A designer of a strong name search will understand both the risk of a missed match and its cost to the business. When designing name search applications, recognize that each data population to be searched may have different risk attributes and costs of failure.

A missed match can be due to human error, because the name search failed to find the record, or because the match was "hidden" in the results set (due to the list being too large, or not in a useful sequence).

A name search, which fails to find a candidate match, either did not cater for some types of error and variation, or did not look exhaustively enough.

The more error and variation that is overcome, and the more exhaustive the search, the greater the potential for finding more true matches. The reality is, finding more real matches increases the amount of work and the cost. It also increases the risk that more false matches will be presented.

The goal of a good name search process is to maximize the true matches while minimizing the false matches. Even after the name search process has been tuned to provide this balance, there will always be the tendency to find more true matches at the expense of introducing more false matches.

In the final analysis, a well-informed decision should establish the cut-off point. If it is decided that no matches are to be missed within the power of the name search, then more human resources will be required to select the true matches from the false. If it is decided that human and machine resources take priority, then the name search can be tuned to deliver to that level.

One of the serious problems of finding too much for an operator to look at, is that the human operator themselves then make poor decisions.

Even good well-trained operators cease to be diligent if they are expected to be searching hour after hour, day after day.

With well designed automated matching it is possible to build systems that mimic the very best human operators looking at all the available data and making decisions that are significantly better than the average human operator can achieve.