1. Speed alone does not add business value: What is the business
value in accelerating a query that currently takes 715 seconds to run and now
runs in 1 second- if business will never use it? (Gartner’s BI nightmare)
2. Clean your BI before
moving to big-data or HANA: There is a huge opportunity to clean your current
BW environments prior to moving to BW-on-HANA. Your Si should be able to
guarantee a minimum of 40% of db reduction. This will reduce your initial HANA
costs by 40% and also reduce your annual support SLA’s by an equal 40% of more
3. Data alone does not generate business value: What is the business
value if you have petabytes of data but little information that drives business
benefits à OUR FOCUS IN THIS BLOG
The traditional
business intelligence technocratic methodology believed that if we collect all the
data then we should be able to meet any information needs. This is how we have
built our SAP BW environments, and other BI environments over the last two
decades. Paradoxically, what our business value audits empirically prove is
that this methodology actually makes it harder to find business value. Sometime
it simply makes it totally impossible.
Moving forward
companies need to focus on business needs and expectations and not the data.
This kind of turns around the legacy methodology on its head but this is now
proving to be core implementation best practice as we redefine best practices
for BW on HANA migrations. Companies need to focus on decisions and information
flow that assists business take better decisions. To use the space age analogy no
one would build a rocket and blast it off with astronauts without deciding
whether the rocket was going to the moon or Mars. Nor would be send a cruise ship
out, loaded with passengers, but without having a clear destination and route. However,
we continue to build our BI environments exactly like that. The only exception to that is Christopher
Columbus who was actually planning to go to India but ran ashore in the
Americas – that is how the Native Americans probably got to be called as Red
Indians. So the key here is ‘Plan your work and only then work your plan.
So the RIGHT ROADMAP
TO BoH MIGRATION of Big-Data analytics is:
1. Identify the business benefits
and value in the form of what information consumers need and in what format =
Business Expectations
2. Identify the data that
is required to meet those ‘Business Expectations’
3. Collect the required
data
4. Architect, Model,
build and deliver ‘Meet Business Expectations’ Analytics
Before we do anything
in the new Big-Data nexus, it is most critical to firstly not get distracted by
the size of the databases out there. Secondly it is critical to let business
define exactly what their needs are. We need to set the stage for extracting
business value from the data and not loading our new environment with redundant
data elements. Rather than collecting data first we need to work backwards from
decision enablement and business benefits and then identify the data required
to support those decision analytics. It is no longer beneficial to take a
technocratic approach of collecting data and just producing interesting
information – hoping business can use it.
Let’s look at a live
business case of a telecommunications company- Goal decrease customer churn
Their legacy reports, ran once a month, provided
customer churn data, i.e. information of customers that had dropped their
services and possibly gone to another provider. This report was a monthly report. These
reports are reactionary.
With access to HANA
the customer developed analytics that predicted customers that might
drop their accounts in the next 15 days. Although, this model was accurate to
around 80% the 15 days the reaction time was not adequate to retain customers. With
the level 1 analytics customer retention increased by a mere 12%. These
analytics though better than their prior reports still did not have any major impact on attrition rates. However, it was the right approach to a business solution....
For level 2 analytics
the first goal was to provide at least a 60 day prediction based on empirical reasons
as to why customers dropped them as a provider. Their Business Value architect recommended they talk to select store managers and business stakeholders. These
were contacted to identify the main reasons customers were dropping. Within
three weeks the front line managers identified 5 reasons that made customers
drop and change to another provider. The first reason was dropped calls when
coverage dropped from 4G to 3G. The second reason was customers being locked to
current phone models by contract, and so on and so forth. It was also noticed that
most of these customers contacted their stores with these concerns, and texted
to their friends for alternative solutions. In most cases the stores had no way to collect this customer issue, not communicate it to the company. So the complaint went untracked and with no response other that that's how we do business. In a matter of 6 weeks they
identified the need for a system whereby store managers would be able to track these specific 5
attributes when their users came into their store. They also sent out emails
and texts to all their users asking them if they faced any of these issues. Thirdly
they started tracking texts on these 5 attributes form their own devices.
By identifying
customer churn reasons, the enterprise proceeded to define clear KPI's that needed to be tracked by the business stakeholders. These new KPI's now enabled sales and marketing to now predict churn by 90-120 days.
Using the new data the analysts could now predict with higher certainty customers who could not only be leaving the company but the exact reasons they would be doing this for. This led to marketing developing specific solutions that were then communicated to their help-line support. The new model not only predict
but also prescribed the exact answers the help-line was supposed to respond with along with solutions tailored for specific issues customers were facing. Using this prescriptive analytics, based on predictive information the company was able to decrease churn by 82%. Due to these new initiatives they not only decreased churn but were able to launch marketing promotions that let them gain new customers to the tune of 13%. The
company also launched Trade promotions and marketing solutions based on
specific customer issues when they reached a certain threshold, for example
free upgrade to newer models without contractual penalties. The second was that the company would take all the transfer burden from competitors.
This approach differs from the traditional data
collection and technocratic approach of IT- collect data --> build reports --> give it to business, i.e. creating reports in assumption of
the business benefits and without direct business participation. This recommendation represents a 180
degree turnaround. Find business needs --> Define deliverables --> Now collect required data --> Build analytics --> Meet business expectations. Now instead of facing data we face business users and
consumers.
Moving forward
companies that continue to take technocratic decisions that are data facing
will end up create more data and not necessarily more information. They will not be
able to enhance the enterprise decision capabilities.
However, companies
that hold their technocrats resources and turn them 180 degree around to face their
business users to understand their needs, with a collaborative approach and business goal, will truly add business value, enhance business
decisions and deliver higher business benefits. They will consistently leave competition
breathing dust as they speed forward and most important of all save considerably
in cost of deployments by optimizing their assets utilization.
Excellent Article - agree 100%
ReplyDelete