Quantcast

The future of BI in two words

What’s the future of BI? Last fall, one sharp source of mine answered, “Two words: Tableau and QlikView. You didn’t hear it here.”

Those are startling words coming from that source, a well-regarded BI consultant known for big-name clients and their big deployments.

At about the same time, a column of mine appeared in Information Management titled “Don’t call it BI” — in which I mentioned Tableau and a few smaller tools. A reader emailed, “You should also become familiar with QlikView.”

My many Tableau-using friends say QlikView is hardly worth a look. Poor visualization! Control panels! Scripting! “It’s so — yesterday,” one emails.

It’s “yesterday” to some yet it’s the future to others. It’s time for a look.

Both Tableau and QlikView promise the same magic: Listen to one pitch and you might think that you’re listening to the other. Each sets itself up against traditional, big-iron BI. Each claims to empower business users by giving them all the data and control they need for free discovery. Each is easy to use. Go inside each tent, though, and you see how different they are.

Metaphorically speaking, Tableau is West Coast. It’s built for discovery by the individual. Just show up and ride on the breeze, the demos seem to say, free as a seed fairy on a meadow. The inevitable mistakes of discovery are quickly undone and forgotten. Create the most dazzling visualizations — “vizzes” — thanks to built-in best practices that nudge you toward beauty and punch.

One of the most attractive aspects is users’ effervescence. They seem to be riding on the wind and solving business problems all at once. Their rapture sweeps me away every time I’m near it.

If Tableau is West Coast, QlikView is East Coast. Its community is bigger, the third-party add-ons are more plentiful, support seems more available, and overall workflow feels more structured. It too is built for discovery, but it’s discovery rooted in community. The “associative experience” reveals relevant data, and you can create your own views and in quick succession ask any questions, anticipated or not. But unless you’re working alone, someone else probably defined the data and its structure for you. This is QlikView’s counterpart to Tableau’s meadow, though it’s more like a manicured garden than Tableau’s unfenced field of daisies.

QlikView’s boundaries may be more apparent than Tableau’s, but I suspect that there’s at least as much power there. I just haven’t yet been able to judge it for myself well enough.

The trouble for me is that I’ve used it alone, as if stuck in a remote cabin. Though even Thoreau might have liked the “associative experience,” QlikView really comes alive only when you link to others.

As in Tableau, any QlikView user can create or modify a workspace, a document linked to one or more sets of data with any number of displays. Unlike Tableau, QlikView isn’t so finicky about data; for one thing, linking to Excel spreadsheets is easier.

I can’t speak with assurance just yet on the differences between QlikView and Tableau Server — more on that later — though I think I see a QlikView edge there.

One other advantage for QlikView is clear: built-in collaboration. True, Tableau workbooks can be passed around in a variety of ways forever. But as with our atomized life on the West Coast, such a community would be for me, the hypothetical manager of a group, too loose for comfort.

Tableau users will shudder, as if about to be extradited back to Maine. “Great, central authority all over again,” they would say. Yet when I imagine myself managing a group, I would feel disabled without a tight, integrated social structure.

“It’s the soft stuff that matters,” TechTarget research director Wayne Eckerson likes to say. Such stuff is what interests me more than anything: Who are these people and how did they choose what they did?

Have most Qlik or Tableau users chosen their tool the way most of us choose spouses, religion, and politics — guided by our relationships? How many software shoppers qualified their candidates with lists of requirements and features and followed through based on evidence? Did they do what a veteran sales person at a large BI vendor sees?: “They gather requirements, they issue RFPs, they visit trade shows, they talk to vendors, and ultimately they pick one because they like its color.”

I think it’s usually about “color,” color being the cover story for something most people can’t quite describe. For now, though, I’m happy to say that at least my first question has been answered: Yes, QlikView belonged on that list in “Don’t call it BI.”

35 Responses to The future of BI in two words

  1. Ted,

    It is probably also worth mentioning that Tableau is consciously about what they call the “cycle of visual analysis.” Discovery happens along the way, questions get answered before you finish asking them.

    It really feels uncomfortable working with Tableau to design reporting “systems,” since that is not its intent at all. No pixel perfect greenbar-to-web browser mentality there. Using Tableau to discover the story in any data is a playful and rapid and in-the-workflow experience. Is it like that for Qlikview; I don’t know, having never worked with it.

    Your comment about boundaries struck me as being less about east vs west as about traditional thinking about left vs right vs whole brain thinking. At first glance, one might think Qlikview – more left brain because of the process of getting to discovery; Tableau more right brain. However, in my experience, simple as it is, I find Tableau responds to and encourages whole brain thinking.

    MANY BLESSINGS!
    Peace and All Good!
    Michael

  2. Hi Ted,

    From my perspective, your timing with this post is perfect as just yesterday I read of the partnering announcement between Datawatch and QlikTech (http://www.marketwatch.com/story/datawatch-announces-integration-of-monarch-report-analytics-with-qlikview-2012-04-23). I find this tremendously exciting as it brings QlikView users the ability to capture unstructured data from report files and many other sources using Monarch technology.

    Of course, that it further validates what I’ve been writing about for years doesn’t hurt either! ;)

  3. Ted,

    Good summary of the BI Tools you outlined above. However, you did leave out 1 very important BI Tool (although I guess it is more of a robust BI solution) , Spotfire. Spotfire takes the strengths of both of these tools and takes Business Intelligence analytics to the next level via Enterprise-class deployment of dimension-free data exploration (without the reliance on IT). Along with the fastest to actionable insight and built in Predictive Analytic capabilities that neither of the tools you reviewed offer.

    We have done a review on Tableau Software at:
    http://bitools.org/tableau-software/
    and will be doing a review of QlikView shortly.

    Looking forward to reading more of your posts.

  4. I think you’ve hit on something interesting about the difference between the two approaches, but I think your line of demarcation is a bit off.

    As someone who is in a highly data-driven profession (University Enrollment Management), and a die-hard Windows User and Apple Antagonist, the implication that Tableau is a west-coast, sandals-and-lovebeads approach while Qlik-view is buttoned down Oxford cloth is off. I’m all East coast and suit and tie on almost everything, and I’m a big fan of Tableau. Let me tell you why.

    I remember both “first time” experiences with Q and T; on the latter, I had insight in five minutes; on the former, I was still staring at a screen after ten, trying to figure out how to get it to work. Perhaps this is a function of my intellectual capacity; I’ll grant that.

    But Q seems to me to be more focused on report creators: Those people who take orders and produce something for others. In fact, I wonder why Q brings in the entire data set…when most reports use only a few variables. So my struggle might be a residue of the product focus.

    T on the other hand, is about exploring data: When you go in, you’re not sure what it is exactly you’re looking for, but discovery is easy: A function of the interface, not scripts and difficult instructions. I see using T more as a function of becoming disciplined about possibilities, not about narrowly tailoring the insight you know you want.

    They’re different, and they serve different people with different needs, but not different cultures.

  5. Having just visited Gartner and reviewing solutions for clients right now, your blog and readers comments are very timely.
    Qlikview has a whole other power and value prop that frankly Qlikview/Qliktech doesn’t quite understand – it goes back to their “simplicity” roots – and most consultants, IT and BI groups miss Q’s strength – navigating data then collaborating and iterating with non-tech, non-BI, non-analyst folk –

    I have tried to do the same thing with Tableau explicitly and a number of other tools – because Q is too expensive for a number of reasons and I reasoned that SQL 2012 and other tools would have caught QlikView by now – but alas –

    Qlikview is an excellent discovery tool and proto-typing tool for a large company’s BI platform of record – it is also a great tool for data discovery and QA –

    If you asked the question, “Which tool will lead to a company finding the most innovative solution?” Qlikview would win – but there would be an asterisk because the innovation would come from a group of decision makers sharing Qlikview insights and iteratively defining and testing an outcome.

    Data structure is too big a deal in every other solution – Once a BI vendor says “hierarchy” the discussion is over for discovery

    Qlikview is not designed for reports and does not perform modeling, predictive analytics or data discovery (clustering et al) but if you have artifacts, derived data, spreadsheets and other data just lying around (Qlikview is forgiving of sparse data) than Qlikview can amaze.

    Just my thoughts – Michael R Hoffman, author Customer Worthy; Director Paragon Solutions, NJ

  6. Ted – Visokio Omniscope is is another global player in this space with Java-based architecture subsequent to, and different from Tableau, QlikView and Spotfire. Omniscope is frequently selected in preference to all three, and users of the others have been known to switch to Omniscope. Omniscope is broader that Tableau, with full drag-and-drop end-user ETL. Omniscope is not as consultant-dependent as QlikView. The average cost per seat of Omniscope-based solutions falls continuously as the number of ‘interactors’ grows, including unlimited remote users on Apple IOS and Android deveices that do not support full Java, as Window tablets do. Omniscope excels on both the hard ‘data techie’ stuff, but also the all-important end-user ‘soft stuff’ too. For example, it works offline with no or poor mobile connectivity, and also refreshes fully… connection permitting. Free to try:
    http://www.visokio.com/omniscope

  7. The trend I see are companies using all of the these tools along with the “big iron” BI, not in place of. For data discovery, associative methodologies, and visualizations, they are pretty slick and have nice interfaces. But against a large datawarehouse, and for enterprise reporting, scorecards, and dashboards, the big vendors still have many advantages and uses and aren’t going away anytime soon.

  8. Talking about this is a little bit like evangelizing religion and trying to convert a Muslim to Christianity or vice versa. It usually doesn’t fly. Although – there do seem to be more converts to Tableau’s religion than the other way(s) these days.

  9. I came to QV from Excel. I am not a script kiddy by any stretch of the imagination, and yet I created better sales reporting than my client (ASX200 company) had ever had before within a day from a standing start. It was so good, that it actually scared quite a few people who loved the greater accountability from as many different viewpoints as possible, but hated the extra accountability from as many different viewpoints as possible. I liked it so much that I joined a QV partner and in my time there, sold QV against Tableau quite successfully i.e. I never lost a sale to Tableau. Given that our sales process was based on a proof of concept process, and was always going up against other proofs of concept, I can only assume that either we were significantlybetter at doing a POC than our competitors, or QV was actually superior in terms of its ability to quickly deliver real results/discoveries/reports/analytics than whatever we were going up against…and then there is the absolutely huge online community that QV has in comparison to everyone else…

  10. I have tested T & Q and am impressed with both. I believe that the discovery idea works fine in both; Tableau needs clean data and Qlik can fix alot itself making it easier to work with non structured data. Making more impresive POC where it can tweak problems with source data directly. Imressive but a POC is a POC and I advise everyone to delete the POC when finished, use it as a prototype and move all business logic and washing/tweaking to the source database.

    I am afraid that organizations will look them into this script ETL and build solutions and business logic that will cost heaps to replace when they want to change vendor in 5-20 years.

    Tableau is great with maps and diagrams which are easy to create, Qlik is more of properties…

    Another vendor, isn’t PowerPivot a competitor?

  11. I’m an avid Tableau user but have lately come across Qlikview for an internal dashboard project for digital marketing. In this specific case I must say Qlikview has a huge advantage over Tableau: the API connectors. It’s apparently very easy to pull data from APIs (such as Google Analytics, Facebook Insights, Adwords…) right into Qlikview and refresh the data automatically. I’m amazed Tableau hasn’t considered this approach yet.

  12. Daniel – Omniscope is often preferred by major advertising agencies over both T & Q and has a wider range of digital marketing direct API Connectors than Q, including many international ad-serving and social media monitoring platforms. More Connectors are being added all the time.
    Free to try: http://www.visokio.com/download

  13. Is it fair to say that Tableau is useful for those who prefer to do all of the data prep and aggregation in the layers before the BI / presentation, and QV has advantages for those who need to work manipulate data in the presentation?

  14. Regarding Daniel’s comment – it is indeed very easy to pull data into QlikView from Facebook, Twitter, Google Analytics/Calendar/Docs/Spreadsheet, YouTube, Sentiment/Text Analytics APIs, Klout and many many other web APIs using the QVSource add-on (www.QVSource.com).

    As far as I am aware there are no other BI tools with such a broad range of connectivity options.

  15. I’ve been looking at all of these tools. In many ways, Omniscope fits the bill for a small business needing flexibility better than the others. However, deep pockets are needed. If the pricing were comparable to T and Q, I’d definitely go for it over the others. I’d go as far as saying it’s the cleverest software I’ve ever used, even compared to advanced engineering and mathematical software.

  16. Ted…..three years ago we met at the Seattle user conference and I mentioned to you that you will witness a cottage industries emerge from companies like Tableau and Qlik-Tek. You and I are now part of that growth – My company provides consulting and training and database optimization services.

    I will admit to being partial to T but I respect what the other players like Q have done and achieved. When it comes to grading these companies with all their features and functions, I firmly believe that one area is rarely addressed….what is it like to do business with each company? There is as much to say about this area as there is about an API or a web client and so on.

    In in the next 90 days a Fortune 1000 company will pick a vendor from the list of the top 3 DV vendors in the world…..Tableau, Tibco Spotfire and Qlik-Tech. That company retained a 3rd party to grade all three based on array of questions that address the technical to the business side. The outcome will tell us a lot about the real world.
    Safe travels

  17. Hi Ted,
    I’m not sure whether collaborative dashboarding app Jolicharts.com belongs to the “Don’t call it BI list” or not, but your perspective on this service would be of great value to us – if you manage to find some time to give it a go some day.
    Btw, I agree with one of the above comments that says that Qlikview is a great prototyping tool. And I don’t know Tableau enough to actually disagree with you, but you’re clearly convincing me to give it a try.
    Best,
    Jeremy

  18. A couple of comments about how QlikView fits in with traditional BI vendors. In regards to reporting, QlikView customer usually find that an interactive report in QlikView replaces several dozen static reports. Aside from the huge time and cost savings of 1 report vs. dozens, a QlikView user has the advantage of seeing something interesting in the report and being able to instantly and visually explore that data in another tab within the same QlikView “app”. Traditional BI just doesn’t lend itself to that usage model.

    I love the quality of Tableau vizzes and how easy it is to create them once you have the data. The challenge is that Tableau, which doesn’t have a noteworthy built-in ETL, relies on clean data from a single primary source, which is a challenge in many cases. If you need to create a data model which combines, say, an Oracle database, a SQL Server database, and a Excel spreadsheet simultaneously, you’re out of luck. Same goes with common business sources such as Salesforce.com or SAP. When you think about time to value, it’s a good idea to consider end-to-end from the original data source to the final business insight. Great visuals are one piece of the chain.

  19. Great article. Comparing Tableau and Qlikview is quite tough as it depends on user needs. I have used both products and like both products. Quite a while ago now I wrote my layman’s-speak write-up comparing the products, http://reports4u.co.uk/tableau-vs-qlikview/ but since then the connectivity options with QV have increased.

    In my line of work, where I operate to very tight deadlines and have to produce reports for different clients using different data sources very quickly, Tableau is a far superior product as it’s much quicker than QV to get started. In a more structured environment, perhaps working in a big corporate BI department, QV could be better, as the reporting capability of QV slightly edges Tableau, but more planning and time is required to get there.

  20. Thanks for the article. My company recently made the decision to use Tableau 7 and Tableau Server. We have been very happy with the implementation. It is easy, nimble and quick.
    I would also like to respond to Terry Martin’s article about Spotfire and Tableau and point out that there are numerous incorrect statements about Tableau in the article.
    Having said that, the key to picking a BI solution is to understand the nature of your organziation, the tools you already have in place and the value of an additional BI tool. The real competition we are all up against is MS Excel!

  21. hi guys, as I currently work for a qlikview elite partner I am bias on the subject. It was only a year or so ago I started playing with both QV and tableau to see which was for me. Qlikview just felt like it was more complete as a full solution that an enterprise could buy and use without needing another BI system already there to feed it nice data.. which 9 times out of 10 doesn’t exist anyway. Tabloue did have the eye candy but I think that edge is being lost now due to the extension object ability being introduced which pretty much gives you the power of open source libraries like jquery and standard web design approaches to make beautiful dashboards. Tableua does have a Disney advantage though in that its very easy to use and can get up and go in minutes…nits just a shame it’s so hard to then share your insights afterwards which Is a QV strength.

    For now I’m still QV, more scope in the projects, more community, and while scripting adds complexity it also makes it a lot more powerful.

  22. Thanks Ted. Explored and played around with Q earlier. Didnt feel it too easy with large enterprise loosely structured data. Most time spent around massaging the data to make Q answer a specific question, and if you change the question ugh. So, now, started looking at T, will see how it goes.

The data industry thrives on conversation. Please submit a comment.