Analytics news and expert advice every Thursday.

# Applying Customer Life time Value To Paid Search

Online marketers have realized that traditional metrics such as sales, brand wareness, attitudes etc., are not enough to show return on investment. In fact, some academic studies have shown that marketing actions that improve short term sales may actually harm long term profitability of a brand.

Further, with improvements in data mining and warehousing, marketers have enormous amounts of data to analyze and convert into actionable insights. Customer Lifetime Value (CLV) was once an esoteric metric that lived in the domain of direct response marketers.

Now, online marketers are adopting it as a way of measuring the long term return on investment of their marketing efforts. As CLV measurement and application is a vast subject, I shall restrict the scope of my article to the mathematical intuition behind CLV and its application to SEM bid management.

**Fundamentals Of CLV Modeling**

The CLV is a metric that tells you, in today’s dollar terms, the value of a customer over the lifetime of her relationship with the marketer. Consider a simple example where up on clicking on a search advertisement, a acquired customer, makes a transaction and makes 3 subsequent transactions by coming directly to your website.

Each transaction nets you $2 profit and the CPC on the click was 20 cents. Here, the customer lifetime value of the customer is

CLV= $2 profit per transaction * 4 transactions – $0.2 CPC = $7.80

Generalizing this to a compact mathematical formula,

Where GP is the average gross profit per customer ( revenue minus cost of sales) per time period

*r*is the retention rate in the time period

*n*is the time horizon for estimating CLV.

- Mkt.Cost is the Marketing Cost

The acquisition cost is the cost of all the marketing dollars you spent to acquire the customer, as well as all the money you spent to retain the customer through programs such as promotions, discounts and so forth.

Note that In the interest of simplicity, I have ignored the cost of capital in the formula. In practice, you may have to include it, specially when you calculate CLV for long, multi-year, time horizons and in high interest rate environments.

**Application Of CLV To Paid Search**

In traditional marketing, the CLV is used to calculate the intrinsic value of every marketing segment. One can afford to spend more in segments where the CLV is high as the profits to the firm payout in the long run. The opposite applies to low value segments.

In search, the CLV could be calculated at the keyword level to determine how much one could afford to spend on the segment the keyword attracts. Consider the following example where the average profit has been calculated for the 2 keywords:

Profit (now) | Profit ( 3 months) | Profit ( 6 months) | Profit (9 months) | |

Armani | $450 | – | – | – |

Designer Shirt | $100 | $150 | $150 | $100 |

Ignoring the acquisition cost (lets say an SEM click), the CLV for Armani with a 1 year time horizon is $450, while that Designer Shirt is $500. Note that the average customer coming through an Armani search is 4.5 times more profitable than the average generic term customer when we consider the first transaction only.

However, the lifetime profitability of the generic customer is over 10% higher than the average Armani customer as the generic term customer in this example makes many more subsequent transactions.

If I were using a bidding strategy which aim to maximize profits calculated only at time of sale (i.e. the first transaction only), I would think Armani was far more valuable (4.5 times more) than designer shirt. If I were using CLV as my profitability measure, the opposite is true.

**Should You Care?**

The bidding decisions made with a CLV strategy and a time of sale profitability strategy will vary significantly when you have a keywords that attract customers of widely varying lifetime behavior. Examples are:

a. Repeat purchase behavior varies widely between keyword customer segments. In the example above, Armani searchers don’t purchase repeatedly whereas generic customers exhibit repeat purchase behavior.

b. Profits per transaction vary widely over a lifetime of a customer. Consider a case where one person buys a weight loss product (profit $50 every 3 months) and another buys a weight loss program DVD (profit $20). However, DVD purchasers tend to buy more DVDs *and* the product in future transactions. In this case too, the profitability calculation will vary a lot based on the method used.

Profit (now) | Profit ( 6 months) | Profit (1 year) | CLV | |

Weightloss Product | $50 | $50 | $50 | $150 |

Shoes | $20 | $50+$20=$70 | $70 | $190 |

However, if your keywords are attracting customers who have similar profitability per transaction and the customer purchasing frequency is relatively similar then adopting a CLV strategy might only give limited benefits.

**Getting Started With CLV**

If all this talk has made you eager to try measuring and optimizing to CLV, I suggest the following:

a. Determine the purchase frequency of an acquired customer and the duration of an acquired customer. This is tricky because estimation the lifetime of a customer is not clear. Theoretically, if a customer has not bought a product for 5 years, there is still a small chance she might buy something in the 6^{th} year. I suggest measuring the cumulative distribution of percentage customers who are new, year 1, year 2 etc. You can assume a 95% cut off to determine the lifetime value i.e. if 5% of customers in your mix are year 3 or higher, then you can use a 2 year horizon.

b. Now calculate CLV at a keyword level for the time horizon you just estimated.

c. If the CLVs by keyword vary a lot then you might want to start optimizing your campaigns by CLV.

Remember that CLV is a lifetime profitability *estimate, *it is not the number that runs your lights (that would be the profit at the time of sale). Hence, to test the efficacy of your program, you must measure the actual profits from your CLV optimized campaigns periodically.

*Some opinions expressed in this article may be those of a guest author and not necessarily Search Engine Land. Staff authors are listed here.*

### We're listening.

Have something to say about this article? Share it with us on Facebook, Twitter or our LinkedIn Group.