OpenI is an open source Business Intelligence application for on-demand or SaaS deployments. Based on J2EE, OpenI is an out-of-box solution to easily visualize data from OLAP and relational databases, where users intuitively build and publish interactive reports, analyses, and dashboards.
First off — Hans Rosling is an inspiration to us all in the business of analytics and data visualization. Not only this story is extremely relevant, but the way he shows the numbers — there is a lot to learn. I will make an attempt here to deconstruct his latest TED talk in terms of what a good BI tool show do, and also how this is a great use case of how great BI users behave:
BI Features Used by Hans Rosling:
The most prominent is the use of Time as a special type of “dimension”. The tool knows that Time will support the concept of a Play button. This is still very novel — most BI tool, OpenI included, treat Time as any other dimension — you can drill up, drill down, set date filters, or date ranges — but that’s about it. Taking a lesson from here, what we should do instead is that the moment there is a Time dimension, user should have the option to “superimpose” Time in “Play” mode within a given analysis — this should result in a Video Player like slider widget appear at the bottom of the analysis with a big old Play/Pause button next to it
Notice how he first presents the data bubbles in dual-axis graph and then transitions it over to a map view. This makes the concept of “background canvas” a dynamic entity for presenting data. How many other choices a user can have (in addition to dual-axis and map overlay) to use as the context in which the data should be presented
He keeps only 1 attribute per axis – country in X-axis, and % of population with HIV on Y-axis, and everything else (gender, per capital income, etc.) is treated as a filter (in OLAP speak). This keeps the visual very clear on its message. I have often struggled with OLAP based analyses, which have multiple dimensions on each axis, which makes sense sometimes in the table view, but the chart-view is completely horrid. Single data attribute per axis is a way to address that
When it comes to drilling further into data, he basically clicks on a country bubble — and it can either split by income groups, or only the specific country goes on a time play motion while others stay the some, etc. — the key for me here is that drilling down is best done at the visual level — somewhere on the chart/graph itself the user should be able to isolate a data group (in this case a country bubble), and have a choice on drilling down or move it back and forth in time
Hans Rosling as a BI User/Presenter
Emotion, emotion, emotion… he is so far away from the stereotype of a statistician making a presentation. He cares about what he’s presenting. The numbers are real people — they get sick, and they can either get better or they can die.. you can feel that empathy as he presents.
Al Gore did this first (that I can recall) in The Inconvenient Truth when he brought a crane ladder to hoist him up so he can point to the tallest bar in the chart that he is showing. Maybe a bit too melodramatic — but it drives the point, and also makes a more visceral connection with the data. Hans Rosling stands on top of a table at the beginning of the presentation to explain the different numbers he is presenting, and the audience is at once connected and engaged
His bringing of the long metal pole to point to the numbers instead of your generic laser pointer (“I have solidified the laser beam”) is another way to get more personal and physical to show how involved he is
Ultimately he has leverages the BI tool to make a presentation, to tell a compelling story. Earlier in my career, we worked on a feature with another BI tool that automatically generated powerpoints from its charts. Yes, it was pretty crude, and didn’t really work that well usabilitywise — but the point is, this was definitely a feature aimed at helping users build a story off the various charts and grahps and analyses. People want to tell a story — the BI tool should help them do that.
Ultimately, watching Mr. Rosling is definitely inspirational — I can only hope that OpenI will one day does the things he’s shown us in this presentation. I’m sure we will get there in due time, but it is the spirit in which BI tools are used, and their ultimate message.. that’s the important thing to keep in mind as we move the product forward.
I am very happy to announce that today we released the beta version of OpenI 2.0. I want to thank our entire development team for all the hard work they have put in for this release.
Please download it from sourceforge.net and try it out. We look forward to hearing your feedback on improving this release as we work on further testing of this version to get to general release.
Create New Project: Create new project option is added in the preferences menu that creates a new project by selecting a new template which can be either the duplicate of the current project or a new template. It also allows the users to define the category, project logo etc. for the project being created.
Multiple Dashboards: an enhancement in the dashboard user interface which allows the user to create multiple project dashboards. this feature is available for the admin user only.
Purge Files: this feature allows the admin user to purge the older files, the user is prompted the file type ( i.e file extension ) and the date , the files with that type and older than the given date will be purged
New Features from OpenI 2.0 Alpha:
Completely re-written based on Spring, Java Server Faces and AJAX
New enhanced UI layout, completely “ajaxified”
New Drag and Drop Navigator UI to move attributes across rows, columns, and filters
Enhanced repository system implemented. Now project contents can be deployed anywhere. No need to have project contents within the “webapps” folder of j2ee servers
Application can be deployed into any J2EE server
QA feature added to validate the MDX statements for all analyses within a project with a single command
Manage Feed feature enhancement – now contains split rows feature and better CSV file parser
Default analysis for project – you can now specify default analysis for a project which is the first analysis displayed to a user immediately after user login
Customizable chart series color (for non pie type chart only) – Now you can define custom color palettes for chart
Excel export enhancement – OpenI now generates excel binary file which embeds chart in file itself (previously it contained a web link to the chart, which could result in broken image links)
Dashboard UI enhancement – Better UI that enables displaying both table and chart view option
Better File Manager – shows file in explorer style tree
User/Role management (requires custom build from source) : now users/roles management can be done from OpenI
Plus there are gobs of bug fixes. Check out the release notes for more details.
Please pass the word around. And as always, we look forward to hearing from you.
A product’s progress is measured not by features, but by user experience.
You can add a lot of value to your product by removing features.
Spending 6 to 12 months with a 4-5 person team to develop the first release of your product is way too long. You can’t wait that long to get market feedback.
These were just a few golden nuggets of wisdom in Dave McClure’s presentation yesterday at SD Forum’s BI SIG meeting. At one point, Dave said — “I come from an engineering background, I used to write code.. and I can’t believe that I’m saying this to you — but all the architecture, algorithms and engineering behind your product don’t matter as much as how strong your marketing is, which starts with user experience.. we all tend to develop a lot of cool features which the user never ends up discovering” — reminded me of that tree that fell in the middle of a big forest (did it really fall?)
Much of Dave’s talk may seem like web analytics 101. He’s got a (cleverly named) “Startup Metrics for Pirates” called AARRR (stands for Acquisition, Activation, Retention, Referral, and Revenue) to measure your overall success. Then as a startup, you focus on improving these metrics one step at a time (here is his detailed presentation):
Make a Good Product: Activation and Retention
Market the Product: Acquisition and Referral
Make Money: Revenue and Profitability
Personally, I get the step 2 and 3, but it is the first step where a great many of us falter because it is not so easy as it sounds.
Part of making a great product is to have inspiration, vision, and an intense passion to solve a particular problem — and then you hope that there are millions of others out in the market who see the problem with the same high priority as you do — and then also hope that your solution beats the competition by being in a completely different league, either by being the first of its kind, or by introducing a completely superior technology, er, user experience.
So can you do all that in 3 months or less? Seems like you have to, if you want to survive in the web 2.0 world (or whichever version we are on these days)
This clearly does not apply to everything. It only applies to a certain class of products/services where customers can experience the
product on the web, and where the quality of user experience can
be measured (even if it is a model of lead generation online and fulfillment being offline)
So if you are building a new product/service today — can you fit it into this model? Well, one way to try it out is to strip away all the features until you are left with just one or two that are your core differentiators, and see if you can create a version of it that can be experienced completely online. But the point is, if you can do this, you have a framework to measure success a lot earlier in the game. One key edge that startups generally have over bigger organizations is the speed of their product development iteration. How soon can you release a feature to your customers, get feedback and go through a series of subsequent iterations and releases?
Dave’s “Pirate Metrics” are about tracking the effectiveness of these iterations, but first you have to configure your business practice so this model can fit. Not a bad thing if you can.
Fellow blogger Bhupendra Khanal has an interesting post that mentions the challenges associated with BI 2.0/Information 2.0 (he also plugs our very own OpenI, which is much appreciated — Bhupendra, may OpenI karma come back to you thousand-fold Software industry, not unlike any other, contains a lot of hype and probably sometimes even more so with all this 2.0 buzz, which probably seems cool to the industry insiders, but is definitely confusing to the market.
Take BI 2.0 (or Information 2.0) for example – what in the world does it mean? Well, turns out, at the end of the day, to most BI vendors, it means more fancy charts and graphs and dashboards, except this time they’ll have rounded corners, larger fonts with brighter colors, and maybe a fit of Flash and/or Ajax thrown in for a good measure to demonstrate live interactivity.
All this is fine and well, but all this is also pure BS if you are not helping your user make better decisions, or informing them of something new.
If BI 2.0 or Information 2.0 is to be seen as the “next generation” (it seems you cant’ escape these cliches), then it needs to go beyond charts/graphs/dashboard paradigm. BI applications and tools need to be rooted in the knowledge worker’s workflow – and should be cognizant of the types of decisions that need supporting.
BI needs to be aware of the domain context – i.e. which industry are you supporting? which area – marketing, finance, operations, research..? Because without this, the best BI can do is to provide nice visuals and hope and pray that the user knows how to translates them into intelligence and action.
But software can be better than that if is stops being lazy. And that’s my hope with our work in OpenI. We certainly started in the charts/graphs/dashboards paradigm, so we are as guilty as anybody. But as they say in any 12-step plan, “acceptance” is the first step — and now, we are moving towards a future of BI software that caters to the root need for intelligence — i.e. not only that you see your data clearly, but you also see it in your specific business context, and get immediate option to act upon it.
For e.g. a marketing analytics BI application – once it incorporates the data about customers and marketing campaigns and resulting purchases — should not wait for a user to define dashboards and reports, but rather already provide a suite of analyses that answer the most typical marketer’s questions – i.e. how effective are my campaigns , who are my best customers/prospects, and what tactics work the best for individual customer segments? And don’t stop there btw — if you have identified some new and interesting customer segments – it should integrate with an online campaign manager tool to immediately launch new campaigns; or publish the list of your most valuable customers to your e-commerce engine or call center platform which know how to treat them in a special way; etc. etc.
Similar scenarios can be applied to other industries and domains too. We as BI software developers just need to imagine differently. And that’s what 2.0, new version, or next generation is all about – next level of imagination.
I attended a talk at the monthly BI SIG meeting at SDForum by Christian Marcazzo from Spotfire, now a part of Tibco. I have long admired Spotfire’s innovations on data visualization front, so I was curious how they see BI from the whole Human-Computer Interaction (HCI) aspect, and couple of things stood out.
First – if we look at consumer-centric data applications (Zillow, Google Finance, etc.) and compare their interfaces to more traditional enterprise BI applications, it’s amazing to see how the latter just doesn’t even attempt to look good.
Why is that?
Because enterprise BI app developers aren’t under the same pressure to seduce their users like consumer data apps. Zillow, Google Finance, et al live and die by the community they create, so for them, user experience in paramount, and it shows. Most BI apps, on the other hand, are almost developed under the assumption that users are under a “thou shalt always use this BI software” executive order, and as such don’t have much leverage in rejecting a software based on poor or sub-optimal user experience.
So they begrudgingly use the BI software for its least interesting/effective use – churn out one report after another. The BI app basically becomes a report production factory.
That brings me to my second point – for BI to be more than charts/graphs/dashboards, it needs to be part of the user’s workflow. Now the term “workflow” means a lot of different thing to differnt people, and has recently become a popular box in BI markitechture diagrams – but to me, it basically means that BI app needs to know the various contexts under which its users are using it, and provide a way to add intelligence/insight to the process. BI app by itself should almost be invisible.
Zillow users think of themselves as a home buyers/sellers, not a real estate data analysts. Typical Google Finance users are checking out their portfolio and evaluating stocks, but don’t think of themselves as financial analysts. So, why in the world BI applications are hell bent to think of their users are data analyst first, rather than understanding the specific tasks they are trying to accomplish more intelligently?
That’s where workflow comes in. BI app needs to understand the nuances of the business domains their users are in, make intelligence available in their task workflow where it’s needed, and provide a clear way to act upon that intelligence. Too often we think of BI as a separate app where a user will do analysis, and then the users will jump to other apps where they can take actions — that’s now how users see the world. And without understanding the users, BI apps can’t really provide intelligence.
It’s time to turn this model around. BI apps should think more like mashups –pull data from any “public” repository with REST like API’s, make anlayses available to share and tweak, and make the resulting insights be integrateable to other apps. The more lines get blurred between BI apps and the rest, more successful its adoption is going to be.
I frequently check out Mosha Pasumansky’s blog on OLAP. I learned from his recent post that Microsoft announced 2 very interesting milestones coming up on their BI roadmap – project Gemini (which will be the new incarnation of Analysis Services), and SQL Kilimanjaro release (a move toward column-oriented architecture). If not anything else, check out the presentation video from the 1 hour 16 minute mark — this is a pretty good presentation of common BI challenges at BI level, and I must say the demo is impressive in how Microsoft is thinking about the BI solution stack going end-to-end from data warehouse to Excel and to a web-based view for general interaction.
As someone who is in the “open source BI land”, I must confess that I am a fan of some of the Microsoft’s BI technologies – namely Analysis Services and SQL Server. Yes, I have reservations around Reporting Services, or embedding BI into MS Office products like Word, or about bloating a solution with SharePoint and PerformancePoint – but as a common denominator, SQL Server and Analysis Services do provide the best price-performance today for a BI backend solution IMHO.
One of the challenges we constantly face as BI solution providers is to call out which is the most common interface for the BI user. In their demo, Ted Kummer and Donald Farmer are right to point out that if left to their own devices, most people trying to do a data analysis will bust out Excel. Solution providers like me don’t like this fact for several reasons (a lot of them may be valid) and try to guide our users towards purely web-based interface to do their analysis. The biggest rationale for this is to avoid various versions of Excel files floating around with multiple copies of data and custom calculations (with no QA) — and so we like to control by having all BI users access data from a centralized web interface, which is reporting data from a centralized repository — and that way, we know that users will be guaranteed a “single version of truth”, and they will be happy.
Is that true though?
In my own open source BI project OpenI, one of the most used feature turns out to be “export to Excel”, so try as we may, there are valid reasons to cater towards a BI user’s natural flow of anlayzing data, and let them get their data into Excel.
And in that sense, Microsoft’s approach may have its merit in looking at Excel as the piece in the front and center for self service BI. Of course, calling it “democratization” maybe far fetched because this democracy will only be true in the Microsoft Office world, but it is a pretty big world of BI users. And for those who would like to stay far away from the Microsoft Office world, there needs to be equally compelling alternate solutions (open source or not).
If not anything, this thinking from Microsoft is worth for all BI practitioners to consider — and see the demo. We may not agree with the exact tools used, but the use case, or the scenario, of a knowledge worker finding the data/information they need, analyzing it in an intuitive fashion, and publishing it for their peers to see — that’s a key part of what we’re all trying to solve. And unless we make it utterly easy and painless, we still have a long way to go.
It almost seems like the big BI companies figured out they had to be a part of something even bigger to justify their already sky-high license costs. Just within last 12 months, we’ve seen Hyperion get bought Oracle, Business Objects get bought by SAP, and now today Cognos got bought by IBM.
And I can’t help but feel that Enterprise BI software has become a dying breed.
If you are a fortune 1000 type of organization, chances are you already have some sort of elaborate licensing agreement with at least one of these acquirers (Oracle, SAP, or IBM). So now, they have one more thing to sell you so that you can have a check mark for your BI initiatives without having to drive too far to a different vendor.
I guess I shouldn’t be so cynical. As BI software, Hyperion, Business Objects, and Cognos are pretty feature-rich, and have many compelling qualities. My problem is with their complex enterprise deployment model, proprietary nature, and staggering cost of ownership that continues standing in the way of making BI available to the masses. And their recent acquisitions further polarizes the world of BI haves and have-nots.
Perhaps this is how corporate world will come to demand a new breed of BI solutions. Back in 2002, when we started our marketing analytics company, we thought (and still do) open source and software-as-a-service were going to drive this new trend. And it’s good to see companies like JasperSoft, Pentaho, Swivel, and LucidEra lead the way in this direction. What’s more important than just looking for open source and/or SaaS solutions to BI, is to demand “openness” — openness in terms of architecture, sharing of insights, and just as importantly, openness about cost of ownership. How many times you have seen a BI initiative where the entire budget got spent just on data integration and/or cleansing? How about the cost of those special consultants who seemed to be the only earthlings that understood the esoteric workings of a popular yet proprietary BI software package?
Above all, one needs to keep in mind that BI is about leveraging all available data to get a clearer picture of what’s going on in the business, be able to focus on the most relevant issues, and make better decisions. Instead of taking a centralized approach of one system doing it all, a good BI system today needs to act more like a network that can connect to various data sources, systems, API’s, web services, etc. What if your BI system acted more like a mashup that lets you combine compatible information sources and cross-reference them as you please? Swivel has an approach close to this except that it expects all information to be uploaded by its users. It could be interesting if Swivel could connect to some common public data repositories (like geographical locations, weather, stock prices).
The key here is to have BI systems that enable mashup of information resources and analyses, a la web 2.0, as opposed to being traditional “enterprise” solutions that need your business to bend over backwards to fit into their proprietary framework and terminologies. Which to me conjures up images of Oracle, SAP, and IBM.
Maybe I liked them better when they were Hyperion, Business Objects, and Cognos. At least they didn’t say — “BTW, we also do BI”.
I received an email from Seth Grimes, a very familiar voice in the BI community, with an interesting question — Do you now see BI as a commodity market?
He was referring to an earlier post about our open source BI project OpenI, where I’d mentioned:
The state of business intelligence software market has been very much controlled by a few big players. The situation is very similar to how the J2EE application server market was before JBoss, or how database server market used to be before MySQL and Postgres emerged as serious alternatives, or how OS market was before Linux. Pretty soon we will talk about the BI platform market in the same manner, because open source and open standards are driving the commoditization of BI as we speak. It is just a matter of time.
That was a few years ago. So, I asked myself — well, how do I feel now? Have I learned anything?
The question is a tough one — something I’ve always grappled with. Ultimately, it depends on what do we mean by “business intelligence”. If we go by the current big commercial players’ definition — then BI is more about a software tool providing capabilities around data warehousing/ETL, OLAP, analytical modeling, and visualization. So by that account, I’d definitely stick to my original thoughts and say it’s a commodity market.
However, a more relevant question might be — does having these capabilities make a business intelligent? In reality, what I’ve seen is that it comes down to an analyst (or group of analysts) who (a) know how to work a “BI” tool, and (b) have some fundamental expertise in the business domain they are analyzing. So, the “BI” tool is more about facilitating the job of an analyst or a general business user. You could argue that by making performance metrics, etc. more easily accessible to a business user, the BI tool is helping them make more effective decisions, but it is making a big assumption that the user knows how relevant the performance metrics are for the business.
In the end, my take on this is that, the most effective BI tools are domain-centric, i.e. they embody some inherent knowledge about a particular business domain — so, not only they are extremely efficient and accurate about compiling all the performance metrics and making them available, they also “understand” the applicability of those metrics and can almost act like expert systems in guiding crucial decisions. This, I don’t think is a commodity market. It needs to be grounded into specific industry domains to be effective.
Software industry, not unlike any other, contains a lot of hype and probably sometimes even more so with all this 2.0 buzz, which probably seems cool to the industry insiders, but is definitely confusing to the market.