Blog

Earlier, I blogged about the upcoming Kilimanjaro and Gemini releases from Microsoft: self-service BI tools that are delivered inside Excel.  David Stein left a comment which prompted me to write this separate blog entry about self-service BI in general.  Basically, David’s comment was that the quality of data won’t be as good because the end users don’t really know what they’re doing with the data.

He’s right, but it’s the lesser of two evils, and it’s a choice that the organization has to make, and every case is different.

Think back to those movies and TV shows of full service gas stations back in the 1950’s-60’s.  I remember seeing snappily-dressed ladies pulling up to full service pumps where the attendant hustled out, started pumping the gas, and proceeded to pop the hood to check the car’s oil and the tire pressures while the tank filled up.  If it needed oil, he’d grab a few quarts, take care of the engine, and the driver sat comfortably in the car the whole time.

Today, when I pull up to the gas station, I fill the car up, but I gotta be honest: I don’t check the oil level and tire pressure every time.  It’s probably okay, but it might not be, and if it’s not, the only person I’m hurting is myself.

I could go to a full service pump, pay more, and get that taken care of.  I could drive happily knowing that everything in my car was operating at its perfect levels, and that I was avoiding disaster.  But I don’t – I save a few bucks and go to self-service.  Judging by the vanishing number of full-service pumps, I’m not alone.

Plus, it’s not always my car: when I’m rushing to return a rental car before I jet off somewhere else in my glamorous (cough) life as a traveling SQL Server expert, I don’t give a rip whether there’s any oil or if my tires are underinflated.  I wanna throw gas in there and get out as fast as I can to make my flight.

BI faces a similar pressure.  With the advent of self-service BI, our end users will have a choice: do they want perfectly maintained data with no chance of disaster?  Or are they willing to save a few bucks, vet their own data (or not vet it at all) and pump their own reports?  If they choose self-service, they won’t have to sit around waiting for the attendee to come by.

Heck, they can build their own BI reports anytime, anywhere, much like I enjoy driving up to an unmanned pump and filling up my rental car without speaking to a single human being.  (Sorry if that sounds antisocial, but the folks at service stations aren’t always the best conversationalists.)  Sometimes, people just want the best data they can get in the least amount of time so that they can race off to make the next business decision.  Not everything can wait for an ETL process design meeting.

Oh, and by the way, there aren’t a lot of jobs available as full-service gas station attendees….

↑ Back to top
  1. Thanks Brent, I don’t think anyone has ever blogged about me before.

    However, I must disagree with you again. Every new BI product purports to do just what you said, give users control over their data. I’ve sat through a number of them just in the past year from smaller, proprietary products, to high end BI solutions from Business Objects/SAP. What they all gloss over is the requirement of creating relationships between the tables and the experienced required to do so.

    I may be a bit cynical as the primary databases I’ve been working with for the last decade are not well designed or documented. Therefore, your analogy doesn’t accurately represent the situations I’ve worked on.

    A more accurate analogy would be to flying an airplane. It has wheels, engine(s), brakes (flaps), gauges, etc just like a car. Both are used for transportation and some cars can move as fast as some airplanes. In fact, until the last few years, I could pick up any number of video games which give you the feeling that you could pilot an airplane. This is clearly not the case.

    Can users manipulate their own data effectively and accurately? Yes. However, the underlying queries must be provided to them if the reports they want are anything more difficult than dragging and dropping (creating links) from one field to another.

    I won’t pretend to have the level of experience that you have, but every one of these DIY data analysis products has been akin to selling snake oil.

  2. Hi David,

    Probably too late in the day (1 month plus) but not all of the DIY analysis products are snake oil – I am talking about ours (Monarch), naturally ;)

    Instead of using databases as the source, we use standard reports, so all the hard work has been done, as regards data cleansing, business value, auditing etc.

    In addition, the data, layout and relationships are familar to the business user, so they are easily able to extract the data they need without having to worry about database-centric concepts.

    Now I don’t claim that we offer high-end BI, most users will eventually export the results of their model into Excel and maybe do some more work there, but it does make the users self-sufficient without the types of risks you are talking about.

    The users (mainly finance) are domain experts in the reports they use on a daily basis, so they can easily identify the data they need and extract and analyse it, without IT involvement.

    In addition, the analysis can always be tracked back to an absolutely authoritative source – the reports they rely on for business operations, as well as audit and regulatory purposes.

    The controls are already in place as regards who is allowed to view those reports, and they are pretty much always archived – so analysis can be checked back for years, if needs be.

    In addition, it can be the only way to analyse some “point-in-time” reports where replicating the business logic on top of the database that generated that instance of reporting data is just not feasible.

    Don’t want this to seem like an ad, but it just hurt a little to be considered a snake-oil merchant.

    Check out our forums if you want an independent user opinion.

    Gareth

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

css.php