• The retrieval of results is much faster. Even with volumes of more than a million addresses the response time is usually less than a second.

• Searching with mailRetrieval can be error-tolerant even without wildcards. This is important to mention, because entering wildcards presumes that the user has an idea where a deviation could be!

• Duplicate check can be carried out automatically without any action by the user for every new entry or change. Hence, even without intensive research by the user the genesis of duplicates is automatically prevented as far as possible.

• Certain identification ensues for example when
  - words are interchanged Franks John instead of John Franks
  - words are abbreviated:  J. Franks instead of John Franks
  - the name sounds different from the way it is spelled: Francks instead of Franks
  - there has been a simple data entry error: ranks instead of Franks

• The results are returned according to the degree of agreement. This facilitates operation considerably and spares endless leafing through result lists.

The maximum technical limit is greater than all of the results that occur in practice. Of course, the address database size to be managed also depends on the available performing resources (CPU and hard disk drive and motherboard). In practice, the volume of addresses stored in the database for real time address identification is not so significant, since this can possibly be distributed to several countries. What is important is the volume of managed addresses per country performing without segmentation.

A practical example for this: some users of mailRetrieval manage address databases with more than 60 million redundancy-free consumer addresses per country.

The concrete performance depends significantly on the resources (CPU and hard disk drive and motherboard). Because Uniserv has developed special error-tolerant searching methods for address identification in real time, this provides high performance with low demand on resources. Here is a practical example:

The management of 10 million addresses without segmentation on a rather small Intel server with 1 GHz and 500MB RAM leads to an average response for an error-tolerant search of all name and address fields of less than 0.5 seconds!

In most applications it is customary that when duplicates are suspected, the user who is entering new addresses or changing them receives a message that can either be accepted or rejected. Through faulty operation or, because final clarification at short notice is not possible, even in such systems the unconscious or willingly accepted entry of duplicate or multiple addresses takes place.

You can track these cases down in a periodical mailBatch run and clarify them asynchronously.