Why BI is a lemon

First off let me avoid a whole class of comment by saying that I personally think BI (Business Insight) offers real value when used effectively.

But it is still a lemon (this is not a good thing for anyone not familiar with English colloquisms) as an industry sector.

Here is why:

90% of spreadsheet users are just pointing and clicking

9% are happy doing their favourite index/offset/vlookup/indirect/external link magic and have no interest in learning a new technology, even if it will make a bunch of stuff easier for them. These may have a BI product but still use their fave spreadsheet app to do the heavy lifting.

0.9% used a BI product at a previous job, or want to introduce it anyway, but are blocked by some IT total blockage (‘No’) or major roadblock (‘identify and quantify the business costs, benefits and risks’)

0.1% are happily whizzing through work at 10-100 times faster than the other 99.9% (using a BI product), and with much better business insight. But no one outside this 0.1% listens to them so they are poor influencers. This group is declining in number.

Of course it doesn’t help that BI keeps trying to change its name to avoid the stigma attached to ‘BI’ and try and generate more press.

Its great technology but like so much stuff in the Office/information worker space, its provides potential benefits to a class of user that have no purchasing or configuration power. And that potential user has an ever harder battle on their hands to convince their IT dept to be proactive and make a change in the hope and expectation it will improve things.

It depts have several teams. The config team doesn’t want the hassle of the install, config maintenance and licensing. The dev team want the users to use their crappy in house .net/DB2/Oracle/job protection system. Until Jobserve is buckling under the job ads for BI specialists IT dev types will continue to be disinterested by it.

Dunno if this will change as we come out of the depression, I expect not. A few mega success stories, but plausible ones for a change might help. ‘How I cured world hunger by implimeting Essbase and clearing up in options’ might do it. I hope it does, because as I say I think there is major value in decent BI. but I think the class of user with the most to gain have little or no interest. Dunno why, fear change?

what Bo you think from what you see where you work and in other areas?




18 Responses to “Why BI is a lemon”

  1. Harlan Grove Says:

    I’ll ask a dumb question: isn’t the classic BI insight putting disposable infant undergarments (neatly side stepping the diaper/nappy terminology ces pit) next to the beer coolers? That is, desperately seeking correlations that might be exploitable? I suppose I should add extrapolating trends based on 3 or 4 data points?

    I had formal university instruction in econometrics. Most of the BI results and analyses I’ve seen aren’t statistically sound. In US terms, it’s always struck me as quantitative analysis for English majors.

    OTOH, if you mean programmable tools for aggregating and comparing categorical data and producing univariate statistics and factor analyses WITH relevant statistics for hypothesis testing, like what a decent stats package provides (e.g., SAS or SPSS which now calls themselves BI packages rather than stats packages), then it can be valuable. Problem is only the 0.1% of users still recall how to perform hypothesis testing. The other 99.9% have no clue what it is, so view it as a waste of time.

  2. Stephane Rodriguez Says:

    “90% of spreadsheet users are just pointing and clicking”

    If BI reports are correctly designed by BI report designers, that is what they should end up with, and that is fine.


    Here is the analogy : why use an object programming language when you can use raw language machine?
    There is a reason why people over time want to use objects. And the reason why is that they can control the life cycle of the objects in question.

    Disclaimer : ex BOBJ emp

  3. Andy Cotgreave Says:

    It sounds to me like you did your rersearch in my organisation. I am continually trying to persuade colleagues that the tools I use are more efficient than Excel and produce better results, quicker. But there’s a vast amount of reluctance to change.

    Bizarrely, while most colleagues are in the mid- to late twenties, the one colleague who is most keen to move things forward is a 65yr old woman! She brilliantly smashes the stereotypes and is a great embracer of new technology!

    Our IT dept are relatively keen but our scenario is exactly as you describe: they need a detailed quantification of everything before committing to expanding what we have.



  4. Bob Phillips Says:

    @Harlan, yes, but seeing as it is those English majors who are the decision makers, it is the role of the BI designer/implementator to make it readable and understandable by such (Rob Collie recently had a blog post where he argued for flash and gizmos in reports to get the attention of such people – exactly the wrong approach IMO, that is pandering to their laziness).

    @Simon, this is a hugely depressing post, probably the more so as I probably think you are correct. Personally I believe, and I suspect Andy does also, that BI is the most exciting tool in the box at the moment. Connecting an easy to use and powerful analysis and visualisation tool like Excel to enterprise data cannot be argued against IMO. Good BI can add so much value to any company. Times such as now are exactly when BI is most needed, to provide the decision makers with enough information to stop them making those crass decisions (such as outsourcing). Unfortunately, in our macho business culture, it seems more important to make instant decisions to considered decisions, hence why the gizmos are so popular.

  5. Will Riley Says:

    Well as a BI architect I guess I’m biased ;)

    Obviously I see great value in effectively implemented BI solutions and preach this mantra as

    1) I believe it
    2) If I succeed, my mortgage keeps getting paid…

    Like Stephane, generally the intended end result of BI implementations that I manage is a “point and click” experience for the 90%, either using Excel (or MSFT’s Excel Services) or something that can easily be exported to Excel (PerformancePoint, Reporting Services)

    I have worked with other BI tools (Cognos, BO etc) but I find that the “total IT blockage” much harder to shift with tools that cost in the region of £1,500 per seat as opposed to £200 per seat…

    Trouble is at the moment that the MSFT “BI for the masses” approach is not really being executed particularly well (epsecially for those organisations who don’t want to/cant’afford to upgrade to the bleeding edge of technology) – the BI offering is still more the “sum of many parts” than it should be.

  6. Harlan Grove Says:

    @Bob – it’s the business majors and the lawyers who run things. The english majors used to be employed to take what the engineers etc came up with and turn it into something senior management can understand. Now companies let engineers etc run business processes as they believe they should be run, but now IT misinterprets the engineers’ etc processes into Business Rules to create BI systems for english majors to use to produce snappy PowerPoints for senior management to discuss in the few minutes between discussing the compensation committee’s latest meeting.

    I’ve used and contributed to decision support systems many years ago. Those included some statistical tests and provided sensitivity testing (oversimplified: varying key metrics about their averages). The BI systems I’ve seen more recently aren’t as sophisticated – they tend either to assume history will repeat itself or that recent trends can be extrapolated arbitrarily far out.

    A BI system that involves just point-and-click is a reporting tool, not a decision support system. If the people using can’t easily assess statistical significance of various metrics or perform sensitivity analysis on the fly, all you’ve got it a reporting system. Still useful, just not the same thing.

  7. Bob Phillips Says:

    LOL. That third (long) sentence (have you been reading Roberto Bolano?) beautifully encapsulates the modern management process.

  8. Bob Phillips Says:

    But I agree on our main point, BI systems are not decision supprt systems, and should not be IMO. They should be able to present enough information in a comprehensible manner to focust decision makers on the areas that need looking at, and may need further analysis.

  9. sam Says:

    I think 90 % of the people equate BI to a PivotTable which is free

    In the future the same 90% will equate BI to a PowerPivot(for Excel 2010) which is …well another pivot Table with a few extra capabilites.

  10. Harlan Grove Says:

    In the end, BI will be as irrelevant as DSS for the really important things. E.g., AIG’s melt down. No computer system is going to point out potentially world economy destroying things like the effects of widespread strongly positive correlation between different risks. What BI or DSS system could have alerted senior management to the possibility that if a tranche of CDOs based on Nevada home mortgages went into the toilet, maybe CDOs based on California and Arizona home mortgages might also be toxic?

    If BI isn’t DSS, then the best if can do is help smooth out quarterly earnings so senior management need not fret overmuch about their bonuses.

  11. Charles Says:

    I started Decision Models 14 years ago to do DSS, and much of my work has been creating simple DSS tools (Pricing, Forecasting, Market Sizing, …).
    But I have seen almost zero interest in creating more complex DSS.

    If management can’t understand it they won’t buy it (except in Quantitative Finance where the reverse seems to be true).

  12. Simon Says:

    hmm kinda thought I might be going out on a limb there, guess not.

  13. James Standen Says:

    I firmly believe that business intelligence (in its classic form of data warehouses and/or Marts, olap, reports, portals etc.) is useful, valuable and I agree that there is a broad bunch of folks who just don’t see the value, perhaps partly because they focus on high profile failures, or they had expectations that were out of line.

    On the other hand, the methodologies that are used can often be very slow (gather requirements, design compelx data models that try to account for everything, build massive ETL jobs with expensive tools and teams of highly trained developers). Sometimes you need the data faster.

    I’ve often thought that there must be a middle ground in there somewhere- for data transformation that doesn’t need a full fledged BI project, yet is a bit too much for just whipping it up in Excel, or trying to hard code a mess of SQL.

    The product I’m working on now is working to build the kind of tool that fits the gap between spreadsheets and ETL/Database to provide a sort of data scratchpad that people could use to quickly pull together a bunch of data, and do stuff with it- like you often do in excel, but more.

  14. dougaj4 Says:

    “The product I’m working on now is working to build the kind oftool that fits the gap between spreadsheets and ETL/Database to provide a sort of data scratchpad that people could use to quickly pull together a bunch of data, and do stuff with it- like you often do in excel, but more.”

    Like a spreadsheet with a built in programming module?

    • James Standen Says:

      More like a simple ETL tool (blocks and connectors) that has some spreadsheet like flexibility, particularly in data typing, and a very visual interface- Venn diagrams for joins, drill down data profiling etc.

  15. Marcus from London Says:

    Of course the reason that neither the business nor IT want to push ahead with BI solutions is that it will force them to come face-to-face with the reality that their data is cr@p. The best BI tools is rendered useless in the absence of data quality and integrity.

    I agree with several of the other posters – after several data mart / BI/ MIS projects my conclusion is that BI is either misused or misunderstood. Either BI just adds another layer to producing pre-fabricated reports or users are simply slicing and dicing without the statistical due diligence Harlan mentioned.

  16. martin rushton Says:

    OT I know but rather than start another avalanche of posts on Excel-L. Did Bob Phillips really suggest a solution using Indirect this morning ;-)

  17. Bob Phillips Says:

    Yes I did Martin, if you know how to conditionally sum across worksheets without using INDIRECT I would love to hear it. It is a fool who refuses to use the only tool available just because it isn’t aesthetic.

    Glad you kept it off Excel-L, I have been drowned by Bill talking to himself recently, the’re all catching the bugover there.

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: