Should you offer “complete” analytic applications?
WibiData is essentially on the trajectory:
- Started with platform-ish technology.
- Selling analytic application subsystems, focused for now on personalization.
- Hopeful of selling complete analytic applications in the future.
The same, it turns out, is true of Causata.* Talking with them both the same day led me to write this post.
*Differences between the companies include:
- WibiData started out with some serious HBase/Hadoop technology, whereas …
- … Causata just changed its underpinnings to HBase/Hadoop …
- … after hiring new, application-oriented leadership.
I know WibiData (client since they had <10 employees) much better than Causata (one conversation ever).
The problem for those vendors and other analytic application aspirants is that it is very hard to offer a complete analytic application. In particular:
- Suppose they want to offer a great solution for, say, website personalization.* It’s hard to do that without offering something that creates complete websites — specifically, complete unique websites. Whoops.
- OK, let’s suppose they solve that problem, drawing a clean line between the personalization and creative parts. Then is it really enough for them to just personalize websites? Shouldn’t they also personalize email? Mobile ads? In-store offers? Shouldn’t that all be tied to campaign design? And by the way, they need the capacity to incorporate almost any kind of data you can imagine, while applying any kind of modeling algorithm that can offer differentiated results.
- On the other hand, suppose they only deliver the common analytic subsystems for various functions? How do they sell that? How do they even demo it? Are they at the mercy of “last mile functionality” partners?
*There are various semantic issues as to whether the correct word is “personalization”, “customization”, etc. In this post, I’m ignoring them. 🙂
My proposed answer starts:
- Even though it’s impractical to offer across-the-board, full-featured, full-suite, highly competitive analytic applications …
- … offer something that purports to be a complete analytic app anyway.
Maybe the “complete” app is, from the customer’s standpoint, at least a “good start”. Maybe you really can deliver an awesome application for a narrow area of functionality — and the customer adopts it with confidence, knowing that she can integrate the core technology into a broader suite if she wants to.
As I’m telling the story, the real differentiation is apt to be in the subsystem, not in the finished app. So for a sanity check, let’s consider when would that might be the case. Examples that come to mind include:
- Small-/mid-market, vertical-market BI. The best example of this may be Google Analytics, for website owners and administrators — but that’s most famous as a free product. Perhaps there are also examples in more conventional enterprise-adoption scenarios. (PivotLink for retailers? I’m not sure how mature their application functionality really is.)
- Any of the four scenarios I outlined in my post on third-party analytics. One notable example is stock quote services such as Bloomberg. But that’s really an information-selling business much more than an analytic-functionality one.
- Price-setting analytics — Zilliant, Vendavo, and so on. Those outfits indeed seem to focus on application fit-and-finish as much as on price optimization expertise. But I’d guess that the most successful companies in that market are still in the 10s of millions of annual revenues; for example, Zilliant recently boasted of its 100th customer.
I don’t think any of those cases are sufficient to undermine my conclusions, namely:
- Making a big business from “complete” analytic applications will in most cases require some heretofore undiscovered insights or conceptual breakthroughs (business model or technology as the case may be).
- Analytic application subsystems are where most of the near-term opportunity lies.
- It will likely be wise to offer “complete” analytic applications even so.
Comments
5 Responses to “Should you offer “complete” analytic applications?”
Leave a Reply
Curt, makes complete sense, we agree.
At Causata, our first complete predictive analytic application was website personalization. We stop short of building or providing a CMS; we integrate our profile, segment and decision APIs with CQ5 and other custom CMS’.
Our second complete predictive analytic application was email content targeting with API integrations to ExactTarget, CheetahMail, Conversen, and more.
We are now working closely with our customers on predictive analytic applications for targeted advertising and CRM. Each application we provide tends to be industry-specific; that makes it easier to deploy and generate faster ROI.
AS you correctly point out, the underpinning of our enterprise-class DMP is an analytic sub-system built on top of our event-driven data store (in HBase), our Identity Graph and Machine Learning algorithms. This is the core of our technical secret sauce.
We don’t think we need to build a CMS, a Content and Ad Server, or a CRM system in order to deliver great value and a complete solution to data scientists, analysts and even marketers.
Thank you again for raising awareness for this subject and related next-generation technology.
“Each application we provide tends to be industry-specific; that makes it easier to deploy and generate faster ROI.”
As analysis is democratized and business users need to make quicker decisions based on their data, having industry-specific applications makes it easier for them to leverage predictive analytic tools.
At a recent NY Enterprise Technology Meetup event we had a demo from Visual Revenue which showed off its platform focused on the media industry. It offers the only real-time analytics solution that is designed specifically to enhance the hand of editors in data driven newsrooms.
You can check out a video of their demo at http://youtu.be/bkCn_nJbey4
[…] application subsystems. Customer-facing analytic applications will have a whole different standard of completeness than more traditional back-office […]
[…] I’ve argued that analytic apps need to be continually customized, which is about as strong a pitch for agility as one can make. And that’s one of the major reasons that packaged analytic apps can’t really be feature-complete. […]
[…] In February, 2013 I reviewed some difficulties around analytic applications. […]