Address correction services Melissa Data vs IntelligentSearch
Address Correctors,
I am an database architect / .net developer for a financial firm and was recently tasked with finding and evaluating an address correction service for free form addresses that are entered by our clients. The 2 front runners in the vendor race were Melissa Data and Intelligent Search. My requirements were to connect to their evaluation engine programatically, retrieve a record set containing data and potential error codes, and store them in my local db. I judged the winner based on the following criteria:
Programmatic ease of use
Data reliability and consistancy
Cost
Vendor Support
Here are my results:
Programmatic ease of use:
Both MD and IS have support for .net web services which came in handy for me.
The IS web service had a few main published web methods. Strings as inputs and they all returned an object of XMLNODE type. This was essentially a pretty little XML document of variable length containing address, postal carrier, and possibly error info for the addresses. It also returned near matches in the same XML node object which is a very big plus. They also had support for "bulk loads". The documentation and code samples provided were adequate to get started pretty quick. The percentage of incorrect addresses reported was roughly 4.1% the sampling was roughly 70,000 distinct records.
The MD web service had fewer published web methods, but there were pretty easy to use. The interfaces were slightly more complicated and messy, with the necessity to declare and pass a larger volume of what i thought were unnecessary parameters. Strings as inputs and strings as outputs. This service (at least the methods I was calling) did not return possible or near matches and the error codes and descriptions were not as detailed as IS. Their documentation and code samples were adequate to get started but unnecessarily complex for the data that actually got returned. MD did however offer more services "under the hood" for name and phone number lookup etc. MD did have support for "bulk loads" also. The percentage of incorrect addresses reported was also roughly 4.1% the sampling was roughly 70,000 distinct records.
This was close but I leaned toward IS here. I hadn't worked with XMLNodes before but with a quick google search and a few tests, I was in. More data came back much cleaner and clearer.
Data reliability and consistancy:
Interesting to note that both vendors had a very similar percentage of invalid addresses for my total data set. What is more interesting is that they only agreed on roughly 2/3 of the addresses marked incorrect. Here's the numbers
Total records: 69847
IS marked wrong 2838
MD marked wrong 2873
Both marked wrong! 1898
The services only agreed on roughly 2/3.
IS seemed to get a few of these addresses wrong. In spot checking some of the addresses that were marked wrong from IS maps.google.com was able to show me a picture of the house! the same can be said of the records from MD. In spot checking some of the records that were one vendor said was invalid and the other said was valid, i was able to find some addresses from google and not others. it seemed pretty balanced. The results of that test are still ongoing.
The data returned from IS seemed to be more consistent in the way that invalid addresses were corrected. I saw fewer corrections in the melissa data returned. The IS data actually told me in return codes which data points were corrected when they were corrected too. big plus.
I definitely leaned toward IS here. Although I did find holes in IS data the corrections and descriptions of what was corrected made this pick easier.
Cost:
Even though I distanced myself from the financial parts of this as much as I could I know that IS is cheaper the MD. With the other features I have spoken about. IS wins here again. There is no monthly fee, it's ala carte. $1000 for 150k searches (without GEO positions).
From a strictly cost perspective for the functions I was calling, IS is cheaper.
Vendor Support:
I had to call IS for support on an account set up problem. They were courteous and within 5 or 10 minutes I had the result I was looking for. Again, their application documentation was adequate.
I didn't have to call MD for anything. It was an email setup and within 24 hours I was able to connect to the webservice and start making data calls. Their documentation was adequate (a little confusing) and I haven't even received a call from a sales person yet.
winner = toss up.
Outside details:
IS gave me a very limited number of "test" searches. 50 to be exact.
MD gave me a 30 day (seemingly unlimited) trial.
The response time of the IS web service seemed to be superior. MD web service would occasionally time out even though their web site claims redundancy and load balancing.
In a reasonably close race I am picking IS.
Good luck with address correction!
I am an database architect / .net developer for a financial firm and was recently tasked with finding and evaluating an address correction service for free form addresses that are entered by our clients. The 2 front runners in the vendor race were Melissa Data and Intelligent Search. My requirements were to connect to their evaluation engine programatically, retrieve a record set containing data and potential error codes, and store them in my local db. I judged the winner based on the following criteria:
Programmatic ease of use
Data reliability and consistancy
Cost
Vendor Support
Here are my results:
Programmatic ease of use:
Both MD and IS have support for .net web services which came in handy for me.
The IS web service had a few main published web methods. Strings as inputs and they all returned an object of XMLNODE type. This was essentially a pretty little XML document of variable length containing address, postal carrier, and possibly error info for the addresses. It also returned near matches in the same XML node object which is a very big plus. They also had support for "bulk loads". The documentation and code samples provided were adequate to get started pretty quick. The percentage of incorrect addresses reported was roughly 4.1% the sampling was roughly 70,000 distinct records.
The MD web service had fewer published web methods, but there were pretty easy to use. The interfaces were slightly more complicated and messy, with the necessity to declare and pass a larger volume of what i thought were unnecessary parameters. Strings as inputs and strings as outputs. This service (at least the methods I was calling) did not return possible or near matches and the error codes and descriptions were not as detailed as IS. Their documentation and code samples were adequate to get started but unnecessarily complex for the data that actually got returned. MD did however offer more services "under the hood" for name and phone number lookup etc. MD did have support for "bulk loads" also. The percentage of incorrect addresses reported was also roughly 4.1% the sampling was roughly 70,000 distinct records.
This was close but I leaned toward IS here. I hadn't worked with XMLNodes before but with a quick google search and a few tests, I was in. More data came back much cleaner and clearer.
Data reliability and consistancy:
Interesting to note that both vendors had a very similar percentage of invalid addresses for my total data set. What is more interesting is that they only agreed on roughly 2/3 of the addresses marked incorrect. Here's the numbers
Total records: 69847
IS marked wrong 2838
MD marked wrong 2873
Both marked wrong! 1898
The services only agreed on roughly 2/3.
IS seemed to get a few of these addresses wrong. In spot checking some of the addresses that were marked wrong from IS maps.google.com was able to show me a picture of the house! the same can be said of the records from MD. In spot checking some of the records that were one vendor said was invalid and the other said was valid, i was able to find some addresses from google and not others. it seemed pretty balanced. The results of that test are still ongoing.
The data returned from IS seemed to be more consistent in the way that invalid addresses were corrected. I saw fewer corrections in the melissa data returned. The IS data actually told me in return codes which data points were corrected when they were corrected too. big plus.
I definitely leaned toward IS here. Although I did find holes in IS data the corrections and descriptions of what was corrected made this pick easier.
Cost:
Even though I distanced myself from the financial parts of this as much as I could I know that IS is cheaper the MD. With the other features I have spoken about. IS wins here again. There is no monthly fee, it's ala carte. $1000 for 150k searches (without GEO positions).
From a strictly cost perspective for the functions I was calling, IS is cheaper.
Vendor Support:
I had to call IS for support on an account set up problem. They were courteous and within 5 or 10 minutes I had the result I was looking for. Again, their application documentation was adequate.
I didn't have to call MD for anything. It was an email setup and within 24 hours I was able to connect to the webservice and start making data calls. Their documentation was adequate (a little confusing) and I haven't even received a call from a sales person yet.
winner = toss up.
Outside details:
IS gave me a very limited number of "test" searches. 50 to be exact.
MD gave me a 30 day (seemingly unlimited) trial.
The response time of the IS web service seemed to be superior. MD web service would occasionally time out even though their web site claims redundancy and load balancing.
In a reasonably close race I am picking IS.
Good luck with address correction!
posted by Steve Bowman at 10:30 AM
4 Comments:
Keep up the good work.
This comment has been removed by the author.
Melissa Data Address verification guides and helps the business people to locate or verifies the address before forwarding or servicing for the same.
Nice Blog. Thank you for sharing.
Address Verification API
Post a Comment
<< Home