markets 
financial services
logistics
and transport
fixed line
telecoms
mobile
telecoms
retail
human
resources
Analytics Professionals
Information Inventor and QueryObject offer a unique proposition to Analytics professionals, delivering true terabyte-class scalability combined with ultra-fast modeling configuration. In today's high pressure environment, our solutions help you to respond quickly to new and changing business requirements.
In essence, Information Inventor offers a number of advantages to the Analytics
Professional. It's been designed to help you deliver robust models in short
order, where the underlying amount of DB-level tedium is much reduced. As
an example, with Information Inventor for Churn Prediction,
you don't need to locate and feed sets of past churners and non-churners into
the model; Information Inventor calculates these automatically.
Further, Information
Inventor uses a proprietary algorithm unlicensed by any other product, designed
to show higher predictive accuracy from the same data sources, and geared
to deliver extremely low False Positives. What does this mean ? Greater profitability
for your business as more accounts are defended, and more "good responders"
revealed.
And with Information Inventor for Automated Business Intelligence dataset
abnormalities are discovered automatically, and pushed out as events. This intelligence
often leads to the formation of new projects and ultimately, refined or new
business process. Our products
deliver a clear Return on Investment message, calculated whilst processing
rated customer transactions - helping to underpin a project business
case.
The fractal compression offered by QueryObject means regular
products, like Business Objects, Brio and even Excel can suddenly start to
work a lot faster,
and it's
this technology that we can optionally use to underpin really high volume
analytics projects. In addition, QueryObject can connect
up all of your discrete corporate data sources into a single holistic view,
for easy data collection.
Example Project
Company Profile "before Information Inventor"
sector: financial services
accounts: 40,000
monthly churn rate: 8% of accounts
monthly lost value: £5,000,000
with Information Inventor for Churn Prediction
project duration: 1 month
days of effort internal (IS/IT, Analysts and Business) : 12
days of effort external (Oceans Blue) : 8
overall classification accuracy: 99.8%
True Positives to False Positives ratio (expressed as percentage): 82%
Information Inventor predicted monthly return: £4,000,000
Implementing Business Intelligence ? Often, you'll be picked to lead new projects in this space; we hope you find our tips below of interest and check out the Business Purchasing link for product evaluation criteria.
1. Communicate clearly
Try to define the specification for the project in business terms, don't
over use technological buzzwords in an attempt to demonstrate your intellectual
superiority. You will be appreciated most through project success, and to
increase the chances of this, tailor how you communicate to the technical skill
levels
of
the recipients.
2. Think business benefits
Get some clear project goals defined - if you have the resources and skills,
try to deliver the project under the aegis of a formal project management
method (see our project management quicklink for ideas),
and get stakeholders together and engaged with project. Remember, this project
is not an excuse to play with technology, it must deliver measurable benefit
to be successful.
3. Deliver an ROI
To demonstrate success, you'll need to measure the impact of the intelligence
you can deliver. This must be in actionable form to be of value. Try to
present your results in terms of a Return on Investment for the business,
especially when presenting results to your business sponsor - someone who's
less likely to be impressed with your superb outlier detection rates or
low false positives/true positives gearing.
3. Actively seek to engage IS/IT
Whether the project is to be hosted internally
or outsourced, these guys are often the last to know what's going on. Get
them engaged with the project early on. At some level you'll probably need
their help, so it's a smart to be on their good side - right ?
4. Put together a risk assessment
You or your project manager need to think
seriously about identifying risks up front - and derisking these with a pilot
solution. Here, you can get some results quickly, with little expenditure
to help to
prove
your case - reputable vendors should be prepared to help you achieve this.
Other common issues are might be : do you have enough internal
resource ? How much funding is required ? Is there a suitable test machine
? What if you screw up the production systems ? Which projects need to be
deferred to action this one ? You get the picture. It's important to present
these risks up front to the project board, not as a "get out of jail card",
but so the whole team can contingency plan effectively.
5. Stage control your project
Build some contingency time into the plan, and measure your progress through
time bound stages and goals. If you don't do this, how will you, or the
project team know if things are going well ? You can review your progress
versus
the plan each week.
6. Don't over-complicate the admin
Ok, for larger projects you'll need to stick quite rigidly to your methodology (be it PRINCE 2 or whatever), but for smaller projects, why not deliver to a subset of your method ?
7. Action your insight
Unactioned insight is
an academic exercise that delivers no real benefit, so make sure your project
plan doesn't end when your graphs and charts appear. Somehow you'll need to
find a way to connect your intelligence with existing core business applications,
be they CRM, Autodialer or Billing systems.
Product Comparison Table
| Information Inventor | Competitive Products | |
| low requirement for internal IS resource to support project | yes | no |
| able to pull in data from all discrete company databases | yes | no |
| intelligent, wizard-like approach to configuration | yes | no |
| model is self-learning, so cost of ownership is low | yes | no |
| highest accuracy, lowest "false positives" gearing | yes | no |
| plug in modules covering all facets of business intelligence | yes | no |
markets
Finding a new customer costs from three to seven times more than keeping an existing one, and for many large companies, up to 95% of profits come from long-term customers. Pricewaterhouse Coopers.