Mobile computers were installed in vehicles to allow incident reports and other administrative tasks to be completed on the road. This has numerous benefits, including immediate filing of reports, which reduces errors and omissions, as well as allowing the maintenance of a presence in the field. These mobile units are costly to purchase and install, so it was important to ensure that they were used effectively.

To that end, a performance metric was introduced to measure the number of reports filed using these mobile devices versus those filed on desktop computers. We captured this top-line performance with the first generation question. As results began to come in there was a large difference in the level of use between groups which raised second generation questions as to why this disparity existed.

Each device could identify both the user which when linked with HR records enabled their job descriptions to be generated. It also generated locations for each user which enabled us to plot them on a Google map.

The Performance Team were surprised to note that over 80% of the reports filed using the mobile devices were done so from the base or the adjoining car park. In addition, the majority of reports were filed by users who were identified as administrative staff, i.e. staff who were permanently based at the base and in reality, only 4% of the reports filed were done so on the road. It became apparent that Mobile terminals had been issued to administrative staff to use, at a far greater cost than normal desktop terminals, to enable targets to be achieved.

Without this context, the project would have appeared successful whereas in reality equipment was under-utilised and not used as intended thus failing to achieve their ultimate aims whilst increasing costs.

A large NHS Trust was failing to meet a national target for Zero Length of Stay. Zero Length of Stay is when people are admitted into a hospital bed, usually via an Accident & Emergency department, only to be released within 24 hours without any treatment, because there was nothing wrong with them. To reduce the numbers of people falling into this category required looking for the reasons why and how people are admitted.

We looked at admissions for patterns in the symptoms. We looked for commonalities: Were the same staff responsible for these types of admissions? Were the circumstances in which the patients were admitted the same? . We found little to suggest any pattern in the staff or symptoms, and on the surface, the admissions appeared quite random and offered us no leverage over the issue.

Digging further amongst the things we looked at was the time of arrival and time of admission, which is when we spotted a trend. The majority of patients who were admitted into a hospital bed, and released without treatment within 24 hours, were admitted after waiting in A&E for between 3½ and 4 hours.

At the time there was another national target which required patients who attended Accident & Emergency departments to have been seen and discharged within 4 hours and it became apparent that in an attempt to hit this target a 'work around' was being employed that adversely affected the zero length of stay target. This was important as the focus for improvement was able to be directed to the correct cause, the performance in A&E and a source of poor patient experience and high cost addressed.

This demonstrates how the ability to ask successive generations of questions, model and visualise them, is essential to being able to interrogate data it until an answer is found. Traditional solutions using data warehouse formats are unable to do this since with each answer the focus shifts and new questions arise that could not have been thought of when the Data Warehouse was designed.

In many instances, because targets change after the Data Warehouse based system was designed and built, the questions weren't even known or relevant.

A large Primary Care Trust was reviewing activity of its District Nurses and found that a significant amount of home visits were being conducted by senior nurses and in an attempt to reduce costs it was proposed that these visits should be conducted by junior nurses whose salaries are lower. As part of this process it was necessary to plan and forecast the likely workload for each team to enable the right number of junior nurses to be recruited.

To answer these questions it was first necessary to build up a multi-dimensional picture of the activity taking place, how long it took and who was doing it. To do this we linked data available in different information systems including: Activity, HR and Payroll. One of the outputs that we generated was a calculation of an average cost per visit by job description.

The surprising result was that, according to the data, the average cost of a junior nurse completing a visit was twice that of a senior nurse! This obviously ran contrary to both intuition, was in direct contradiction of the received wisdom and the proposed approach of reducing costs by increasing junior nurse numbers. The Trust asked us to investigate further.

We linked into more data, connecting Training and Timesheet data, to fully apportion 100% of each nurse's time, to include behaviour, training, annual leave and sickness. What we uncovered was that due to their lower level of skill and capability junior nurses were often visiting patients in pairs to cases where a senior nurse would have attended alone, and they were taking an average of 30% longer to complete the activity than their more senior colleagues. In addition, junior nurses spent significantly more time training, and therefore spent less of their available work time delivering results.

The outcome, with all these factors taken into account, was that the recruitment of additional junior nurses was put on hold and the focus switched to forecast the recruitment requirements for senior nurses.

By being able to join up data held on incompatible systems, a multi-dimensional picture can be generated that allows simple powerful metrics that can be used to check perceptions and beliefs and often deliver new insights that allow better decisions to be taken.

We work with a UK based business who manufacture and ship goods around the world.

As with many businesses they have various different systems gathering and producing information on the essential elements of their business especially sales, manufacturing, logistics and finance. The products they supply are part of the manufacturing chain, and so the logistics element of the business is essential to satisfying the customer's just-in-time production requirements and to keep their own delivery and warehousing costs to a minimum.

The basis of the order system is similar to many we have seen. When an order is placed customers are given a date by when the order will be delivered. For efficiency, it was practice to ship parts of an order early, if possible, by placing it on a vehicle already travelling to a nearby destination, that had space and to load trailers the night before, ready for departure at the earliest moment the next day.

In a desire to make improvements, the client had invested in a system that placed barcodes on each item. This enabled monitoring of the chain from manufacture to delivery, measurement of orders delivered on time and the ability for customers to track their orders through a web site. Over the next few months, the system showed that the number of orders not delivered on time had risen significantly. It was immediately evident to the management that numbers were not a true reflection of reality and so there must be something wrong with the new system.

What came to light was that the system was unable to cope with the practices of loading the night before and part shipping and, unless the order was shipped in full on the target dispatch date, orders were recorded as late or failed. Management then took 2 actions. The first was to approach the system providers to change the way the system was working. Their response was that they had provided what was asked for. Ripping out the system wasn't an option as, aside from cost, the order tracking system provided to the customers would then have to be withdrawn. The outcome was that a redesign of the system was started to attempt to fix the issue, the weekly cost of which was large and the timescale expansive.

Meanwhile management informed staff of the issues and explained how the system was working and that the numbers couldn't be trusted. However the reaction of staff was to change their behaviour to the new context the system had created.

Firstly, when orders were loaded the night before, customers were receiving updates which suggested that their orders had been shipped. To stop, this preloading was stopped and vehicles were loaded the same morning. The effect was that vehicles left depot much later and could therefore make fewer deliveries in a single day.

Secondly, shipping part orders was stopped and as a result, vehicles were leaving depots with unused capacity, increasing the number of journeys being driven, reducing vehicle utilisation and increasing the amount of space required and stock time in the warehouse, until the entire order was ready to be shipped in full. As a consequence logistics and warehousing costs rose significantly.

When we found out what was happening we were able to offer a new option that saved the customer money and created a system that worked in a short timescale. As a consequence management stopped the original providers from trying to fix the new system and let us loose.

Our solution was to use our know how and software to draw data from all the systems to create a picture of any order compiled from input from each system. This enabled us to place an umbrella system over all the data, apply new rules to the data and to use this as the source of the feed of information for the customer tracking system and management reports. Within a few weeks of starting, accurate information on orders was achieved, the consequence being that cost minimising behaviours were returned to and costs reduced.

Most importantly, customers now have a true and accurate picture of their order status.

A large UK insurance company has a growth strategy of acquiring other companies.

Each acquisition presented 3 main challenges all based around accessing, integrating, processing and reporting large amounts of data, in various formats, from a wide range of systems, containing vital information about the customers and their policies.

• To undertake due dilligence in advance of purchase, the acquisition target needs to be fully understood in terms of the size, nature and value of the business.
• The new business needed to be integrated after purchase into the existing at the lowest possible cost to enable reporting and managment.
• To be able to report to the FSA in an accurate and timely way.