Often we need to enhance internal data by using the services of external vendors. Most organizations already use vendors in some capacity. For analytics, they could be a source of additional data, partner for a specific project, or act as analytics consultants. Often organizations have certain personal information that they acquire through the existing relationships/transactions, however, they would like to help their marketing better target customers/donors. This is where this post comes in. Before purchasing external data that enhances your existing data files consider the following questions:
- What is the vendor’s match rate? The data that you will provide to the vendor (usually full name, address, phone, and DOB in some cases) will need to match to their file and then the vendor appends additional data points. Keep in mind that the overall match rate will be different when the match rate of additional append variables. For example, the vendor might have a match for John Smith, born Jan 1, 1950, but not his wealth, pet ownership, or car driving habits. The good overall match rate should be 80% or higher, while specific categories might be much lower than that.
- What would you be using it for? Do not buy additional data without having clear use cases and plans. For example, before buying a propensity to buy variable, test its value for a group of prospects half of which become customers and half did not. For a phone number, check the reach and donation for a group that has appended numbers vs those that do not. Lastly, having a use case plan will allow to justify expenses and learn valuable lessons regarding the efficacy of the purchase.
- How transparent is the cost structure? Usually, there is a cost to append per 1000 records, however, some vendors would not be transparent about additional fees. For example, some of them will have a high minimum purchase amount, far in excess of what you need. Others might offer free items, but only from one survey/dataset. For example, dataset one has phone numbers, but if you want to know if those are work or personal, then they will charge you a fee from another dataset. Always be careful with pricing as some of it is hidden.
