The two sides of BI
As is the case for most important categories of technology, discussions of BI can get confused. I’ve remarked in the past that there are numerous kinds of BI, and that the very origin of the term “business intelligence” can’t even be pinned down to the nearest century. But the most fundamental confusion of all is that business intelligence technology really is two different things, which in simplest terms may be categorized as user interface (UI) and platform* technology. And so:
- The UI aspect is why BI tends to be sold to business departments; the platform aspect is why it also makes sense to sell BI to IT shops attempting to establish enterprise standards.
- The UI aspect is why it makes sense to sell and market BI much as one would applications; the platform aspect is why it makes sense to sell and market BI much as one would database technology.
- The UI aspect is why vendors want to integrate BI with transaction-processing applications; the platform aspect is, I suppose, why they have so much trouble making the integration work.
- The UI aspect is why BI is judged on … well, on snazzy UIs and demos. The platform aspect is a big reason why the snazziest UI doesn’t always win.
*I wanted to say “server” or “server-side” instead of “platform”, as I dislike the latter word. But it’s too inaccurate, for example in the case of the original Cognos PowerPlay, and also in various thin-client scenarios.
Key aspects of BI platform technology can include:
- Query and data management. That’s the area I most commonly write about, for example in the cases of Platfora, QlikView, or Metamarkets. It goes back to the 1990s — notably the Business Objects semantic layer and Cognos PowerPlay MOLAP (MultiDimensional OnLine Analytic Processing) engine — and indeed before that to the report writers and fourth-generation languages of the 1970s. This overlaps somewhat with …
- … data integration and metadata management. Business Objects, Qlik, and other BI vendors have bought data integration vendors. Arguably, there was a period when Information Builders’ main business was data connectivity and integration. And sometimes the main value proposition for a BI deal is “We need some way to get at all that data and bring it together.”
- Security and access control — authentication, authorization, and all the additional As.
- Scheduling and delivery. When 10s of 1000s of desktops are being served, these aren’t entirely trivial. Ditto when dealing with occasionally-connected mobile devices.
The set of business intelligence vendors that have prospered without noteworthy platform technology is approximately {Tableau}. Candidates I omitted from that set include Spotfire (didn’t get far before being acquired by TIBCO, and perhaps not afterwards either) and Xcelsius (I’m not sure how far it got before being acquired by Business Objects).
Just as platform technology has been essential to BI innovation in the past, I think the same will remain true in the future. For example:
- ClearStory is throwing a lot of platform-side tech at what amounts to BI for third-party data.
- Alerting and metrics management is a long-standing opportunity, and definitely calls for platform-side effort.
- I think the same is true of BI integration with predictive modeling as well.
Bottom line: BI innovation usually depends upon serious platform technology.
Related links
- I observed in January, 2012 that analytic technologies tend to be adopted departmentally. I reiterated that for the specific case of BI in my “Things I keep needing to say” this week.
- Endeca was another BI vendor whose UI differentiation was based on a proprietary DBMS-like engine.
Comments
11 Responses to “The two sides of BI”
Leave a Reply
I would categorize the two sides as: 1) BI infrastructure and 2) BI tools
BI infrastructure is necessary but insufficient on its own.
BI tools address different needs of different audiences (e.g., analysts, IT, scientists, end users) and complete the last mile for turning data into meaningful insight.
Most (vendors) want to be a platform (play). The problem is that there are very few platform slots at a given prospective customer.
Better to start off as a spoke (solving a specific problem/need) before trying to become a hub…
Ken,
I fear you fell right into the confusion this post was meant to combat. 🙂
My point is that it’s tough to be a relevant tools player without building a lot of tool-specific infrastructure.
Now, I know that you are among the marketers who believe that, even if this is true, vendors should usually shut up about the infrastructure differentiation and talk rather about the business benefits. And you know I think that you take that view a little too far. 🙂
But my real point here is a technical one. For your tools/infrastructure distinction to match vendor reality, we’ve have to call analytic RDBMS, ETL, etc. “business intelligence infrastructure”, which I don’t think is a helpful way of phrasing things.
Did not know you thought so low of me. 🙂
If you are (primarily) an infrastructure vendor, you should absolutely talk about your infrastructure differentiators by all means (especially if you are in a well-defined category with clear competitors).
Sometimes it makes sense to lead with infrastructure differentiation. Sometimes it does not. It completely depends on the audience (and the knowledge/skill of the person discussing those differentiators).
Not a low opinion, Ken. Just a disagreement that has played out across more than one company. 🙂
But at the moment I’m also a bit put out that you’re responding to my post as if you hadn’t read it. My core contention in it was that your distinction is a false dichotomy.
Why doesn’t Tableau Server qualify as “platform” technology?
I’ve never been aware of anything differentiated on the platform side at Tableau.
Fair enough. So, in Tableau’s case, BI innovation is driven by the UI and supported by an undifferentiated but “good enough for most” piece of server technology?
That’s my impression, although I’d change your first “by” to another preposition such as “in”. 🙂
Do you think UI innovation is more replicable than platform innovation?
It’s easier to copy, unless there’s some implementation issue that makes it hard to copy. And the implementation issue would likely fall under “platform” technology.
That’s in the BI world. In other scenarios, the implementation issues might, for example, be in the area of OLTP application software.
[…] suggested in the past, approximately, that the platform technology side of business intelligence is more significant than the user interface. That formulation, however, doesn’t exactly capture what I believe. To be more precise, […]