Tuesday, March 17, 2009

Business Intelligence trends: HP's take

Buried amidst the consolidation of business intelligence (software) vendors, HP appears to base its solution on Knightsbridge, which it purchased in 2007.

It can't be too surprising if HP (and, previously Knightsbridge) aren't familiar names in the BI field. They do not figure regularly in industry reports and forums. HP's profile is quite low - so low, in fact, that they're not even mentioned in Wikipedia's list of BI tools. I expect their marketing strategy is limited to supplementing their provision of whole-of-business solutions to the marketplace: when selling to enterprises, something akin to "Oh, and we also supply business intelligence solutions - and consulting services. No need to go to the market for that". (However, for a counter-view on HP's BI profile, see this blog by Shawn Rogers.)

Notwithstanding, they recently released their take on BI trends for 2009, as follows.

Trend #1: Consumerisation of IT
In effect, business to adopt consumer-level technologies such as facebook and twitter. BI-specific effects in collaboration, visualisations, new data sources.

Trend #2: Post-Western tech economy
"Emerging regions" will transition from being simply "suppliers of low-cost talent" to being developers of best practice and global standard-setting consumers. Benefits to BI in terms of innovation - in analytics, unstructured data, etc.

Trend #3: BI importance increases; data governance and quality to become critical
So says HP - but see BI Survey's comments on less than expected adoption. However, it's to be expected that in hard times businesses would turn to BI for efficiency gains.

Trend #4: BI Buyers more scrutinising
- linking projects to business outcomes.

Trend #5: Market demands lower BI complexity
Commodified BI, standards for data marts. Here, HP contradicts its own earlier comment about SaaS/Cloud issues not yet figuring prominently in BI. They note it under "consideration".

Trend #6: Analytics moves to the front office; business users get greater sophistication
- including data modelling in the hands of business managers - scary!

Trend #7: Data integration increases in importance
Consolidating data from traditionally disparate sources; increasing focus on enterprise-level information management strategies.

Trend #8: A blurring between data warehouses and operations systems/data
A need for realtime operational reporting - enter, data hubs, Enterprise Service Buses, etc.

Trend #9: Convergence of structured/unstructured data
- this is a brave call, as business are only just starting coming to grips with the unstructured data buried in documents, notes, etc. My call is that it will be some time before unstructured data sees much effective use, let alone convergence.

Trend #10: CEP (Complex Event Processing) comes of age
- this seems to be an amalgam of alerting and data mining, nearing real-time.

As they later note: "To make the most of BI, first you need to get the data right". Much as that sounds a truism, it is a point that needs to be hammered at every opportunity, from data modelling to quality/management/governance.

HP's full report here.


Anonymous said...

Thank you for this post. Some excellent points.

However, on point #6 about putting data modeling in the hands of business managers your comment is "Scary". I would say "excellent idea".

It have always argued that the data model is a business model not a systems or IT model.

IT have tended to keep this in their ivory tower away from mere mortals "who could not possibly understand it".

Does it matter? Would it be of benefit to the business? Definitely!

I saw a very specific instance of this in a finance company where, due to a lack of understanding of data structures, the business managers misunderstood what their unique products were.

A basic understanding of unique identifiers are (and I do not mean codes or primary keys) would have enabled them to both sharpen their marketing and their product performance measurement.

So, yes, get data modeling into the business and let its power be realised there. Lets remove the mystique and demystify the subject.

If I understand it anyone can!

S Simmonds said...


Good on you. Yes, I agree with all you say.

And that mention of business managers not understanding their product is quite salient.

(However, you could argue that a manager should ipso facto know their product anyway - equally, it could be said that all too often they don't - not to that extent.)

I would reiterate that giving business managers control of data modelling is scary - but I'm not saying that it SHOULD be thus.

Business managers should by rights know their business inside and out, including their own data (and by extension, their data models). But that does not mean that they do.

Some business managers are quite analytical - that type is more easily to be trusted with an understanding of data models.

But to date, the area of data modelling is rather arcane for your average business user, even one that gets the general idea.

Some managers _are_ bridging that gap. But I would say that it's quite encumbent on the data modelling tools (and the analyst purveyors) to become far more accessible to business.

That is the challenge of IT professionals.