Thanks to Tim for explaining that the addresses found on the TomTom are actual street addresses and not approximations such as ZIP code centers. On the subject of extrapolating destination locations, I’d like to add that Tele Atlas offers both address ranges and specific address points (which are the exact positions of verified building numbers). By using the combination of the address range with the address points, Tele Atlas customers will both find more addresses and be routed to their destinations with terrific accuracy (either the exact destination, or an extrapolation between known address positions). This serves both the navigation customer as well as GIS and geomarketing companies. It also positions Tele Atlas as a good choice in the emerging mobile phone navigation market, where a pedestrian will want to be guided to a doorway or a store front.

I don’t believe the test was flawed. In order to navigate to a particular given address, that address needs to be able to be found via the device. Yes, the results could be a little bit different with the same map database in a device from a different vendor, but probably not by very much.

There were other aspects of the test that I didn’t document in the text above, but perhaps should have. After an address was found on one device, I pulled it up in the respective “map” view from each manufacturer. Then I pulled it up on the second device and also looked at it in the map view. This way I could make sure that they were at least agreeing that the street/house number was in a similar location. This was pretty easy to verify by looking at nearby intersections.

If I found that the address was in different spots on the street in the two devices, or if the locations looked to be more than about 1/4 mile apart, then I would try to get further independent verification. For example with the business addresses I could lookup the website of the business and look at their text directions to confirm which streets the address would be between. I think in only one case out of 500 locations was there a discrepancy I could not verify. So in these cases there are likely no instances where the address would have been off by 5-10 miles. If they were then that error was present in both devices.

Typically the first step in using an auto GPS device is to find an address to navigate to. Having the device recognize that address is a critical step. If the address doesn’t exist in the database then the GPS will not be very useful to the user. This is the part being talked about here.

You are correct though that there might have been differences when actually arriving at the “found” locations, and I acknowledged this in the last sentence of the “Notes” section. As you mentioned, driving to them would one of the few ways to confirm. My original plan was to do a driving test and scale back the number of locations tested, however there could have easily been a geographic bias (towards either vendor) if one vendor just happened to have mapped that area better. For example my home town is much better mapped by NAVTEQ than Tele Atlas, but that is just one area.

Thanks for dropping by and sharing your comments!

First, both Navteq and Tele Atlas have multiple names for each road (postal name, common name, honorary name, etc.). While this information is different for each map vendor, each application vendor uses different amounts of the data. As an example, one vendor may compile all of the different names for a street into the end-format on the device, while another application vendor may only include the first two names to save storage space.

Second, the map vendors handle addresses differently within their database. Neither is necessarily better as each method has its preferred use and is evident based on the history of the two companies.

Navteq has historically been a supplier to navigation companies. Because of this, they only include the street addresses if the address or address range has been verified in some fashion. Tele Atlas includes street addresses in a similar fashion, but it does something different for postal addresses (which both companies receive a list of, as do many companies, from the post office) that it does not have a source for.

Tele Atlas geocodes unknown addresses to the center of the smallest boundary they can match to. As an example, if your house is not in the Tele Atlas street address list, they would place a reference point for you house address in the center of your zip code.

Tele Atlas’s approach is extremely beneficial for GIS and geomarketing uses, but it can make consumers of navigation devices rather unhappy when they are told they have arrived at their destination only to be 5-10 miles away from the actual address. Navteq’s approach on the other hand will give an automatic unknown address error but may cause more problems for GIS and geomarketing uses. Which is the best way? It depends on the use.

If you would like to actually test the address accuracy of each map vendor for navigation systems, you need to actually drive to each of the addresses to see if the address is actually where the map vendor says it is. Do not simply assume that because a system says that they have found it that they system knows where the address actually is located.