Business Intelligence During the Recession- Cutting Cost But Still Getting Results

It makes sense that knowing what's going on inBusiness Centered Projects
your business will help you make better decisions.Now some of my former colleagues in IT will be
This obvious bit of news has been held up asbe making cautionary noises on the next
proof of the validity of "Big Business Intelligence"category- Business driven business intelligence
for years. "It costs what it costs" say the bigprojects. While it's true that letting the business
business intelligence software vendors "you can'tbuild databases and servers without IT
afford to fly blind". Well, its true that flying blind isinvolvement is a good strategy for pain and
a bad idea. But worse is to fly blind after spendingsuffering all round, the world of information is not
a few million dollars on a data warehouse or dataonly about databases and servers.
mart project. Better be blind and keep the moneyThe goal is to improve your businesses ability to
in that case, if you ask me.take action based on factual information- this
In its predictions for business intelligence for 2009most likely will involve computers and software-
Gartner states that "Although IT organisationsbut don't ignore activities that are less technical,
excel at building BI infrastructure, business usersbut get you towards that goal. One of the key
have lost confidence in the ability of them tothings that hampers your ability to get data is
deliver the information they need to makeinconsistent data definitions. These definitions
decisions. ". They further predict that at leastoften have nothing to do with any IT system,
20% of business intelligence budgets will benor the IT department. Simply having the business
controlled by the business rather than informationclean up that product naming rule, more clearly
technology departments (IT) by 2012.define the cost accounting structures or give all
Ok, so how to best allocate your budget?their administrative people standardized training on
how to categorize expenses are examples of
1. Have a plan.first steps that will in fact improve your ability to
2. Make it balanced- from small to big.compare and map data. In my experience this
3. Keep focused on the goal not the technology.aspect is often underestimated. You can spend
To have a plan, make a list of every potentialmillions on a Master Data Management system,
initiative and project with an aim to have abut often the first thing to do is just to get the
diverse set of activities/developments thatright people to fix some of the worst issues in
optimize delivery and cost. Not just thethe existing systems.
technology projects- everything. Remember thatConsider this- it's possible to imagine a "successful"
the prize is not more servers, or nicer reports, ordata warehouse project where millions of dollars
having the biggest, fastest data warehouse ongo into creating the data warehouse and
the block. It's making better decisions using betterinfrastructure, but the actual ability to compare
data.data across regions was due to the work done
First of all, have a wide range of different sizedby your own business people to standardize
initiatives. Don't put all your eggs in a single datadefinitions as part of that project.
warehouse mega project basket. Mega projectsOnce the definitions are clear, maybe you can do
are difficult, expensive and risky. Wait till theat least some of the comparisons in a more cost
economy is overheating again to get on that bus.effective tool- and keep the millions you spent on
On the other hand, don't imagine that lettingservers and consultants.
everyone get on with it in Excel will solve all yourSelf serve data- empower the analyst
issues either.And finally, Don't neglect the mini and micro
To help ensure that you have balance, makeprojects, and giving support to the individuals
three categories of projects and try to havedoing them. People are smart. Give them the
multiple projects in each one;capabilities and tools to get at data and
- Projects with IT heavily involvedunderstand it and they can do impressive things.
- Projects that are done by the Business notTraining for example; we all know how much time
involving new technologyyour analysts spend in Excel making reports. Have
- Use of Self Serve and helping individual analystsyou ever considered giving them access to some
do it themselvesspecialized Excel training to make them more
Often when we think about business intelligenceeffective? Not a run of the mill Excel training
projects we just focus on the first category. Thiscourse, but a targeted, analytical skills builder. No
is not only about technology- sure the IT projectsmatter how brilliant your technology and systems
are going to be important, but don't neglect other,are, its likely they'll still spend some hours every
often far less expensive and still effectiveweek in Excel. If they were better at it, they
activities.could get more done, and do it in a more
The Classic IT Projectsstructured way.
So first category- Projects being largely run by ITWhen the data size or complexity does exceed
(and I used to be in the IT department, so I haveeither Excel's capabilities, or your analysts desire
respect for how hard they typically work, and theto do VBA programming, there is a new breed of
challenges they face).tools becoming available that allow users to
Set up limits in your mind, both minimum andaccess data directly in new ways on their
maximum for the amount of time you thinkdesktop.
reasonable before an IT project delivers realThis is my favorite area because its the area that
benefits.my current activities are focused on, building new
Minimums? What? Isn't a short project a goodtools for the micro and self serve capabilities.
idea?The advantages for these types of desktop tools
Sure- but think about it- have a reality check inare;
your process. If IT, or a vendor tells you that- Lower or zero hardware cost because they use
you'll have a full fledged data warehouse or datathe existing analysts PC.
mart in less than a month (and I know vendors- Ability to let analysts stay close to their data
who will do this), you have to ask yourself if itsand understand the results they are getting.
reasonable. Maybe it is- but more often than not- Very fast delivery. Because this is a free-form,
the vendor makes attractive claims that would beflexible environment, the delays from the time
true only providing the data in your system hadconsuming process of requirements specifications,
zero defects and you used the systemtechnical analysis, budget approvals, purchasing,
EXACTLY as the software vendor recommended.installation and commissioning required in a server
When the data quality issues that exist in everybased "Big BI" system just don't happen.
real system appear, and your variousNow, desktop analytics is not suitable for
customizations in, say, the sales system areeverything, and its not going to replace the need
revealed- the BI vendor will say "oh, well, if itto have more traditional business intelligence
were not for the data quality problems etc. wesolutions. Plus, it can suffer from the same sorts
would have had the reports in the time we said."of issues that Excel sheets on their own do- need
Right.for cross checking, and a tendancy to create
Another reason to flag very short IT projects ismultiple versions of the truth.
to challenge the need for a data mart at all. If ITFor this reason, it should be done in partnership
say they can build a data mart in two weeks ofwith your main stream business intelligence
development time, do we really need to buy allsystems and IT department to ensure these
those servers and software licenses? Is a dataissues are addressed or at least not completely
mart overkill for such a trivial data manipulation?ignored. Just as you shouldn't close your books
Who is using this data? Is there a less expensiveeach month in Excel, analysts working in a data
way to do it without increasing the amount ofsandbox like tool need to be aware of its
hardware in our data center? (More on this inlimitations. But if they find trends or spot patterns
category three.)in the data that can then be validated, they may
And maximums? I think that you simply can'tdiscover more than the million dollar warehouse in
accept anything that doesn't have a deliverable bya fraction of the time.
the 3 month point. Period. Now is not the time toAnother thing to consider is this: How much of
launch projects where the first report is a yearthe cost in your previous business intelligence
off. The project can take a year, or even more-projects was due to the features requested by a
but get deliverables quarterly or better, andsmall group of business analysts?
monitor them closely. Be able to ramp down orIf these analysts have the tools to get at this
pull the plug on projects going off the rails.information themselves, you may find that some
Another type of IT project that is often notprojects can be eliminated, and those that stay
considered but should be is a decommissioningmight be simplified, reducing cost and delivery
project. Take a look at the various reports andtime.
databases that you have. Are they all being used?In the current economic environment, no one can
If not, shut them down- free up hardware,afford the business intelligence mega projects of
storage space, CPU cycles, and maybe even ITyesteryear. But with a balanced common sense
maintenance effort. I think a lot of companiesapproach, using proven techniques, practical
would be amazed how much work is doneapproaches and some of the new desktop tools
keeping systems alive when they are no longerbecoming available, the data can flow at a much
needed or used.lower cost.