First in a series on Online-Offline Data Integration
Michael Lewis and Brad Pitt made “Moneyball” a great movie, but also a popular business concept. I’m no Brad Pitt (pause) – but marketers can use some of this same thinking – by focusing on net yield. Rather than bidding up PPC keywords to get more sales volume, only bid those up that have the best net yield, as measured by Customer Life Time Value (LTV).
Customer LTV is a subject marketers love to talk about but rarely implement. Certainly you have heard conference speakers profess that knowing the LTV of your average customer is central to effective marketing practice. The reasons for this disconnect could be many. LTV is sometimes viewed as strategic in nature: something you only need to do every year or so. Kind of like doing your bucket list – not something you pay attention to daily. Or it may be that LTV is rare for a more practical reason. Without an integrated marketing database, marketers have huge challenges bringing the components of LTV together in one place.
Customer Lifetime Value uses many metrics – average sale$, COGS, retention rates, marketing expense, the present value of money, and others – to calculate the profit you will realize on the average new customer over time. The method balances inputs (costs of acquiring & servicing customers) with outputs (profit). For a detailed discussion of Customer Lifetime Value, see our May 2012 blog posts: “Customer Lifetime Value: How much are your Customers Worth?” and how one business used LTV to assess their PPC costs in “Effectively Using Customer Lifetime Value in E-Tail.”
“What If’s” to Consider
In the world of digital marketing, consider your PPC keyword charges as your customer acquisition cost. What if you could do your pay-per-click bidding with LTV values at hand? Goodness, what a concept! What if your bidding was based on the ultimate LTV of a converted customer, rather than just the value of their first purchase transaction? What if you found that you can justify higher bids on certain keywords? For example, some keywords might convert high value new customers into repeat customers with multiple visits!
Professor’s Corner: Measuring the Net Yield of PPC Keywords
Our project: DMSI is in the early stages of a research project to study this phenomenon. Our research thus far has been with a growing gift client, who until 2008 had a pure online presence before adding a catalog to their mix. The client uses the highly respected RKG Group, Charlottesville VA, as their Search Engine Marketing (SEM) agency. In fact, RKG has also blogged on this topic – see RKG Blog for their view. Here are some of our early findings.
- Big swings in LTV. As the project has progressed, it is clear that LTV calculations vary significantly by keyword, a good reason to pursue this project. The influence of highly variable input components – many of which are obtained only from offline analysis – strongly indicates that online-only ROI analysis (i.e. Google Adwords reporting) is incomplete. The baseball analogy here would be that on-base percentage is more significant than batting average.
- Attribution challenges. One of the clarifications we needed to make early on was the difference between a new customer conversion, and a purchase conversion from an existing customer. Our attribution model focused on customer conversion, in a two step process.
Step 1 is done by the SEM agency, to match a PPC click-thru with a subsequent customer order.
Step 2 is done by our WiseGuys software, to insure the customer order is truly a customer conversion transaction, i.e. the first transaction in a series. More on this in a future post, but suffice to say at this point the analysis is equal parts art and science. - Keywords versus Keyword Categories. We first calculated LTV by keyword. This turns out to be too granular and overkill. Our plans now call for calculating LTV by keyword category (perhaps using what Google Adwords is now calling “Label”).
- Allocation of Keyword Expenses. We distributed all PPC charges for a keyword, in a 6 month time frame, against the number of 6 month converted customers. Further discussion suggested including only converted keyword charges for assignment. In this scenario, keyword charges that led to orders – but not customer conversions – would be ignored. This seems to be a fine point but nonetheless worth further study.
- COGS. This component is probably the least variable in our approach, usually 50% of retail price.
- Repeat Buyer distribution. Some keywords may generate only one purchase – e.g., Graduation gifts – others generate many repeat purchases. Obviously the latter contributes to a much higher LTV than the former. These are the keywords worth higher investment.
Stay tuned to our blog as we move forward in our research. We will continue to publish more results from our own research, as well as reader comments and input from other practitioners.
