Milking Oracle for all it's worth

By ADAM GIFFORD

Some day towards the end of October, Fonterra manufacturing operations manager Max Parkin knows his 25 factories will process upwards of 70 million litres of milk.

That is what the dairy industry calls the "flush" - the peak in production that comes as spring turns into summer.

At last year's flush, those factories were working at more than 90 per cent capacity.

There is not a lot of headroom. The milk has to go to the right factory and be turned into the form that will give the best return.

Over the year the company will turn 13 billion litres of milk into 2 million tonnes of products, worth up to $10 billion and accounting for almost a quarter of the country's export revenue.

Most of that is from commodity items, so keeping manufacturing costs as low as possible makes a significant impact on the returns to farmer shareholders.

To help Parkin and his managers achieve that, Fonterra has built an Oracle data warehouse which measures more than 25 key performance indicators covering all facets of the business, from milk production to the yields at the other end.

How well individual factories are doing, the reliability and throughput of the manufacturing processes, legal compliance and even workplace accidents are visible at a click of a mouse.

The warehouse, fed by a nightly data transfer from the production database, as well as inputs from more than 20 other systems, builds on an earlier data warehouse built by the former New Zealand Dairy Group.

Oracle Portal was added to the front end, to give more than 100 managers and supervisors ready access to the system rather than having to get IT staff to generate reports.

"Through the portal, we can link direct to the factories, so we can graphically display volume of milk by the minute, so know at any time if we need to send trucks elsewhere," Parkin said.

When the Dairy Group Oracle applications were rolled into the other legacy company, Kiwi Dairies, a major exercise was undertaken to identify the most useful performance indicators.

"We wanted a single version of the truth," Parkin said.

One criterion was that all data should come from applications, so there was no chance for manual manipulation of data.

Parkin said the reporting structure and improved information flows meant if something went out of kilter, it could be picked up far earlier.

Oracle account manager Gael Stoddart said many organisations struggled to get real value from their data warehouses.

"The way Fonterra is getting data out and using it is the best we have seen."