What constitutes a top company? What are the magic ingredients that differentiate the great from the merely good? That is what we try to discover in the annual Top Companies survey. And as regular readers will note, the process has been evolutionary.

Two of the basic ingredients are a five-year earnings track record and a reasonable size - so that investors can be confident that an operation is sustainable and has critical mass. And we also try to add our own insight into the process, rather than relying purely on numbers.

A quantitative analysis is essential but it can throw up some unwanted anomalies, such as the infamous Tigon of a few years back, for example. That is why we try to add value by drawing on our pool of highly experienced in-house writers and, where necessary, drawing on other talent as well.

The quantitative financial performance is measured, as always, by BFA McGregor, which aims to eliminate inconsistencies in the way that companies report their financial performance by standardising treatment of the figures (see "Definitions").

BFA McGregor has an excellent pedigree in providing this information, and the association with the Bureau of Financial Analysis at the University of Pretoria goes back over a quarter of a century. Their methods are rigorous - on the odd occasion that we think we have found a loophole or an anomaly, a quick telephone call invariably puts us at rest.

Of paramount importance in the quantitative analysis is identification of companies that have a consistent track record - we're not interested in one-hit wonders. And, in line with Warren Buffett's view that "our favourite holding period is forever", we concentrate on long-term performance. It was also Buffett who said that if you wouldn't be happy, for whatever reason, to be holding a share in 10 years' time, you shouldn't buy it.

About one-third of the overall score (quantitative and qualitative) is derived from a five-year assessment of internal rate of return (IRR) and compound growth in earnings per share. In this way, we incorporate factors that are within management's control (earnings per share) as well as those that are more market-derived. IRR measures a company on its returns to shareholders - both dividend flows and capital appreciation in the share price, also known as total shareholder return.

Certain large market-cap companies like Exxaro and Kumba have been listed on the JSE for less than five years in their present form, because of recent important corporate activity. Though they are undoubtedly worthy candidates for inclusion on the basis of performance, the five-year criterion excludes them for the time being. It is also worth noting that Telkom only became eligible for inclusion this year, having been listed in 2003.

We omit companies with a market capitalisation of less than R1bn. This is lower than the JSE's own definition of small cap, but we feel that it's a reasonable hurdle to apply. The rationale is that we seek to ensure that we consider as potential blue chips only companies that have the mass to be real investment prospects.

The selection process for the top is iterative. Firstly, we filter out the smaller companies of less than R1bn market cap, and those that don't have a five-year earnings track record. Using the rest as the universe of companies, we rank them on their past financial performance, using a combination of return on equity in the latest year, IRR in share price over five years and EPS growth over five years. This reduces the list to a top 40.

That list is then subjected to qualitative assessment by the FM. Factors included here are how the company is managed (and is likely to continue to be managed); its corporate governance procedures and culture; the quality of its communication with shareholders and stakeholders; and, crucially, its "investability" in terms of present value and future prospects, and whether the share is tightly held.