careers
contact us
Nearly every organisation employs some form of business intelligence these days. However, most organisations still complain that their systems are not working for them the way they would wish.
One of the main complaints is that by the time the data gets to the user it is no longer relevant to the business need. This is particularly the case where there is a wide range of data sources that need to be collated or combined.
Why is this when many organisations have spent millions of dollars implementing data warehouses and creating departments to analyse the data?
The simple answer is that collecting the current data from different sources into a single data warehouse and statically reporting on that data is insufficient for most business needs. Its not that the data is not there it’s that it is not easily accessible or usable by the users who need to make the decisions or take direct actions.
In recent times this has spawned the promotion of advanced search tools to fill the gaps in users requirements. However, even the use of such tools can fall short if the basics of business intelligence have been overlooked. These can be summarised as:
- Capture the right data – common faults are data either captured from wrong point or the data is insufficient to draw an accurate conclusion.
- Identify the agreed source of truth for each data type – alternate sources can be good for quality checking but a single source of truth must be identified.
- Use external data sources to qualify internal data sources – industry wide data helps to qualify trends and anomalies in the data.
- Profile all data activity – a focus on transactional data rather than non-transactional or behavioural data may miss the real source of an issue.
- Identify who the end user(s) are – who will use the data and in what format is it useful.
- Identify what the end user needs to use the data for – is the data to be used as background analysis or as a call to action?
- Identify how the end users are going to attempt to interpret it – what tools are available to manipulate or further analyse the data?
So once you have the basics covered how do you make this under utilised resource really work well for your organisation and your customers?
One way is to make the information available quicker. Since the late nineties a number of organisations have experimented with using “near real-time”1 business intelligence for parts of their overall solution but the results are inconclusive at best, with most organisations merely speeding up the delivery of data into the system so that reports and analysis can then be analysed by the users (“user analysis driven” BI).
Although speeding up the data flow is definitely an advantage it does not necessarily provide a faster input to business decision making. In order to do this some work needs to be done to identify trigger data or an event that can then be alerted to either an individual or embedded into to a business process for action (“process driven” BI).
This often requires some considerable work in formulating hypotheses and defining specific actions or outcomes required including alternative options (should the initial option prove either unsuccessful or inappropriate). Expect this to involve combining both the “near real-time” and the historical data in order to identify more complex responses.
An example of this is supermarkets using RFID data in advanced trolleys combined and historical buying patterns on the store card, then comparing this with products that have recently been put in the trolley to make suggestions for additional items to be purchased to go with the already selected items. The customer can see this as a benefit in reminding them to buy products they may have forgotten and the store has a benefit in that the customer buys more products.
The ‘event driven’ approach outlined above is not just useful for real-time or near real-time data. One bus company collated data from sensors on their buses engines and ancillary equipment with component failure records, hours of service and other historical data to produce better service schedules and predicted component failures. This resulted in better reliability and better service to their customers.
The other thing to remember is that this type of event driven business intelligence requires regular review of the hypotheses in order to account for changes in market or environmental conditions or updates on emerging trends.
It is also important to note that there is still a place for the traditional analytical business intelligence reporting for background, historical trends and more static information as well as its use for combining with real-time data to define events.
In summary the four things to remember are:
- Get the basics right.
- Consider ways of improving the speed at which data is processed through the system.
- Consider combining “real-time” or “near real-time” data by using event driven triggers as calls to action or inputs to business processes.
- Formulate hypothetical action plans to respond to defined events.
In short if you want to really make the best use of your business intelligence just implementing a traditional data warehouse is unlikely to satisfy your business needs for the future. However, incorporating business intelligence into your day to day business processes may not only make better use of your data but has the potential to improve your customers’ experience.
Julian Donohue is a Practice Leader at the Customer Experience Company a management consulting company specialising in improving clients' returns from their sales, service and marketing investments.
1 NB: Whilst true real-time data may be necessary for some processes such as rolling stock breakdowns in railways, in many cases near real-time or just more timely data will suffice.