- How We Help
- Identity Resolution
Q: What do the neighborhood pizza shop, the big box department store, a Fortune 500 financial institution, and a national non-profit organization have in common?
A: All four need to know as much as they can about their customers and prospects, and they need to know how to reach them.
The only way to gain this information is through data appending: taking the information you have and turning it into much more. A list of names and addresses can be transformed into a telemarketing list complete with demographics and purchase behavior, for example. In layman’s terms, a data append is when a vendor receives input data, matches it to a master database, and returns a file containing both datasets.
A forward append begins with a name and address, and adds a variety of additional information about an individual. This can include phone numbers, email addresses, vehicle information, property data, demographic information, and much more. Forward appending is the most common form of append, but it is not the only one.
Another common type of append is called reverse appending, whereby one field of data is used to return all other data tied to it. What if a call center has a list containing phone numbers of callers who expressed interest in a product, but lacks name or address? A reverse phone append will fill in the blanks. Using the phone number for exact matching to a master file, the call center can obtain the name, address, phone type, email address, and even complete demographic information associated with that particular number. Based on this additional information, they can segment and send postal mail or email to the prospects, trusting they are marketing to people who are interested in their product.
Reverse appending is available for virtually any other distinct dataset, from email address to Vehicle Identification Number. Typically a reverse append uses exact matching, whereas a forward append can use exact and/or fuzzy matching, which is explained in further detail below. Because of the nature of exact matching, reverse append is more accurate and takes less processing time than a forward append.
Phone numbers, email addresses, and demographics are the most frequently appended elements. However, there are thousands of other elements that specialty companies can provide, from motorcycle ownership to social media propensity. If you can think of it, it likely exists as an element that is available for append.
Appending is not just a one-time process. Some of the country’s largest companies run monthly or quarterly appends to follow their clients’ behavior and demographics; as these items change, so do their needs as consumers.
For example, a bank benefits from knowing that their client has recently married, because direct marketing to a newly married 30-something looking to buy a home is different than to the single 20-something who lives at home with their parents.
There are two ways to obtain appended data: through real-time lookup or batch processing.
In real-time appending, a set of information is sent to a server, which completes the match and returns the additional data elements – all within a fraction of a second. This is perfect for companies that need information on the spot, for one record at a time. Examples of applications for real-time appending include a retail store signing people up for a customer loyalty program or a customer call center instantly routing inbound calls.
Batch processing refers to processing a file with a consistent layout, whether it contains 100 or 100 million records, for simultaneous matching.
Batch processing can be completed manually or automatically. In a manual append, a processor will load the data into a database program like Oracle, write code containing match instructions and logic, and then execute the process. The processor needs to be sure that the input file is ready for append, with all necessary fields containing quality data (no unknown characters, extra delimiters, or foreign addresses, for example).
Manual appending leaves room for human error and variance, but it can be advantageous for complex, custom match jobs. A reputable company will add an extra quality control step to ensure that both system and human processor have completed all steps correctly. At one time, all data appends were manual. Because of the potential for error or slight variances in methodology between processors, larger companies have created automated systems that complete tasks from basic matching to complex, multi-step fuzzy matching projects.
For some automated batch systems, it’s as simple as 1-2-3.
1.) Users load a file to an FTP site
2.) The vendor’s system receives the file and performs the append
3.) Users receive an email when the file is processed and back in the FTP folder for pick up
How does the system know what to do? Many automated systems offer complete flexibility in initial configuration including input layout, match levels, output layout, suppressions and more. Once the append project is configured to meet client specifications, the automated process runs itself. However, each data provider offers slightly different options and will have their own, unique way of handling automated batch appends.
For a forward phone append, input data would need to include, at the most basic level, a street address and a zip code. City and state should not be required; a quality data provider will have tables that infer city and state from zip code.
First and last names are important for tighter matching. Without a name, you may input an address and get the intended person’s roommate’s phone number returned. Some applications need the tightest level of matching; in skip tracing, for example, the goal might be to find one specific person, not their spouse or roommate.
In a reverse append, the required information is a lot less complicated: one field for exact matching, whether it is phone number, VIN, IP address, email address, or other input data field.
The first step is standardization. If your file lists an address incorrectly, or different from the master database, there won’t be an exact match – even though the addresses are clearly the same to human eyes:
Standardization converts every input record into a standard format. A good vendor database will already be standardized, so the input data needs to look the same. If the master database refers to every “Street” as “St” and every “Place” as “Pl,” then standardization is vital to ensure proper matching.
After the input data is standardized, it is compared to what the vendor has on their master file using a given set of match logic. Typically, the vendor will have varying levels of matching available. Some allow the client to choose match levels, while others are rigid about using all of their match levels — and while these companies flag the output data with match levels, clients are not free to choose which levels are acceptable for their particular application.
Matching at the individual level refers to a match using first name, last name, and address.
Matching at the household level refers to a match at the last name and address.
There are two ways vendors perform data matching. In exact matching, the input data matches the vendor’s data exactly and to the letter; in fuzzy matching, there are variances between the two data sets, but the vendor believes the match to be correct. For example, a fuzzy name match:
There are several ways to match names, but the most commonly known matching algorithms are Soundex and Jaro-Winkler. Soundex looks at the phonetics of a name (Smith vs. Smythe), where Jaro-Winkler compares two data sets (in this case, names) and assigns a score based on the similarity of the two pieces of information. Data compilers with advanced capabilities will often go beyond these algorithms and create their own proprietary match logic.
After the match process takes place, the standardized records are typically changed back to the exact same format as the input file; standardization is for the benefit of the append process, not to improve deliverability (address standardization is a product available from many vendors – but it is separate from data appending).
This illustrates the classic quality vs. quantity dilemma. If a phone append rate is higher than expected, you should ask how recently the appended data has been validated and how tight the match logic was. If the append rate was lower, you may be able to loosen up the match logic to see how many more matches you might receive. Understanding this is particularly important when evaluating two or more data vendors to ensure equal comparisons.
Your provider should be able to tell you exactly how the information was matched. Common matching levels include Individual, Household, or Address level matching. Some vendors will allow you to customize this and others will hold more rigid guidelines. It is important to know ahead of time if multiple match levels are available and if your project can be customized according to the various levels.
A client-focused data provider will allow for flexibility on both input and output layouts, as far as format (comma delimited, fixed) and fields allowed. Understanding specific input and output information from the beginning will save you time and resources long-term.
As mentioned in the article above, there are specific matching algorithms available for fuzzy matching. Make sure you understand how your provider matches data to ensure a quality append.
While you may be looking for a specific data element, it is likely there are other elements available that could help you further segment your consumer audience. Check with your provider to gain a full understanding of all elements available to maximize your data append project.
The source, quality, and quantity of the underlying data are critically important to understand prior to selecting a data provider. Ask whether the data is from an original compiler or aggregated from several compilers. Find out about whether the data is from double-verified, trusted sources. Understand how the database is built and maintained so you can be confident in the accuracy of the data you receive.
Receiving verification dates within the appended data gives you the power to filter out older records or use all records, depending on the goal of your project. The more information you have about your data, the more accurate decisions you can make.
Confidence scores can indicate levels of perceived accuracy of the data. Many providers will include these scores on appended records to provide you with their degree of confidence in the data provided. Confidence scores are another way for you to filter through appended information and make the best decisions for your initiatives.
Flags within data can be very useful and provide significant cost savings. Removing deceased individuals from calling or mailing campaigns will result in immediate savings. Flagging or removing wireless records, phone numbers on the Do-Not-Call Registry, and opt-out email addresses can save significant money in fines or potential lawsuits.
When you are evaluating new vendors, ask if they provide opportunities for testing and evaluating their data append service. Many vendors will allow you to test both the data and the append process prior to committing to a larger project.
Particularly critical if you have large projects, it is important to understand how robust your data provider’s system is. A strong platform will be able to process millions of records per hour, giving you the best chance to meet your project deadlines.
Many data vendors who offer an automated batch processing platform or real-time API make these systems available 24 hours per day, 7 days per week. Many even include off-hours support to help with potential system challenges. Others may have different hours of service. It is important to know the availability and support of the batch processing system.
Simply put, ask whether your data provider has a back-up plan if the platform is experiencing problems. Some data providers offer redundancy, meaning they have two identical systems processing the data on different servers. If one is not working, the other immediately takes over until the problem is resolved. For you, this means no project delays due to unforeseen technical challenges of the provider.
With much buzz about data breaches among the largest US companies, you need to understand how your data provider is securing your data. Security among database vendors is a constantly changing science, with the top providers always staying ahead of the curve.
Taking security a step further, specific training and data handling procedures are necessary when a provider is handling sensitive Personally Identifiable Information (PII). PII protection processes go above and beyond standard data security practices. Your provider should be able to provide detailed information in their handling of PII data.
This may not seem like a critical question, but a good Account Management support team can make or break a project. Having a dedicated support team will ensure your project criteria and deadlines are met.
Data appending is a practice that has been utilized for years as a way to further identify, verify, and segment consumers. As technology and data collection has advanced, so has data appending. From lengthy manual appending to receiving appended data in subsecond time, the pace of processing has been driven by the speed of consumer demand.
Data appending will continue to evolve as marketers and fraud specialists refine and improve processes for consumer validation and segmentation.