[Blog post] David Gee offers his practical tips.
Big data. It’s the hype-fuelled juggernaut that few
CIOs can afford to ignore for any longer. But jumping on the bandwagon
is not going to be easy. Here are some practical ways to make sure your
organisation is ready to make the most of the deluge.
Consider a study tour
Alan Grogan, partner at Advanced Capability Solutions in London, says
that letting third parties in on your big data plans can be vital “as
this will help you later on if you need their buy-in and support with a
project”.
But he adds a cautionary note.
“IT alone won’t be able to help you with commercialisation of your
data into information and then business knowledge,” he said. For this
you will have to do some personal study and network heavily to get
detail on user cases both in your sector and across other industries.
One way to accelerate this learning process is to seek out some
expert guidance. An option is to undertake a ‘field trip’ to a handful
of leading-edge analytics organisations based out of India and/or the
USA.
Sri Annaswamy, founder at advisory firm Swamy & Associates, endorses this approach
“There are a number of options, with varying degrees of usefulness.
You can buy a standard industry analyst report (Gartner, HF etc.) or a
white-paper from a standard consulting firm (KPMG, Deloitte etc.) But
neither option would provide the ‘hands on’ learning associated with
being on an analytics factory floor during a field trip” he said.
Another learning to pursue is getting an idea of what an onshore and
offshore analytics centre of excellence will feel like and indeed how
this could work for your own enterprise. There is nothing like seeing an
offshore centre in action and getting a sense for the talent pool that
is in place.
The degree to which your business will consider ‘partnering versus
owning’ is a key question. But certainly does not have to be answered
immediately. This can definitely wait as you and your team learn by
observation and also by osmosis.
Learn by outsourcing
Ok, so you want to learn more about this strategic topic – big data.
But you’re also aware that this skillset is in short supply and that
no-one wants to expose their thinking or share information at this
stage.
For those of us that are true-believers but lack the requisite deep
experience, then the best way to learn is to consider outsourcing big
data functions. From my own experience, I used this approach many years
ago in another domain as I was establishing a customer contact centre
for Asia Pacific.
This may sound counter intuitive, but it is actually the fastest way
to learn what best practice looks like. From this position you can then
decide whether you want to grow your capability internally or continue
to outsource.
There is a saying that when the processes are a mess then you will be
outsourcing a mess. It is true, but you can also learn from working
with great partners.
Picking my partner?
For those of us that have been around a few seasons there are some
common sense criteria for selecting a suitable partner. In essence you
want to be buying high quality data analysts who have done this before
in industries that you rate as being leaders of this new category.
Grogan has narrowed his advice down to four tips:
- For your specialist areas and projects you should have dedicated –
ideally named – people who support only your business. This is not just
to ensure analysts who ‘get’ your business, but also to protect any
intellectual property.
- Be careful that the vendor lets you exclusively own any new
processes they develop for you. This may sound obvious, but I’ve heard
of companies effectively signing away their business models.
- Be careful of vendors using very fancy phrases, promising the latest
technologies and overstating their capabilities. Be mindful that it’s
known for consultancy and technology companies to ‘wash’ their
capabilities with other more basic capabilities, and then re-outsource
the difficult stuff.
- Contact any potential vendor’s current big data clients and ask how their needs and wants have been supported.
Who should ‘own’ big data?
As a CIO, should big data even be your responsibility?
Alan Grogan says the way to approach this question is to forget about
‘ownership’ and ‘territory’.
For big data to succeed there needs to be
a strong focus on the customer and an understanding that the culture of
the organisation will evolve with this transformation.
“Big data ideally needs to be sponsored by a CEO (local, regional or global) who is the custodian of culture.
“In terms of driving the agenda, the owning technology area
(CTO/CDO/CIO) should really engage with each business area and function
that stands to benefit. Over time, as big data processes become more
embedded in the organisation, it’s not unimaginable that ownership will
split between functional leaders (head of product, head of sales, head
of finance) etc.”
He recommends that appointing a joint sponsor is the best approach,
within the function or area which will fund the project. It’s critically
important that this strategically positioned executive straddles the
entire organisation and reports to the CEO.
Self-service philosophy
Big data demands self-service from day one. This is what will drive
success. It is a long way from the traditional data warehouse and data
mart approach and IT will have to evolve to work within this paradigm.
Let’s remember that the single-most important purpose of big data and
analytics in an organisation is to enable end-users to go directly to
the data source and carry out the analysis and visualisation they
require directly, without relying on an IT-department-based BI and
reporting team to prepare reports for them.
This approach shouldn't result in a lack of governance but it is
radically different from how it was approached in the past. In effect,
it means those of us in IT will have to let go of a little control.
One proof-of-concept is not enough
Cloud options have lowered the barriers to making a start with big
data. But it is also important not to rush into anything that hasn’t
been thoroughly tested.
The concept of testing big data through a proof of concept is clearly the recommended path.
In fact Annaswamy recommends enterprises don’t just run one POC “but
three to five POCs in parallel, utilising different potential strategic
analytics partners”.
He explained that this “helps accelerate proof of the value of big
data and analytics to the executive and in a tangible manner as well as
uncovering the specific capabilities of different service-providers in a
defined timeframe and cost bound manner”.
What changes to the organisation are required?
One can argue that big data is naturally already the job of the CIO
or CTO – so why would you hire a chief data officer or chief analytics
officer? I remember the rise and decline of chief knowledge officers
that have come and since gone from the ranks of corporations.
My advice is: let’s not let this happen again, so IT, including the CIO, have to embrace this change.
There is still a need for an appropriate framework – with end-to-end
data stewardship to drive data quality and make sure all data falls
within a governance, risk and compliance framework. The most important
thing is to get executives to grasp the importance of big data and
analytics for every part of the company’s business.
Business gap analysis
A gap analysis is not a business case. Instead it is designed to
illustrate how real the gap between your organisation and the industry
leader(s) is – and how to bridge that gap.
Annaswamy maintains that the length of the review depends on the type
of analytics project. In the case of a simple cross-sell uplift or an
attrition problem project then it could be within 12 months.
However if it this is a complete revamp of your analytics platform,
to say an enterprise data warehouse with the incorporation of a
Hadoop-based system for unstructured data plus visualisation and newer
software, you could be looking at a three-plus year scenario.
Copy cat – who should I benchmark?
Benchmarking against peers is the trickiest aspect because anyone
that is doing this well is not going to share their intimate details.
Recently I was speaking to a CIO in Singapore who worked in an
industry completely removed from my base in financial services. Despite
my best attempts I could not convince him to divulge big data specifics.
But one hint is to look for organisations that are customer focussed,
have made significant investments in this area and have appointed chief
data or analytics officers.
In Australia it is understood that Suncorp and CBA are the most
advanced in this space, but watch out for Walmart, AIG, Amazon, Google
and Pay Pal.
Reskilling my team
I’ve been really curious about what skills are really needed to
manage an outsourced big data capability. The most important part is
managing the diversity of staff – standard IT developers, programmers
and BI techies won’t do!
You will need engineers with strong statistical-modelling experience,
behavioural scientists to pick out the insights and operations people
to plug the insights into specific interventions.
There is an assumption of course that vendor management today is a
given. Most organisations have been through the IT offshoring/BPO cycle
for several years now that they have reasonably sophisticated
procurement and vendor management teams.
Reinforcing this view was a further insight from Grogan that
relationship and listening skills are all key. The biggest challenge is
simply getting the outsourced capability stood up.
