Can a user really do BI from the desktop? Analysts at Gartner (and Forrester, and elsewhere) have predicted that 2011 will be the year of self service business intelligence, as users demand tools to help them access and understand company data. The users’ motives are easy to understand: identifying and responding quickly to trends and patterns in business data is vital …
There's plenty of non-Enterprise only BI tools.
OBIEE is an overcomplicated piece of krudd, tied to the horrible Oracle stack.
Products like Panopticon, Qlikview and Tableau enable BI analysts from corporate to SME environments to get a concise overview of their datasets, from their desktop.
I think you'll find that users already do. Business Objects has been on the desktop for years, and even Microsoft offer desktop BI through SSRS/SSAS embedded into SharePoint.
The difficulty here is twofold
1) "User" is quite a large group - data analysis is not the easiest of tasks so some training should be first in line.
2) What's the back end? BI is - at it's base - a federation of data, either from different sources, or areas within a single source. This means ETL/Data Quality/Warehousing all of which are specialist skills. The user is not going to be setting up their own "Kimbal-Approved(tm)" OLAP Cube any day soon.
However once the backend is set up, cubes built, universes defined, then yes - there is no reason why desktop BI isn't already a reality - the answers might be wrong, but you'll get them!
Anon - as I work in the BI space
There is nothing really stopping anyone gathering and understanding BI. Physically getting the data has always been the hardest part, especially when you have third party tech support companies and poorly chosen, unfit for purpose systems bought on the cheap, or legacy systems from partners who have been assimilated.
Having had some experience in the analysis and reporting of timely data, along with more in-depth insight, I can honestly say that it isn't difficult to do with traditional tools, but the new tools becoming more commonplace have their own problems. While you may centralise the process of data acquisition and reporting, you have to do a lot of work to convert your old ways to the new way. I am not yet convinced that it is better either. In some instances it has led to an increase in work to achieve the same result. I am currently threatened with this translation, but as per the usual large corporate timescales, what was supposed to be a painless 3 month transition has turned to a 15 month barren patch with no sign of the new reporting suites.
Ironically, I have also found that the data that gets reported on is often uncomfortable reading for some of the decision makers. This leads to relevant business data being sidelined or "reworked" to suit their own agenda. While this will cause long term issues, short term that manager gets a promotion.
For me, it is not so much about the tech, it is all about the managerial buy-in.
IME people have been doing this to some extent or other for years - in fact from before it was called business intelligence. The only change is the sophistication and scale that is possible with modern systems.
The key bit is not so much technical, but personal - do the people doing it understand what they are doing or what the data they are analysing actually represents.
Again, my experience is that savvy managers will request the "IT people" to create them a report, and then they can extract the data they want and import it into Excel (or whatever) with a few clicks. This is one area where systems have improved - it generally being easier to get at your data than it used to be.
It's at least 20 years since I was first involved with this question.
BI tools have come a long way (in some respects, in others they seem curiously to have regressed). But I suspect that the answer to the question is still "Probably Not". The way corporate data is stored is usually the problem.
What tends to happen is that somebody in an office becomes a BI super-user and either helps everybody else with their reporting problems or supplied them with canned reports. It's still BI as a specialist function, but the specialist isn't part of the IT department.
I agree (although my experience is also 20 years out of date). Our solution was to train up some of the DBAs to answer this type of query - who knows the way data is stored better than these guys? This only works if you don't recharge IT costs - otherwise you run into the problem that 'my people are free (to me), but I have to pay if IT do it'.
Ok, I am a programmer so it's slightly easier for me but almost all the BI work I do which can be anything from a simple tot up of e-comm sales to complex analysis from multiple data sources and web services I use Ruby as a script and put the data into MySQL.
I used to use Excel a lot and have made some nice financial analysis packages in the past but it's not suited now with so many different types and formats of data source, in my view.
A BI person, with a few days of Ruby/SQL training could, in my opinion, be extremely productive, and with a bit more training could automate those scripts to do lots of fancy things.
The problem with desktop business "intelligence" is that it's only as good as:
a) the quality of the available data
b) the understanding the end user has of the derived statistics
Typically, the available data will not be reporting exactly what the user imagines (because data gets bent to fit the collection method) and the user's ability - and in particular the ability of the typical spreadsheet jockey - to calculate what he thinks he's calculating is notoriously over-estimated:
http://www.eusprig.org/stories.htm
In my business (insurance) it's mostly the IT dept. where the problem lies, having built poorly specced data repositories and not understanding either what the data is or the importance of it to the company.
Their insular 'you wouldnt understand or you'd break it' attitude has prevented users obtaining the data they need for BI and even led to users seruptitiously obtaining the data they need without anyone in IT being aware.
My experience may be different from the majority but I think most people who really want, or need, better BI are substantially more computer literate than most IT professionals give credit for.
This post has been deleted by its author
What, for lambs sake, is "doing" BI ?
"Aw-wight Dave"
"Aaaawwww-wight Stan"
"Whatcha doin ?"
"I'm doing some intellgunce"
"Nice one. I hear Mike gone dun some brilliant intellgunce the uver day n'all"
"Nice one"
Second thoughts, I have a horrible feeling I don't even want to know, nor what that God-awful and over-used phrase "business intelligence" is supposed to convey. Is it really not possible to speak normally in whatever commercial environment this guff is spouted ?
It's not about if they can or they can't. BI is a concept. Tools are something that allow you to explore the concept and generate results.
What is really important is how you score the reliability of the results produced. I.e. Are the results accurate and relevant to the business. Garbage in, garbage out. Which of your staff do you trust?
This post has been deleted by its author
In my organisation at least, people already do using mostly Excel to produce custom reports, and this is causing me massive headaches. To be able to produce a report and / or business intelligence thing (KPI, Dashboard etc) I think you need to understand the data your working with. I can't count the number of times there have been disagreements over something because someones custom report says one thing, and another users says another.
IMHO, IT should be responsible for setting up OLAP cubes, reports etc so that there is a single cohesive report structure for an entire organisation. If key knowledge workers need more, let them access the cubes through things like PowerPivot and SharePoint.
As computers have got faster more data has been collected.
Some analysis is now possible on the user desktop in ways it wasn't before.
In some ways... snore... blah blah.... Excel, Qlikview, LogicXML... blah blah... R... BIG DATA!... Hadoop, Google... snort, dribble... MYSQL, cubes, RapidMiner, SAS. Some more chat about products. "Users are best placed to understand their data than analysts in some cases"
That pretty much covers it right? Speaking of which anyone seen Amanfrommars recently? I miss him.
One of the key problems I see is that while BI (as an initiative) is business-driven, and is essentially an evolution from allowing IT to produce the required reports (and waiting months for them to be written) the underpinning technology is still very much in ITs ballpark. So the gulf between the business objectives and ITs delivery capability is wide - and getting wider, due to the introduction of new technologies (or at least new versions) which IT have scant experience in.
The whole concept of centralising and optimising data within a single data model is such a step forward, that this needs to be completely understood at a business level before you can really reap the rewards. Executive buy-in is just the start:
Where's your logical data model?
What's your strategy for only extracting each item of data once from the source systems?
Where's your strategy for ensuring that every dept doesn't simply write the same report as everyone else? (Why have 100 reports when you could have 10?)
When a new project comes along, how will you verify, out of the data they request, which is already in the EDW and which isn't?
If you're extracting data from multiple global data sources, but those sources will be busy during their respective daylight hours, how do you manage the collection, transformation, updating, reporting, backing up & maintaining of data from 3+ global regions, all into the same data model? Which parts of that model dictate regionalised data (i.e. for region-specific backup windows) and which parts are a single entity?
Who's going to write the OLAP cubes? The users? If so, who supports them? You've spent $10m on those EDW and ETL platforms, and now Maureen from Finance is going to monopolise the whole thing with a badly-written query.
Looking at the presentation tier, who writes the BI dashboards for SharePoint? All the products are so new that IT keep being asked for sysadmin rights to be granted to all the servers - how do you educate these people to develop via the API when their boss is screaming at them to produce "a simple report" by tomorrow?
Until the business and IT learn to really talk to each other and respect the other's needs, timescales etc, then while self-service BI might be moving forward, the concepts of standardised, maintained, efficient, supported, operational IT are threatening to go backwards.
Desktop BI has been around for years..
with tools such as Essbase.. Oracle HFM and the use or any web base reporting tool at the back end Infrastructure,
puts desktop BI/SaaS at the finger tips of any business users..
allowing any standard user the eazy for manintating there own olap cubes, datasouces and reporting
without any IT support or intervision..
but the drawback depends on
if a business has been designed to run at a three tier support..
from Advanced users, Key users and to the Nervous users..
similar to the way todays ITIL services are being run..
if you require BI consultation/ BI Architects
and an ondemand access to our Oracle BI LABs
We're here: support@bi-labs.com
I've seen Disco in use. You still have to set up stuff for the users but it cuts down on some tedious reporting work for devs when they want things summarized in different ways.
If they know how to use it, the users will make heavy use of it. It's the know how that's a problem. And bad users means nasty queries.
Unfortunately, many ERP and CRM systems that I have experience of are designed for transactions and not data mining. Therefore, unless seperate images or infrastructure is planned in, the regular daily users tend to get poor performance.
Also, if dedicated BI data and infrastructures are not provided, power users inclined towards "ad-hoc" queries tend to lock up systems for critical minutes (which may be the critical life cycle of a data user).
User centric and user serviceable BI has been around and working as long as I've been in BI. To say this is the year of it is laughable.
The core problems in BI remain the same as they have for years :
Quality & Accuracy - golden sources. One version of the truth.
Timeliness
Relevance - common metadata definitions across the Org, handling unstructured data, handling commentary.
The major problem right now is vendor consolidation in the industry has produced these ridiculous behemoth systems that are over complex and overly expensive.
if http://www.ebay.com/itm/SiCortex-SC072-PDS-72-core-48GB-MIPS64-Cluster-System-/290568230232, then certainly yes. In fact, I was able to do some data analysis on 1.2GB RAM notebook using in-memory sqlite DB and this was 10x faster than production Oracle (because all data fit in RAM nicely).
.. though there is a lot of crappy oh-look-at-my-nice-n-shiny-2-seconds-to-make-a-report tools around which sure confuse the customers and the issue.
Problems are: crap data (by far the most important in terms of time spent), loose definitions (a good way to confuse users from different parts of the company), vendors' promise that aggrating years of incoherent and incomplete data can be done in months (weeks even!).
Once the data is cleaned and the dictionary is rubber-stamped, which crummy drag-n-drop tool is used to make a couple of tables or pie charts doesn't matter. The desktop tool doesn't matter. It is the back-end data scrubbing and management that brings the value.
Of course some tools like SAP BW have a crap back-end *and* a crap front-end, just to prove that anything is possible in BI.
There are a vast array of technologies out there that make Business Intelligence look easy… but anyone that has had to deploy any of them knows that making it look easy requires a scary amount of investment upfront in preparing & presenting the data, and providing frameworks for users to make it easy for them. And that this is not just an upfront investment, but an ongoing one to ensure the data remains fit for purpose and the frameworks keep pace with what users are doing.
But the fundamental questions come down to what you consider “Business Intelligence” to mean and who these “Users” are. If by “Users” we mean the masses within the Enterprise, I’d argue BI should stand for Business Information. Without wanting to be too harsh on the masses, they are usually data illiterate, don’t want to have to think too much about what they’re doing, and that exposing them to the [often] daunting collection of capability offered with modern-day BI suites is just asking for trouble… the opportunity to get the wrong answer is too high and getting someone skilled [if they have the time??] to unpick what they’ve done is tricky, if not impossible. So the masses needed to be safe-guarded with such well-guarded boundaries that “Self-Service Business Intelligence” usually becomes the ability to explore “Information”, not the ability to deliver “Intelligence”.
I built a BI suite from my desktop thirty years ago. It wasn't difficult once I added the data warehouse. So yes there's no doubt that desktop BI is possible.
There are two questions that I still don't understand. The first is why any company would hire an executive who isn't capable of desktop BI. The second is why the universities would award a degree to anyone who couldn't do it.
Of course they can "really do BI from the desktop" the problem is a bit more complicated.
Making tools easier to use and more integrated does not resolve the fundamental problem: in order to produce accurate information from data you need to understand the structure of the data and the effects of transformations you apply.
Users have had access to spreadsheets for a long time - and in my experience, the quality of the tools they produce using such spreadsheets varies greatly. I've seen millions of pounds lost by a business due to a single bug in a spreadsheet application created by a user (tool was never tested, never documented).
Actually, I think BI on the desktop is a great idea - after all, the further we can keep *some* users from production systems the better ;) Once again, from personal experience, I've seen people without extensive IT training bringing a production system to a halt by running badly behaved / innefficient applications on transactional systems.
So we can give the users tools which are easy to use, we can train them in data structures and development processes, train them how to test their applications, provide them with version control systems and document management systems....at what point do they cease to be 'users' and become 'developers'?
The biggest challenge with these systems is getting data into them in such a way that keeps integrity and is effortless to use.
If a fault occurs with a product, is the servicing engineer going to fill out a database form with 50 fields, or are they going to just write an e-mail?
You've got to make data entry rigid enough to make interpretation easy, but not so rigid that some things become difficult to record. As soon as that happens, the end user thinks "stuff it" and switches to e-mail.
Desktop BI is great for visualising simple data and can be quite compelling – some of the Tableau and QlikView demos are great. However these vendors are still selling snake oil – as anything that requires access to a central data repository or heavens forbid – joining disparate data sources together and then the project becomes, quite rightly an IT lead solution and the visualisation becomes a relatively small part of the overall project.
Desktop BI is also a cottage industry – somebody becomes an expert in knocking together reports and then they become the resource for the department. The accuracy of the data being loaded may be questionable especially if it is being manipulated to generate new measures and KPI and it’s easy to end up with multiple versions of the truth.
It is an unfortunate fact but BI solutions become complex quite quickly and for the most part are pretty dull – ETL, Data warehousing, snowflake schemas, multi-dimensional objects – all to get a chart telling you that the sales are going down the pan.
One final thing - I’ve not seen anything really new in the last few years and dashboards seem to be the state of the art. Where is the intelligence in Business Intelligence?
Metadevuk is quite correct. Users will never be able to produce reliable reports, fully-reconciled across all the tables and ledgers in an ERP system such as Dynamics NAV or AX, unless they have a single data warehouse/cube/mart (whatever jargon you use) to supply the data.
Creating such a data warehouse is something most IT technicians believe they alone can create. The reality is that they can't without thousands/millions of £££'s and months/years of effort and untold cash is wasted by companies allowing them to try. Without such a source users will alwys be arguing between themselves as to why their different reports show different figures or trends and the differences are nigh impossible or at best very time-consuming, to reconcile. That source also needs to be produced and maintained automatically, responding to changes in the underlying database table structure. Without this element you're back into being tied to techies updating databases.
Many vendors brush over or ignore the issue of real-time reporting slowing down transactional processing in businesses of moderate size and complexity. Separating the reporting and transactional engines is the only way to avoid this clash while providing near real-time reporting by allowing fast incremental refreshes of the data warehouse during the day. Simple reporting from one or two ledgers may only require a data-scraping solution such as have been referred to in other postings, but as soon as inventory and multiple currencies, for example, are introduced such solutions show their innate weaknesses.
Most BI projects are a goldmine for consultants and service providers who claim to have the magic skills needed to link the thousands of tables needed to provide a data warehouse and the additional skill needed to know where the data resides in these tables in order to produce a report. Their interest is in keeping the end-user tied to their apron-strings and milk them for fees. They peddle easy to create glossy 'dashboards' but fail to explain how many hours of customisation work is required to assemble any screen which involves cross-ledger transactions.
If only there were a solution that addresses such issues. There is, at least for the MS Dynamics ERP systems - PrecisionPoint. Other solutions address other ERP systems, but not in the holistic way that is achieved by PrecisionPoint in the Dynamics world.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
