Geocoding is the process of finding a location (i.e. coordinate) for an address or other place name. The reliability of geocoding typically relies on two things: the accuracy and completeness of the address database; and the quality of the matching algorithm.
MapPoint’s address database is currently updated on an annual basis. Therefore, if you have the latest version of MapPoint, then you have one of the best address geocoding databases in its price category.
The quality of the matching algorithm is more complicated. Although the US has a formal method of formatting street addresses, street names can vary a lot. For example, in the Dallas area, “LBJ Frwy”, “Lyndon B Johnson Freeway”, “I635-E”, “Interstate 635″, and “I-635W” all refer to the same highway. As a further complication, the US has at least three I635 interstate highways (Kansas, Missouri, Texas). As well as freeway abbreviations, cities and states can also complicate matters. Both Los Angeles and Louisiana are commonly referred to as “LA”. A human can usually use context to differentiate between the two, but not always. An automated geocoder never has context. So if MapPoint’s geocoder is having problems finding your street address, try to formalize the formatting and remove any abbreviations.
Most, if not all, consumer geocoders use a process of interpolation. This saves processing time and disk space by only storing the street numbers for the ends of street segments. Individual street numbers are then interpolated between the ends of these segments. As well as efficiency, this also has the advantage that the geocoder can make an intelligent estimate of the location of a new-build street number. However, the interpolation is itself an explicit assumption about address spacing, and also requires an assumption about the positioning of odd-numbered and even-numbered addresses.
Geocoder interpolation is good enough for most applications (e.g. deliveries and route finding), but it is not accurate enough for emergency response or even fine-scale urban planning.
As it stands, MapPoint does not have the ability to “reverse geocode” – i.e. to find the location or street address for a coordinate. If you are only interested in “places” then you could create a pushpin at the coordinate and search for nearby places. An alternative is to use custom programming to perform a search around the coordinate. This can be slow and time consuming.

We are using MapPoint 2010. We have mapped a large number of locations using street addresses in a linked file. Typically 20-25% cannot be found and we must make a selection from the list returned. We have now added latitude and longitude to the linked Excel file. If we run the update will MapPoint use latitude and longitude first for a more accurate placement?
You will need to create a new link and to tell it to use the longitude,latitude data fields instead of the street address.