BI Excel and browsers

Theres a bit of a debate on one of the others posts about this so I thought I would pull it out into a main post. 

Heres what I remember of the recent history of BI.

In the 90’s all the BI vendors were tripping over themselves to get us off Excel and onto their proprietary client software. (*note)

In the 00’s they finally realised we had no plans to dump Excel and get tied into their client. They then tripped over themselves to deliver us Excel add-ins. Suddenly BI uptake improved.

In the last couple of years the BI vendors have started trying to move us onto their browser based proprietary clients. Are they succeeding where they failed in the 90’s? I have no idea, do you?

* note : Arbor swam against this tide with their Extended SpreadSheet dataBASE, in my view this is a key factor in the success of Essbase over the competition.

Personally I think the march to browser based clients may be misguided. To get the power of Excel to the users, these browser apps will need a fair bit of calculation grunt. I accept that deployment may be slightly easier, although I think most enterprises have got MS office deployment and maintenance pretty much nailed by now.

I struggle to see how providing a browser client with the same power as Excel, is any better than Excel? If its trying to save the licence cost, and the maintenance cost, I’d say the reduced productivity of multiple unusual interface will more than outweigh any savings. The big benefit of spreadsheets (you can get Essbase add-ins for OpenOffice Calc) is they can be used for many many things. One investment in learning the interface and functionality pays back many times.

Maybe there is a market for a BI provider that only has a spreadsheet add-in as a client, and then passes on the savings in development costs to clients?

Business intelligence is only useful to people who have the influence to affect corporate activity. I’d say these people are analyst and management types, not a massive number in any one organisation. Browser based apps seem better suited to wide scale deployment, lower functionality apps target towards administration activities (the classic example seems to be submitting expenses). Maybe its a fashion thing?

I’m not saying the browser thing won’t happen, what I’m saying is that I’m not clear on what the benefits will be for the users, the admins, and even the BI companies. If you can see the benefits then post a comment.




18 Responses to “BI Excel and browsers”

  1. Stephane Rodriguez Says:

    “I’m not saying the browser thing won’t happen”

    In the BI industry, it HAS happened. All major BI suites have seen the fall of their rich client alternatives. And new entrants (small vendors, open source) use web clients by design. In this industry, hardly anybody other than Microsoft is still pushing rich clients. We all know why.

    “If you can see the benefits then post a comment.”

    I can give you many, but since this blog is about spreadsheets, let’s say BI software fixes the “spreadsheet hell” problem. How, you ask? Single version of the truth, which materializes by a combination of several things, most importantly the semantic layer.

  2. Jon Peltier Says:

    “Business intelligence is only useful to people who have the influence to affect corporate activity. I’d say these people are analyst and management types, not a massive number in any one organisation.”

    One of the efforts of Total Quality and Lean Manufacturing is creation of a “Visible Workplace”. Part of this means each cell (manufacturing unit or department) has a public location where their key metrics are posted. It could be things like weekly tracking of parts produced, scrap rates, or similar quantities. Typically these are printed (or hand-drawn) charts and tables tacked to a bulletin board; it seems to me that these could be printed out from a browser-based BI system as easily as output from “private” worksheets or as drawn by hand.

    Such a browser-based system would be far broader than the circle of corporate influencers you mention. A decent BI system should be linked into the databases that contain relevant data down to the level of production and scrap rates of an individual cell.

  3. Will Riley Says:

    The march to browser based clients is misguided, I agree, although I think that MSFT have actually done quite a cool thing by allowing XL 2007 to sit on Sharepoint 2007 allowing the client PC to see the richness of the 2007 app rendering without the requirement for the software itself. bviously the client requires a CAL for MOSS, however MOSS can deliver far more than Office….. at least in terms of BI.

    I admit that I am quite a convert to the MSFT BI toolkit. The more i work with it the more i like it. In overall terms I think it could make massive inroads in to the SME market over the next few years and for good reason….

  4. Harlan Grove Says:

    BI delivered via browser is an end product. Unless one is a developer with a developer’s licenses and software, all one gets to do is load the end results into a spreadsheet. And in doing so, there goes one version of TRUTH.

    If you’re delivering a ‘dashboard’, fine. Nothing wrong with displaying it in a browser. If you’re providing a report generator, not quite so good unless your users want those reports as FINAL products. However, if your users need to generate reports to get data/information that’s INPUT into other calculations, then browser-based systems begin to suck. IF user customizable output from the BI system could be loaded into, say, Excel like a web query, wonderful. How many systems provide that?

  5. Marcus Says:

    I’ve been involved in a few web based BI implementations. Users *always* want to know if they can export a web report to Excel. This is not a question of whether the BI providers provide this functionality, but an expression of preference by the users. I’ve never seen a case where a user didn’t want to copy a chunk of data from a browser, tart it up in Excel or paste it into a presentation. I’ve seen companies spend millions on a BI implementation only to have users consistently export reports to Excel for their own requirements.

    “Single version of the truth”
    We’ll maybe. Once a user copies/exports a report to Excel, enriches (read: diddles) with the numbers and then presents them, our one version of the truth theory just got flushed away. Yes, the original verso in of the truth is still in the warehouse – but the board of directors isn’t looking at that, they’re looking at some-ones enriched extract.

    My experience has been (and everyone’s is different), that too many businesses got burnt or disillusioned with all the hyperbole they swallowed by BI vendors in the past. They’ve also found dealing with their own IT department painful enough to avoid them where possible. So where possible, many business units try to maintain control of the sanity by keeping their BI/MIS implementations self contained. Many places use Access, some SQL Server and nearly all use Excel as the client. Over time I’ve also seen business users getting more adept using QueryTables, writing SQL queries, MS Query and PivotTables to external data sources to maintain that self-sufficiency.

    P.S. for anyone around London – how healthy is the job market for people with Excel and banking experience?

    Cheers – Marcus

  6. Harlan Grove Says:

    With regard to diddling truth from the BI system, consider a simple example from Financial Services: revenue forecasts for a global corporation by currency. Something BI systems can do easily. Now consider someone in the foreign exchange unit within Corporate Finance needs to manage foreign exchange hedging. Maybe there are some BI systems that can use matrix time series of foreign exchange rates and perform the necessary canonical factor analysis to see if there are any arbitrage opportunities or gaps that need to be hedged against, but that sort of analysis is better left to the math, stats or physics PhDs who know the mathematics rather than to the typical BI developer.

    In this example, the forecasts would be useful to some as FINAL RESULTS, but they also are useful as INPUTS to other systems, often involving calculations that the average BI developer isn’t qualified or competent to implement.

    [Very little is as irritating as developers discussing misuse of their precious programs’ output by mere users when those developers are ignorant of the full spectrum of analytical tasks their users need to perform. Maybe 9 out of 10 users are just reformatting, but the 10th user is often doing things you probably can’t do for them.]

  7. Marcus Says:

    Interesting perspective Harlan. I’ve not encountered traders utilising BI systems as the data is usually at too high a level for their requirements and nearly always too late (i.e. not real-time).

    They more often use a combination of the respective trading system, mathematical modeling with the assistance of the resident quants and a measure of their own appetite for risk (capacity to gamble).

    Regards – Marcus

  8. Harlan Grove Says:

    I didn’t mean traders. Insurance companies and merchant banks can have variable future cashflows in many different currencies, and may not be able to hand it all off to a trading department. For those types of companies, cheap liquidity in every currency is the goal, and that’s easier to achieve with hedging than trading.

    I should have said data probably wasn’t headed for a spreadsheet in such a scenario. There are less sophisticated examples of users needing to take data from the most efficient source and perform more analysis on it. Often BI systems are the most efficent source because most business computer users outside IT departments are NEVER given user accounts on company database servers, so BI systems with customizable report generators are often the only tool available to pull company data.

    Then there are those of us who work close to point-of-sale who need to work with our own company’s data AND customer-supplied data (not available in the company database, so not coming from the BI system), e.g., calculating political risk insurance premium that depends on customer’s current and changing international holdings. That requires something more than an a BI backend and a browser.

  9. Stephane Rodriguez Says:

    “BI delivered via browser is an end product.”

    Since when have you used a BI product?

    Cognos, Business Objects (to name the two leaders) provide the entire report life cycle (including design, real-time preview, view, analysis) within a web browser.

  10. Simon Says:

    Excel VBA banking in LNDN is lively, any Bloomberg or Reuters is a benefit
    rates 3-600 GBP a day roughly
    Are you thinking of coming over?
    try here:
    and here for background info:

  11. Simon Says:

    I think its important to separate the client from the server in this discussion. Data versionitis is not an excel problem its a ‘storing data at the clients’ problem.
    I suppose I see BI as more than a few static reports, its the interactivity that makes me see it as an analyst tool. Static reports, browser, fine. interactivity, Excel, fine. Wothwhile interactivity and browser is what leaves me unconvinced.
    There are plenty of Excel based systems that store all data on a server, I think that these with the option to do disconnected analysis in a familiar app is more attractive than not having Excel.
    As Harlan mentions its misplaced arrogance of developers to believe they can second guess all the analysis and reporting and investigation that an information worker might do. Anything that could match a spreadsheet for this, would in fact be a spreadsheet, whether rendered in a browser or standalone. So whats the point of the browser?
    I’m not saying spreadsheets are the ‘best’, and will stay the best client, but I think they are the best for most experienced users that are familier with them. (second best tool for everything)(Of course I accept I am likely to be biased, and I wonder if I’m missing something?)

  12. Stephane Rodriguez Says:

    I never knew there could be such a religion about spreadsheets…I don’t know if that has something to do with those making a living of maintaining spreadsheets.

  13. Marcus Says:

    If the richness of web GUIs can be enhanced, web interfaces are a strong contender over Excel as a BI interface. However, I think that’s still a way off – all the MIS implementations I see tell me that Excel is still the preferred medium by a long shot.

    Hi Stephane,
    ‘religion’ may be too strong a term – perhaps Churchill’s ‘fanatic’, being “one who can’t change his mind and won’t change the subject”, is more apt :P

    Hi Simon,
    Yes – we we’re looking at coming over about 7 years ago. By the time I was able to regain by EU passport, the bottom had fallen out of the market. I’ve been watching jobserve but was seeking confirmation from people on the ground as a sanity check.
    A lot of my work has been in risk (market, institutional, operational and some credit), front office development and MIS systems with a variety of back ends. I was looking at perhaps Feb next year, potentially earlier. Thanks for the links.

    Cheers – Marcus

  14. Simon Says:

    I wouldn’t call it a religion, if anything I would call much of the browser stuff just a case of the emporers new clothes.
    The BI industry is world renowned for its bullshit and buzz word bingo. I just wonder if they have jumped on the Web 2.0 (another rather BS concept) band wagon.
    I make much of my living providing usable info to clients, I am not clear how a browser interface makes that better or easier than a s/s interface (for the users I typically deal with). Your are right on the maintenance issue, for sure that is something that needs managing.

    Marcus, I would suggest Feb/Mar is an excellent time, I reckon there is a bit of a boom just prior to the April financial year start. Anyone else see that?

  15. Stephane Rodriguez Says:

    “The BI industry is world renowned for its bullshit and buzz word bingo. I just wonder if they have jumped on the Web 2.0 (another rather BS concept) band wagon.”

    The semantic layer is what is at the origin of the BI niche in late 80s, now worth over 15B. Business Objects is a 1.4B revenue company whose semantic layer is a critical part of its value proposal.

  16. Harlan Grove Says:

    With regard to BI as end product, I’m thinking dashboards and canned reports. The kind of thing IT develops for senior management.

    I also mentioned report generation features. I use Business Objects’ WebIntelligence through the company intranet. In theory I could perform ‘analysis’ in WebIntelligence, but as a practical matter it’s more efficient to write query results to files, use Excel or an editor to do some simple text transformations, then use a stats package to perform statistical analysis.

    Maybe Business Objects provides libraries/add-ons for general linear modeling, time series, survival and factor analysis, but no one in the IT department has shown any inclination to tell anyone outside IT whether they’re available or not.

  17. Stephane Rodriguez Says:

    It’s fine for me if you have a preference on not using those tools for doing all what you do with Excel. As long as you don’t mean to imply you can’t do this stuff in the first place, which I think you were implying in a comment above, it’s ok for me.

    “but as a practical matter it’s more efficient to write query results to files,”. Depends what you mean by practical. I mean, there are people out there who write assembly code. It’s more practical to them. But let’s not discount the immense power of the semantic layer (“Business Objects” is actually the marketing name of the semantic layer).

    “Maybe Business Objects provides libraries/add-ons for general linear modeling”. Yep. There are hundreds of ISVs out there. One of the reasons is because this company took it to heart to behave much like Microsoft, by providing comprehensive SDKs throughout the entire product line. And it worked! In fact, if you take a look at job boards, there are plenty of BO professional related positions.

  18. Harlan Grove Says:

    Re libraries: they’re useful on the company intranet ONLY IF (1) there’s some indication they’re installed, and (2) there’s available documentation for them. Otherwise, they’re useless.

    As an employee who isn’t an IT department or consulting developer, there are no AVAILABLE BO libraries. Ergo, for me, it’s more efficient to use other tools. And I rather prefer getting answers today than spending months making a case for an IT development project.

    As for examples, a Google web search for +”Business Objects” +”general linear modeling” returns a mere 3 hits, two of which point to the same PDF file. So all the BO ISVs selling general linear modeling libraries don’t bother having their own web sites? Or try +”Business Objects” +”survival analysis”. Actually, that gives a few hundred hits, many describing stats packages that can fetch data from BO. I’ll let you see if there are any that mention how to perform survival analysis IN BO.

Leave a Reply

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

You are commenting using your 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: