If you asked a new hotel staff member what a guest was worth when he or she first walked through the front door, you would get responses like $150-$200 or even a bold guess of $300. When you tell them they are actually worth over $50,000, disbelief usually leads to an explanation of how this is factored. If a guest has a good experience, he or she will likely visit twice a year for at least 10 years and would also tell five other people, 20% of whom would stay as well, and hopefully a similar dynamic would prevail.
Now ask that question of the person who is running the email program for that same hotel, and what type of answer do you think you’d get?
LifeTime value of a customer (LTVC) is a technique that has been used for years to guide marketers, providing profitability views of a customer over time. The academic research on this topic is extensive and deep, and the more sophisticated formulas would make a particle physicist jealous. They include a range of components such as NPV (net present value) stochastic models and Iso-value curves (a way of graphically depicting customers and their future values despite past behavioral differences).
Even with the more sophisticated modeling involved, you should know one thing: Your customers simply will not behave according to your predictions! LTVC modeling is a guide at best if used in the traditional sense, as it can’t adequately reflect the nuances of day-to-day consumer behavior.
So, how do you really know what customers are worth, how much to spend on them, how much to discount, when to discount in the lifecycle and which channels to spend the most on? LTVC only gets you to a certain point of decisioning.
For years, my approach to “email, social, mobile, display, search” and LTVC was that there was a channel proxy that goes into this model — the only way to show the long-term value of the activities that you handle day in and day out. I also believed you must be able to tie back to LTVC, even if your measurement is proxy-driven (e.g cost of sale, CPC, CPL ). I also did this to show that many companies didn’t accurately reflect the front-end costs of acquisition or project these out long enough to understand the value of channels at a point in time.
I tried to simplify the formula for a single purpose: to show the original value of an email address. This was easier to justify when it was a variance between sending a piece of mail vs email. If I’m ever confused about attribution or the chaos of marketing in general, I refer back to this model and build my logic from there. When I look at this today, I am challenged by two primary issues:
1. How has the “referral” value in a LTVC model changed with the dynamics of social and mobile?
2. How do we apply this within this model based on new measurements available?
Many apply RFM (recency, frequency, monetary), yet in reality we operate in a world of profit and bottom line.
What’s really cool about tomorrow is, we now have mainstream buy-in to customer word of mouth and referral and ability to measure in many forms. This in itself brings new life to channel proxy values to a LTVC model, and the rationale to apply it more than once a year at budget time.
I’ll leave you with this thought. If you’ve never calculated or don’t understand the merits of
LTVC= [(M-C) x (PxY) – A + (A x N)] X F — you’ll have difficulty showing the incremental value of new emerging channels on what a customer really means to your business. Learn it and translate it for your efforts.