Even if you think the phrase "lies, damned lies and statistics" is a meaningless platitude, don't mention it to John Sall. A co-founder of business intelligence software company SAS, and now head of its JMP business unit, the softly spoken American is passionate about how statistical data can be harnessed to improve business processes.

"We help people innovate by exploring all their opportunities in an efficient way," says Sall. "Once you collect the data, we help you analyse it and find the patterns."

JMP's software has two main applications - running controlled experiments and analysing statistical data derived from industrial processes, including experiments.

"The analysis part is all about looking for processes that go out of control and seeing what patterns or defects are associated with that," explains Sall. He gives the example of a printer who finds that they are suffering a problem with banding - lines where no ink appears across an image or text on the printed page.

"That could be a function of factors like ink temperature, paper type and the various interactions between them," he explains.

The printer could solve the problem by tampering with the different variables until he hit upon a solution, but it would be highly inefficient.

"Experimentation is basically trial and error," says Sall. "Changing one factor at a time is very inefficient because you are only looking along one dimension."

The semi-conductor industry has embraced these techniques to get its production processes to the highest level possible. In the early days of the industry, back in the 1960s and 1970s, the defect rates for chips were extremely high.

"How many transistors fail on a chip now?" he asks rhetorically. "If you consider that there are thousands of transistors on a chip, they operate literally billions of times without fail."

Although high-tech industries may have been the first to approach statistical analysis and the design of experiments in a structured manner, Sall believes that all industries can benefit from the application of statistical techniques. Sall is also unperturbed by the threat posed by low-cost economies to the manufacturing base in locations such as Ireland.

"The drivers of efficiency, most of them are technology-led," he says. "Low labour costs may help you make some things easier. But take computer chips as an example, that is driven by the technology that goes into them. Developing the technology of production and the manufacturing process is key. If you are going to lead, that is where your leverage is. It's the key to competitiveness."

If JMP is not a household name, it's no surprise. It's a subsidiary of SAS, the supplier of business intelligence software that prides itself on being the world's largest privately owned software company.

SAS has a variety of top tier customers in Ireland, including several government agencies such as the Central Statistics Office, the Revenue Commissioners, the Department of Social and Family Affairs and the Department of Health and Children.

In many cases though, its private sector customers don't want their names mentioned in public as they feel it could compromise any competitive advantage they get from using the software.

It's a similar situation with JMP. Sall tells how Procter & Gamble was a long-term customer but didn't want that publicly revealed as it was using the software to tackle production line reliability issues. However, it was so successful at putting its own house in order that it has partnered with consultants Bearing Point and now sells its own expertise in improving production processes.

Sall co-founded SAS in 1976 with Jim Goodnight and two others who subsequently left the firm. Goodnight is still chief executive, while Sall had a variety of senior positions in the company before heading up the statistical instruments division of the company, now called JMP Software.

SAS's software helps large corporations to spot patterns in the huge amount of data stored in corporate databases. Applications of the technology range from combating money laundering to drug development.

Although the core ideas are similar, Sall says that SAS is about enterprise systems and processing vast quantities of data, while JMP is a smaller tool employed by individual engineers at the project level. "Think of JMP as a spreadsheet for data analysis while SAS is a database for data analysis," says Sall.

Sall was the principal designer and developer of the JMP package, originally released for Apple Macintosh computers back in 1989. Until now, it was primarily sold and marketed in the US and Japanese markets, but SAS has decided to put a concerted push on it in Europe.

Despite the low level of investment, there is a surprisingly large user base for the software, with 200 companies using it locally including tech giants Intel, Dell and HP. Sall will be in Dublin next week to speak to Irish JMP users and showcase the latest version of its software - JMP 6.

It features a range of improvements around the design of experiments and data mining, but the innovation Sall is most proud of is the software's support for split lot design experiments.

"When you have an experiment, some factors are easier to change than others," explains Sall. "To change one, you may have to disassemble a machine and switch something internally, which could take the best part of a day. Another factor may just require you to turn a knob.

"That imposes a desirable structure on the design of the experiment - some things you will not want to change very often."