I fast and easy….like Tableau. Tableau is the best new analysis tool invented since Lotus 123 hit the market in 1982. I literally have not seen another tool since then that was as significant….until Tableau. Excel has become a list-storing tool for me. I am a Tableau Consultant but like Andy I started as a customer. Qlikview seems like new-old BI to me.
How do they compare regarding maps? Is it simple to add your own maps?
Tableau is hands down the best when it comes to maps. BUT, I’m not usually a fan of map visualisations, I do feel that they are massively over used.
Is it true that it’s easier to make a a centrally designed dashboard in Qlikview, while Tableau has a more exploratory nature?
For me, it really depends on the audience and how engaging your readers are likely to be in the data. Tableau definitely lends itself more to exploratory, journalistic and one off analysis. I think that is reflected in the fact that going into many of the major financial services organisations in London, most use Qlikview, Business Objects and Cognos – which are essentially ‘old world BI’.
For exec dashboards, I still feel Excel does a better job in terms of flexibility – adding in glossaries, definitions, commentary etc is much more fluid in Excel than either of these tools, but of course much more error prone.
‘On demand’ BI in large organisations still has a lot to answer for, not in terms of producing pretty charts (thats the easy bit) – but actually creating a business culture where dashboards are used effectively and are engaging to end users.
How can I share a Qlikview chart in my blog?
Now, why would you want to do such a thing?
Have you considered Dundas Dashboard yet? They created the component technologies that are now found in Microsoft’s flagship BI technology. They can extract data from excel and have an open API for extracting data from anything else, plus they’re ahead of the curve on a lot of dashboard BI & display data in real-time. Oh, and the dashboards are fully customizable so you don’t have to settle for a shitty gradient.
They have some examples on their website:
http://www.dundas.com/dashboard/online-examples/
I’ve been using them with my company for 2 years since we got off Tableau and haven’t had a complaint.
Dundas Dashboards would be my recommendation too, just because they seem to accomplish everything you’re looking for in regards to embedding dashboards and creating custom maps. However, I’m not sure what you mean by “…it’s easier to make a a centrally designed dashboard in Qlikview, while Tableau has a more exploratory nature” since all of these software options touch on both aspects (Dundas included)
I’d always recommend Tableau first. There are many examples on my site of how versatile the software is. On a side note I would not recommend Excel with PowerPivot and Sharepoint as a method for visualising data as the lack of printing capabilities is a real drawback. I’m looking to use PowerPivot + Tableau Server for my company’s future dashboarding projects.
We went through a Tableau vs. Qlikview (and others) selection process last year, chose Tableau, and have been very happy with the results. Often I’m spending a huge amount of my day obtaining and wrangling the data and have limited time for real analysis. Tableau’s interface makes the analysis incredibly productive and satisfying. An advantage of Tableau’s UX is that it lets me fail faster. A lot of my work is about looking for patterns, Tableau enables me to check out dozens to hundreds of possibilities in very little time, views that would take a lot longer to set up in other tools.
Even though Tableau might not have some of the fine-grained controls over dashboard appearance that other tools like Excel do, it gets me to the my real goal of useful output faster. Tableau delights me in a way that no other piece of PC software ever has, and for me that’s a really great thing to be able to say about something I’m sitting in front of for hours at a time.
In terms of learning Tableau and making connections, Tableau the company publishes a huge amount of free training material, there’s Tableau Public, and an active Tableau community. That was an important factor for us since we’re relatively remote. The ability to download a Tableau Public visualization or a packaged workbook file (for Tableau Desktop users) is a massive advantage to learning, you can see something you like and take it apart within seconds and figure out how to do it yourself. Given the quality of your designs, I have no doubt you’d pick up Tableau very quickly and work would soon be coming to you.
@Jorge,
+1000 to everything Andy Cotgreave, Joe Mako, and Dan Murray already said! Beyond the utilitarian consideration of paying the bills, you should think about the elusive notions of deep satisfaction, and freedom, and protecting your brain. All these things are beyond bits and bytes and features. They are baked into the DNA of Tableau. Not so much any other tool. Period.
@Martyn,
Your point that, “Tableau definitely lends itself more to exploratory, journalistic and one off analysis. I think that is reflected in the fact that going into many of the major financial services organisations in London, most use Qlikview, Business Objects and Cognos – which are essentially ‘old world BI’,” confuses me. If you mean to say that executives and IT outfits in this vertical embrace old world BI at the expense of rapid, speed of thought insights, I will take you at your word, not being myself a participant. If you take a look at Tableau’s customer list on their web site, though, you will see a growing number of names in the banking and financial services sectors.
The tools you cite seem to be inherently designed for ‘reporting’, which in itself can be argued IS actually a one off exercise, if building the report ever gets done. The business model for these vendors seems to be built around something other than analytics per se, which an exploratory and recursive activity.
Tableau, is built around the mental processes inherent in what they call the Cycle of Visual Analysis, which is key to understanding the world. This approach is exploratory, as you say, but not one off by any means. And no ouchies for the brain!.
So, generally, Tableau gets out of the way and let’s you get your work done. It never surprises me that when I sit total newbies down with Tableau they are answering questions and synthesizing present moment conclusions and recommendations for our clients within minutes. Often what was intended to be an introductory training module ends up being something like, “Go away and leave me alone, I get it. I am having too much fun getting to the root causes of [insert issue here] and formulating recommendations from my data. I will reach out if I need you, but don’t count on it.” Gotta love it.
@Dan,
As you know, our experiences have been almost identical. You said it all, sir.
@Andy,
I, too, am a recovering Excel junkie.
Hi Excel junkie & anarchist,
A great post you wrote.
I don’t have much experience with Tableau, but Qlikview is definitely a good technology if you’re okay with the idea of an install on your hard drive.
Indeed, fat client Qlikview is a affordable (about EUR/USD/GBP 1000 one off + 20% maintains) and powerful tool.
However, it’s in my opinion rather complex to operate, ie build dashboards.
Plus, if you’re looking for an online business intelligence tool (SaaS) to foster collaboration with your peers / colleagues / customers / …, Qlikview becomes very expensive, too expensive.
I know you don’t want to expand the debate to other BI solutions, but since I’m an anarchist too, I’m not going to abide by your authoritarian stance that highlights market leaders only 😉
Combining flexibility, power and collaboration doesn’t have to cost what Qliktech asks for – at least as I see it.
First of all, let me introduce myself: I’m Jeremy Fain, one of the cofounders of environmental reporting SaaS vendor Verteego. We went through a typical make / buy or partner process, looking primarily at buying or partnering rather than buying of course. You’ll see that eventually, we chose to start a business unit, Jolicharts.com, to target Excel power users – like myself, and other ISVs (independent software vendors) – like Verteego.
On top of Qlikview (that we pushed at a number of customers of ours), we had delved into a number of other solutions including (not exhaustive):
– Pentaho / good solution / heavy to implement –> show stop / very 1990s-looking / not too expensive though
– JasperSoft / same as Pentaho, somewhat less heavy
– BIRT (Actuate) / good solution / too complex to implement –> show stop
– BIME (We Are Cloud) / the best solution we actually tried / still a little complex to implement + we experienced some performance issues (perphaps solved by now) –> show stop / very smart business model
– Zoho / excellent but didn’t match our needs
– Chart.io / good but not for Excel users
– GoodData
– Spotfire (Tibco)
…
None of these (including Jolicharts) have data compression algorithms as powerful as the ones provided within Qlikview.
However, Qlikview doesn’t foster collaboration, it’s expensive (you need an online solution to work with your group, and Qlikview server is costly), and you need a BI consultant to implement your dashboards.
This is why we decided on developing Jolicharts.com, as a spin off of Verteego because we believe many people are currently facing the same need as we did.
Like Qlikview, Jolicharts enables Excel users to connect their Excel files in a wink, and start building their dashboards immediatly (no BI or computer skills needed).
Unlike Qlikview, Jolicharts enables its community to share (for free) their data visualization dashboards to collaborate. Here are examples of public dashboard: https://jolicharts.com/dashboard/dashboard/publish/de1dc5f3e70d70d794249b431f4ba49f
You can also share privately, just with your colleagues.
Jolicharts is also dirt cheap, and even potentially forever free (if you invite folks). We’re still alpha, and you’re obviously very welcome to give us a try.
Looking forward to your feedback, and thanks again for the fish,
Jeremy
Tableau is awesome for analyzing data and very, very good for creating content.
For disseminating content, be prepared for a lot of frustration no matter the software. No vendor seems to have a good off-the-shelf toolkit for this yet.
Hi,
There seems to be a lot of Tableau fanboys here which is making the conversation quite biased. I use both Tableau and Qlikview. Those who say that their having a hard time building dashboards in Qlikview, well I think they haven’t done it. I manage now a global dashboard of my company that consolidates 32 kpis from 7 major divisions and I tell you its all about good design. Many qlikview users are pisses with the menu interface, I admit I am one and the ugly default objects. The secret is making your objects transparent so they can have a clean effect. Send me a message and I will show you a good execution of a Qlikview dashboard. Now tableau for me is for quick visualization of data but haven’t used it to combine different data sets. I think it the object designs in tableau are much better but it usually requires that your data is in perfect format. In qlikview you can do a lot more at a script level like adding inline values. I think both are very useful you just have to tap the strengths of each.
If there’s any learning curve whatsoever in using Tableau or Qlickview, my thoughts are to stick with R and consider D3js for dashboards.
It’s true that both require a little extra time to learn but the payoff is huge. Both will give you full control over the look and feel of your graphics along with full control over the user interactions.
Thanks so much for your great comments. I would like to see a more balanced discussion, but you scared all those Qlikview users away 🙂 My fault, probably. If you are a Qlikview fan please add a note below, I need your perspective too!
Let me take advantage of (temporary, I hope) blissful ignorance and play the Devil’s advocate here.
1) In data visualization, there is an ugly side (data cleansing, ETL) and a glamorous side (visualization). Apparently, Tableau focus on the glamorous side only, while Qlikview offers both ETL and visualization tools (yes, they don’t look that good).
That’s enough to heat up the discussion about the role, limits and skills of a data visualization expert. How much overlap there is between visualization and ETL? Should we improve both skill sets or focus on one of them?
ETL is a thankless job and I’m glad Joe Mako brought it up. The way I see it now it that if you choose Tableau, not much will change in your ETL processes. Tableau just seats there, showing off its glorious charts. If you get Qlikview instead, you will be able to replicate some of current ETL processes. And if you are using Excel and Access for ETL (I use them a lot), it should be much faster (that’s what I hope). If you see ETL and visualization as two completely different processes requiring different skill sets maybe having Tableau and a dedicated ETL tool is a good idea.
2) If I’m a great salesperson, that’s because clients love and trust me, not because I spend all day exploring sales data. I just want a sheet of paper with basic results, competition activity and outliers that need to be explained and addressed. You, the data analyst, can do it for me and for everyone else in the team. It’s your job. You are good at it, so please let me good at mine and don’t waste my time with data analysis.
Not everyone wants to be empowered. I should cook, but I have no talent or interest in cooking. If I have to cook, I just use the kitchen robot. Most people need data visualization like they need a foreign language: being able to communicate with the locals and avoid basic mistakes. They are not and don’t want to become experts. Take-away: know thy users.
3) Qlikview is formal, control-freak, data-centric, Linkedin. Tableau is informal, democratic/chaotic, vision-centric, Twitter. As a blogger, I love Tableau. As a consultant, I know that different clients have different needs, and they probably are more interested in a formal and self-contained (ETL) approach like Qlikview. Not all of them, fortunately.
@Andy: I know how enthusiastic you are about Tableau, and rightly so. I believe you spend a lot of time in data cleansing. How do you see it? Do you see it as an integral part of the data visualization work or should be… humm, delegated?
@Joe I like the idea of failing faster… A recommendation system would present the buyer with “Tableau buyers also bought [enter ETL tool here]”. What would you write there? Also, do you share Andy’s view regarding scripting? Now that I’m learning R, I don’t find scripts that bad…
@Boyan I love the way I can use slicers to synchronize tables. I’m sure there is a lot more to learn, but it’s time to visit new lands.
@Kelly Thanks, kind words.
@Martyn I once wrote that thematic maps are the pie charts of cartography. I suspect you agree with that.
@Marty, Slava, Dundas is on the wrong side of contemporary data visualization principles. Changing defaults is one thing, but fighting against its nature is well beyond what I need now.
@Jeremy Thanks for all the info. I’ll take a look at Jolicharts. I’d like select the tool(s) I’m comfortable with and can answer the needs of a broad range of clients. Obviously I will need to worry about money, but not yet.
@RKW Yes, I’m worried about that too.
@Noah I’m not sure if I want to learn two programing languages. I had to choose one and I chose R.
[Update] By the way, how do they compare regarding real time monitoring?
@Jorge
I’m stealing that quote on cartography! Totally agree. There are SOME cases where they are truly useful, but I do feel that many who use them do so because they think they look cool rather than that they’re necessarily the best way of representing the data.
I am still not settled on an ETL tool that I feel meets my needs. I am not content or satisfied with any of the ETL tools that I have seen (maybe that means I should scratch my own itch and make one that fits my needs).
Currently my general process is to build a proof of concept in Lyza, exploring the business logic, and then have developers code the logic in a language like C. I am on the lookout for a better ETL solution.
I choose Lyza because it only has 4 elemental steps, instead of hundreds of single use nodes. Lyza does not attempt to do everything, only basics. As with Tableau, Lyza enables me to fail faster with a supporting interface and immediate results. Another big plus is that I do not have to worry about the arrangement of my steps on a canvas, and Lyza also automatically takes care of the other busy work, like meta data.
There are a great deal of downsides to Lyza, it crashes frequently, drops code, uses 10s of GB of hard disk space daily, no null values, can get confused when reordering steps, and other strangeness that I have to work around every time I use it.
But I am willing to put up with all this for the way it enables me to work, it works with data the way that I like to think about data, and that is the same reason I choose Tableau, it matches my perspective.
In regards to working with real time data, Tableau is not well suited for this task, Tableau is not designed to deal with steaming data, Tableau issues queries and then renders the results.
For real time data visualization I would recommend Processing or Cubism.js and for ETL of real time data you would need a CEP (Complex Event Processor) like Streambase.
As for scripting, that is a need Tableau has not directly addressed, beyond the scripting abilities with Tableau Server. I believe for Tableau it is a benefit that you do not need scripting or code to accomplish a great deal of very interesting analysis, but it does create a barrier when the desired operation cannot be achieved within the user interface. Scripting can be a great benefit for removing the limitations of an application.
Also, I used the comment from on your website, but did not get a reply, in this post there is a link to https://www.excelcharts.com/dashboard/ as an example you would like to recreate, but there is noting on that page, maybe only members can see the content. I would like to see the Excel file and attempt to recreate it in Tableau for you.
@Joe, thanks for your excellent comments. I’ve been learning a lot from you this year…
You can see the dashboard in this video (it’s an older version, but you’ll get the idea) and an image of a second version here.
I’m currently trying to replicate its basic functionality in QlikView. Obviously I can’t avoid an Excel frame of mind (that’s why I chose Qlikview to start this), but it’s interesting to fight that. I don’t know anything about QlikView, so it’s challenging too.
I’ll do the same with Tableau next week(s?). I will gladly share the file and the data with you, but let me try it out first. This is how I like to learn. I’m sure you can come up with an elegant version in a couple of hours, but that would change the game…
@Joe in the dashboard there are several charts displaying the data for a single year. But there are also line chars that display the entire range. These charts must be “half-linked” meaning the year is not linked but they must display the selected countries without duplicate filters (selecting countries in two lists).
Took me some time to find a solution and it was fun. But I’m getting a little irritated about some design choices (surprise, surprise!) like a shadow under symbols in the legend or the chart that keeps re-sizing depending on legend width.
Having a color legend that is a multi-select as well, is not possible in Tableau, they would have to be separate objects on the dashboard. Everything else in the video looks fairly straightforward in Tableau (parameters, actions, only relevant values quick filter, etc).
As for the version 2 image, the UI in the bottom right corner would have to be different, a legend cannot be directly placed on top of a chart area (can use an annotation to get the effect), and the spark bars/line/text in the bottom right might take a little effort to get exact in Tableau.
Clearly, Tableau is not designed for endless customization of the interface, combined with the way I use Tableau, this is a good thing, it removes the paradox of choice.
The really interesting stuff in Tableau for me is with their table calculations.
Hi Joe,
Juste curious about your ETL search : have you tried Pentaho Data Integration (aka Kettle) ? I’m not affiliated with them in any way, but I have found working with it to be very rewarding. The community version is free and works well for me.
Edouard,
Yes I used Kettle for a some time, and I like their approach as a Data Engine, and there are a lot of great capabilities build in, but its focus does not align with my needs, because for nearly everything that I need in ETL, I would have had to write custom code in Java to accomplish. Additionally, for the built-in steps, there was little consistency, and I found most of them difficult to understand. There are many other details that do not fit my high expectations either. I ended up spending more time dealing with the application barriers than working with my data. I have recommended it others that need a full feature ETL solution, but it is not good fit for my needs at this time.
@jorge this comment thread is fascinating. i like hearing your progress. You asked in an earlier comment what I think of the scripting/ETL part of the process. It depends on the level of transformation. For simple metadata/cleaning up, Tableau’s fast – rename columns, create groups and aliases, etc. That’s done in an instant at any part of the analytical process. I don’t have to decide at the start. When it gets more complex, I do the ETL elsewhere. Tableau has not yet begun to address ETL – right now, we stick to visualising your data. Your better off using a specific tool for ETL and a specific tool for visualisation (like Tableau!). As soon as you lock yourself into one tool for ETL and data exploration/visualisation, you better hope that product is good enough for both parts of the process, because if it isn’t, you’re locked in.
@joe and jorge
I’d love to get a copy of the data for the sample dashboard. Joe said you can’t do multi-select filters on a colour legend. That’s only mostly true. You can do multi-select for highlighting from a colour legend. If you want to do multi-select filtering from a colour legend, you can do this by creating a worksheet that looks and behaves like a colour legend. Once you have that worksheet, you can use Dashboard Filter actions to do multiselect from it.
Good luck and keep us posted on the progress!
Andy
Andy, sorry I was not clear, I was referring to the the interaction in the video Jorge linked to around the 0:30 mark. It is legend that is also a multi-select check box and each item is a drop-down combo box. If you can remake that interaction in Tableau as a single object on a dashboard, I would love to see it 🙂 The point I was making was that Tableau is not endlessly customization in the way Excel is with the use of VBA.
@joe – Ah, i see. My fault. Not sure I could hack that together in Tableau! Not in v7, at any rate…
Here is the my comparison of Tableau vs. Qlikview:
http://apandre.wordpress.com/2012/04/24/tableau-vs-qlikview/
Joe Mako, you mentioned above, “From what I can tell about Qlikview, one of their big features is filtering based on selection, this is something that takes more effort to reproduce in Tableau, but in most cases on a dashboard it is two clicks away to use a worksheet view as a filter.”
Do you know how to show (in Tableau) a listbox and show the frequencies (like I can in QlikView) for a subset of data filtered on 3 or 4 filters on the page?
I used to work for QlikTech and I currently develop Qlikview aps to help me get quick answers to business questions around sales and marketing (I combine salesforce.com data with existing excel spreadsheets). I love Qlikview because I don’t mind writing some easy SQL code in the script editor to enable me to do some pretty powerful analysis and drill downs. ROI by lead source and lead category and country are just some of the quick analysis that I do. I am more of a numbers person, so attractive presentations of data are of secondary importance. That said, I have been hearing great things about Tableau. I am going to check if they have some getting started videos on youtube so I can see how easy it is to use. Perhaps someone can reply to a few questions I have. Does Tableau require a data warehouse, or can it access and combine data from different sources (spreadsheets, salesforce, Quickbooks, ERP’s and other databases) without the need of a data warehouse? Does Tableau perform its calculations in random access memory like QlikView, and if not, how does it do it?
Hi Mark
Tableau has lots and lots of video tutorials available, all free. There’s a bunch of on-demand training videos: http://www.tableausoftware.com/learn/training?qt-training_tabs=1#qt-training_tabs
And our YouTube channel has lots of related content:
http://www.youtube.com/user/tableausoftware
Tableau has very little in the way of requirements for data. We can connect live to just about any major datasource or import data into our own fast data engine. We want the user to be able to connect in the way that is convenient to them, not dictate one particular method. Here’s the run down on all our data connections:
http://www.tableausoftware.com/solutions/data-sources
Calculations are, where possible, passed on to the datasource to be calculated, as databases are generally optimised better than Tableau could ever be. Some calculations are calculated in Tableau itself, though. There’s no simple answer to that question! Tableau does not explode and swallow up all available RAM – it is architecture aware and in the cases where data/calculations require massive amounts of processing, Tableau will slow down rather than come to a stop.
It’s never simple to figure out which “tool” best fits. As an intelligent consultant told me one time “it depends”.
I believe it boils down to a few critical details….
1. If you’re looking for a “tool” that can address traditional query calls from a clean set of data or datawarehouse then Tableau may be the right fit as a visualization tool.
2. If you are you committed to leveraging a “platform” to help you drive your data journey with or without clean data QlikView may be the right fit.
QlikView’s full potential is realized when instituting a strategy for all information workers to drive their own business discovery. The bonus is being able to leverage the entire QlikView platform for data governance as well.
Don’t get me wrong, Data Marts & Warehouses are important but wouldn’t it be interesting to create associations and explore your data while making decisions before waiting on a 18-30 month Datawarehouse project.
Recently I read a blog stating that each QlikView user is like the Christopher Columbus of their data. Columbus was looking for the Far East and bumped into a place called America. The point is that true data analysis allows every user to explore and allow their own navigation, eliminating assumptions and producing successful results.
http://community.qlikview.com/blogs/theqlikviewblog/2013/01/10/with-qlikview-you-re-the-christopher-columbus-of-your-data
Enough said on this. Let’s all prove it to our customers in the field which QlikView has done for 20 years. 26,000+ customers in over 100 countries….
Well said chris sault. Being QV fan i cant agree to everyone. Blog is about QV vs tableau. I could see many are marketing Excel 2013 and many of their tools. If you say QV is not so good, then can anyone tell me why tableau has not entered leader quadrant of gartner while qv is there for last 4-5 yrs.
http://bi-review.blogspot.in/2012/11/impressions-from-tableau-in-comparison.html
see above link. It is worth discussion.
Still no decision??? 🙂
Kelly, I’m writing my own data visualization (e)book, and my priorities changes a bit, and I keep hearing so many good things about Tableau 8 that I decided to wait for it. Meanwhile, I’m still playing with Qlikview.
As a QlikView consultant, the real question here is what do your clients want?
I came to QV from Excel because I could download a fully working personal edition for nothing. That’s right, free. Here’s what I love about QV:
1. It has its own very powerful ETL engine that allows you to
a) connect to multiple data sources within a single application. One client connects to 20 different customer facing systems in one application to get a complete understanding of what drives customer satisfaction.
b) transform, fill in the gaps, or otherwise augment the data as it is being loaded into the application. One client uses this to create monthly tracking docs that join together information in a totally useful way that you would never put into a data warehouse.
c) create highly optimized and compressed data stores that can be used by multiple different QV applications – almost like a mini DW.
2. QV uses a columnar database structure as its data back end. You don’t see it, it does not affect what you do at all, and yet it leads to compression rates of at least ten times, which allows transaction level data to be loaded without pre-aggregation. That means no building data cubes, and a much lighter hardware footprint because your I/O doesn’t get hammered anywhere near as much as the competitors.
3. QV’s pioneering work in “in-memory” processing combined with its columnar data structures means that if you want a faster analysis (although the “instant on” nature of it is pretty fast anyway), just add RAM. One client was looking to implement a DW/BI solution using Microsoft stack tools eg. MS-SQL. Fully half of their budget was for hardware – storage, I/O etc. Even after paying for QV licences, the savings they would have realized on the hardware alone would have put them in front. But the savings on the consulting would have been astronomical, because dashboards and reporting solutions could have been developed in 2 weeks, instead of 12 months.
4. If you can write formulas in Excel, you can create charts in QlikView.
I have heard it said that Tableau is sexier to look at, but that it needs to be coupled with a serious ETL engine if you want to connect to multiple data sources in one application. QlikView is both an ETL engine, and a Visualisation engine, and in fact, because of this, QV is often used as a prototyping tool to scope out requirements for DW and other “older BI” solutions.
Finally, a little known fact about QlikView is its write back capability. It can actually write data back into the underlying data sources if you want to. Imagine what you could do with that kind of iterative data exploration capability…
This is an excellent discussion. I am relatively new to Tableau (less than 12 months) but have 20 years experience using BI tools such as Cognos and SAS. I was getting a bit jaded, to be honest, particularly with how complex/over-engineered so-called ‘Enterprise BI tools’ such as Cognos 10 have become. Tableau reminds me of the excitement we felt when we first discovered Cognos PowerPlay/Transformer in the mid-90s and started building OLAP cubes graphically. The fun is back! For the past couple of months I’ve been beta testing Tableau 8 and OMG it blows Tableau 7 away. I can’t wait for Tableau Public to be upgraded to version 8.
Regarding ETL, or let’s call it ‘analytical data preparation’, I know Tableau are pushing their improved ‘data blending’ capabilities, but for me, when I get into Tableau my mindset is creativity not data preparation… I like to prepare the data prior to import into Tableau, and my tool of choice is currently SAS Enterprise Guide. (Yes, ‘poor man’s ETL’ I know, but it gets the job done and it’s easy to use). I’m currently writing an article on SAS/Tableau integration techniques, which I will be posting to my blog shortly. I have already documented a way to prepare SAS dates for easy consumption in Tableau, which is an area several customers have asked me about: See http://www.see-change.co/setting-up-sas-dates-for-input-into-tableau-desktop. Other ‘Agile DW’ tools/approaches I’ve heard about but not used include WhereScape RED.
Addressing the Q: “Does Tableau perform its calculations in random access memory like QlikView?” my understanding is yes – at least this is the option I choose – but it can also pass the processing to the DB if required eg Teradata (which QlikView cannot, as I understand it).
Something not to underestimated about Tableau is the power of it’s calculation engine. We were recently working on some fairly complex time-series based calculations using Base SAS and SAS/ETS, which resulted in quite a lot of SAS code. We were able to produce the same logic in a Tableau calculation in about 3 lines of code.. Similarly, a specific chart calculation in Cognos 10 required 3 Studios and over 60 mouse-clicks to produce. In Tableau, it was about 9 mouse-clicks to produce the same thing 🙂 So don’t dismiss Tableau as being ‘light-weight BI’, because it is more powerful than you might assume.
One thing I love about Tableau is that when you have a question on a technique to produce a certain visualization, there is usually someone in the Tableau community who will answer it right away, and even provide a Tableau Workbook with the solution. As good as the tool is, I think the helpful user community makes all the difference.
Hello Jorge,
I’d be curious to learn how this evaluation turned out for you. I also did light proof of concepts with both Tableau and QlikView and recognize strengths and weaknesses in both products. Here are some bullets from my experiences.
Tableau Strengths
Easy import of tabular/formatted data
Ease of creating and manipulating visualizations
“Calculated Fields” available for data manipulations
Very modern training, support and user community (posts from Joe and Andy and direct feedback from Tableau users were very helpful in my learning process)
Tableau Weaknesses
Limited ETL capabilities (this was a big one for me, very complex calculations and disparate data sources)
Limited scripting and functions available for data transformations (kind of related to the first point, still dealing with data complexities)
QlikView Strengths
Easy import of data
Nice visualization and association of analyzed data
Scripting option for data transformations, which I want to explore more.
Very hands-on support from reps and tech support folks (very popular with business folks).
QlikView Weaknesses
Dialog box driven (per Joe’s comments) to get to the visualizations
Legacy-feel of training videos and online support options (per your comment about their video graphics)
My attention was diverted to another internal project, but I’m fairly certain our firm will eventually require a BI/Visualization tool and Tableau and QlikView are my top two choices right now. Any follow up from you on this post would be appreciated.
Regards,
Al D.
I am in the final stages of writing a book, and Tableau 8 will be available very soon. I expect to write a follow in the near future, maybe in April.
Thanks Jorge. I’m actually signed up for a Tableau 8 tour event in Chicago. I’m trying to keep tabs on the progress of both firms.
Thanks for your blog.
Hi mark, did you ever get the answers you were looking for?
We are currently looking at qlikview v tableau here so we have similar questions…
I would give SiSense Prism a chance, enables individuals who are not professional developers to create and share interactive business reports and dashboards, which can then be viewed, customized, extended and drilled into by colleagues, executives, customers, suppliers and/or partners.
ElastiCube (SiSense’s technology) is disk-stored, which means the limit on its size is only tied to the size of available disk-storage, not RAM. Similar to other in-memory technologies, ElastiCube utilizes in-memory query processing for speed of calculation. But in contrast to these technologies, ElastiCube’s query processing engine (Elastic IQ) only loads and unloads data to and from RAM on-demand. This means RAM is only taken up by data you need access to right now, but the rest of the ElastiCube data is stored on disk, available instantly when requested.
I have no too much expirience for giving my own recommendation , but I have friends who advised me to use Tableau instead of Qlikview, the reason? as you mentioned the community behind Tableau is a key factor, today’s communities are good if you want to get the best of a tool.
I heard of Qlikview is better in some situations, but if you get the right help on Tableau thnx to the comm. so it’s not necesarry other.
I hope I’ve could helped you.
Erwin
Hi there,
I used to work out of a QlikView consulting shop prior to landing my current role as Risk Manager in Singapore. Did some prototyping with Tableau and generally I have a positive impression of Tableau as a company (even though I blew an interview for some techie role. no regrets though.)
Some perspectives.
If you have a well mannered data warehouse and want to get basic answers quickly, I would say definitely go with Tableau.
If you have data lying all over the place and need to get some answer to a question you had not previously thought about, then use QlikView.
And if you need to do some analysis and bet your house on it, use QlikView to pull the data together, and build a quant model in R. And always use VaR.