Many Internet retailers and social media companies keep very good track of both their cost of acquisition and estimates of the long-term value of their customer base. As long as the long-term profit stream (discounted back at the appropriate cost of capital) of the average customer is much higher than the cost of acquisition (\"COA\") of the customer, management feels pretty good. They monitor the two numbers and try to improve both of them as part of their management activities. Specifically, they will test or experiment with different customer acquisition modes and try to find the ones with the lowest COA.\n
The problem with this thinking is that it is only true \"on the average.\" The beauty of Internet businesses is that they don't need to just look at averages, but can disaggregate customers. For many online customer acquisition modes, for example, the Internet site - maybe an affiliate or search site - that the customer came from is available. In many other cases, codes or separate Internet URLs can be used to capture differing offline acquisition modes such as magazine or broadcast advertising. It also usually makes sense to ask customers how they got to a site as part of any initial customer registration. With this ability to disaggregate customer acquisition mode from other customer data, a different picture can emerge.\n
In one example analyzing two years of customer data, it was found that magazine advertising (with an attached coupon) brought in $474 in average revenue per customer and median revenue per customer of $108, while word of mouth (using coupon referral codes) brought in $574 in average revenue per customer and median revenue of $128. The COA for magazine advertising was about $40, while for word of mouth it was less than $20. In this case magazine advertising was both more costly than word of mouth, and generated lower long-term customer value. Without the ability to disaggregate customer acquisition by source, this insight would have been lost, and it should be obvious that only looking at aggregated data might lead to very different - and less efficient - customer acquisition resource allocation decisions.\n
Companies need to keep disaggregated data on each customer, ranging from the different ways customers get to sites, to tracking distinct on-site activities, including purchases. Advance planning is needed to be able to accurately acquire and productively relate disaggregated customer information to the rest of the customer data. Once the data is gathered however, it is relatively straight forward to estimate the COA and long-term customer value (\"CLTV\") by cross tabbing the data. More complex statistical analyses can then be done to isolate other factors that might be influencing COA or CLTV but that are not associated with the mode of acquisition. Factors like sales or promotions that applied more to one cohort than another may require some statistical adjustment, but simple cross tabbing will usually get most of the value from the disaggregated data.\n
In sum, we believe simply measuring aggregate COA against CLTV is no longer sufficient. Instead, we recommend analyzing disaggregated data that marries how a customer was acquired to how much he or she purchased over time. The resulting analysis will show that some customer acquisition modes may attract much more valuable customers than others, even if they may cost somewhat more. Conversely, one may find some low-cost acquisition vehicles bring in lower-valued customers. Regularly analyzing disaggregated customer data in this fashion has the potential to increase the productivity of customer acquisition efforts by orders of magnitude.\n
Do you have a readily available source of disaggregated customer data that might improve your company's customer acquisition performance?"