Calculating ROI 
(Return on Investment)

Jim's Intro:  An important thing to understand about database marketing metrics is they can be defined differently by different people.  The critical issue is:  within the company, does everyone agree on the definition?  

A common definition of ROI (Return on Investment) involves looking at the cost of a marketing campaign relative to the profit generated.  This approach comes from the original definition of ROI as used by finance people rather than marketing people.

A Return on Investment in financial circles has a very specific meaning:  How much did I invest in an asset to the company, and what was my Return on the Investment?  This is the same "return" concept used in the stock and bond markets, as in:  "My stock portfolio had a Return of 20%."

If you invested $1000 on January 1, and on December 31 you had $1,200, you generated a 20% ROI - your Return on the Investment of the original $1000 is $1200 - $1000 = $200 / $1000 invested = 20% ROI on an annual basis (excluding transaction costs).

ROI in marketing can be defined in the same way, though a financial purist (CFO?) will tell you marketing typically involves expenses (income statement), not assets (balance sheet) so you really can't have "ROI" under a strict financial definition of the term.  But the general idea is the same, what you are trying to do is create some sense of the value of a marketing expense, a "yardstick" so you can compare one marketing program to another.

Let's say you're selling widgets with a price of $100.  50% of this $100 is margin, what's left over after the cost of the product.  It costs you another 10% of the $100 to take the order, pick, pack, ship, and deal with returned widgets.  So profit on a sale is $40; the other 60% of the product price of $100 goes to cost of the product and processing the sale through the system.

You put $1000 into a campaign to try to sell these $100 widgets.  You sell 30 of them at the profit of $40 each, for a total profit of $1200 on the campaign.  But this profit is before the cost of the campaign, so you subtract this cost, $1200 - $1000 = $200.  This is the Return - $200 extra dollars came back to you on your original marketing investment of $1000.  Your Return on this Investment is $200/$1000 = 20%.  This is "simple ROI," a straight calculation of Return without factoring in the element of time.

Why does this matter?

Back in the finance department, where they have cash, they can invest it.  In computers, in software, in people, or in bonds, or their own stock.  If your marketing programs can't generate a higher ROI than say, very safe government bonds at 6% annual ROI, then the company should not invest in the marketing programs.  The company as a whole is "better off" investing in the bonds, or in people, software, buildings, or whatever delivers a higher ROI for the company.

How do you calculate ROI on a banner campaign on a site where no products are sold?  For the sake of simplicity, we'll say page views cost nothing to generate and all we look at is revenues from ads on the site.

There has to be a value metric around somewhere.  If you sell ads for $20 CPM, and you're 50% sold out, you're at a net $10 CPM.  Assuming 1 ad per page, each page view is worth a penny in revenue ($10/1000) at a $10 CPM.  How many page views did the banner campaign generate versus the advertising cost to generate them?  If you spent more than a penny to generate a page view, you lost money - negative ROI.

If you have some other more complex revenue generating methods, try using Total Revenue / Unique visitors.  Divide total revenue over some period of time by the unique visitors  in that time period.  Let's say you come up with $1 per unique.  How many uniques at a value of $1 did the banner campaign generate versus the advertising cost?  If it cost less than $1 to generate each unique visitor, you made money.  If it cost $.80 to generate each unique, you ROI is:

$1 - $.80 = $.20; $.20/$.80 = 25% ROI

There's more to campaign ROI than just the "front-end" revenue generated versus the ad spend.  What about the future value of the customer, repeat behavior?  You can get a good idea of how to measure this residual value by taking the tutorial Comparing the Potential Value of Customer Groups.

The above examples ignore any other costs to the organization.  To be more accurate, you would want to at least figure in operational costs.  One way to approximate would be to take total operations cost for any period and divide by the number of unique visitors served in the same period.  Let's say in a month you serve 1 million uniques and your operating costs were $100,000.  That's an average of $.10 per unique.  So true ROI for the campaign would have to include the cost of servicing the incremental uniques you generated.

Let's say you spent $80,000 in the campaign.  It generated 100,000 uniques, which generated $100,000 in revenues ($1 per unique).  But the cost of serving each unique is $.10, so you have to net this cost out:

100,000 uniques x $.10 = $10,000 cost

$100,000 revenue - $10,000 cost = $90,000

$90,000 net - $80,000 campaign = $10,000

$10,000 / $80,000 = 12.5% ROI

Now, do people really think this way?  Sure, offline.  Online, the cost of serving incremental page views may be next to nothing.  But there is some cost - they have to designed, stored, and bandwidth used to serve them.  The point is, this model doesn't "scale to the sky" without additional costs.  At some point as you grow, you have to buy more servers, storage, bandwidth, and so forth.  And hire people to service all of it.  So to calculate the true ROI of a campaign, any additional costs must be taken into account.  Costs frequently overlooked might include design, agency fees, answering phone calls, clearing credit cards, and so forth.  In remote shopping, these costs can be huge, such as to pick, pack, and ship.  They must be included in the overall calculation to measure true campaign ROI.

One way to handle this is to get agreement on a "flow-through" number with the finance department (if you have one).  What percentage of each dollar in revenue actually "flows through" to the bottom line, to cash flow, or to EBITDA?  For example, in the cellular business the number that matters to Wall Street is called EBITDA Margin, and it can range from 25% to 45%, depending on the growth stage of the company.  If at your cellular company it's 30%, then any $1 you generate in revenue has to cost less than $.30 to generate or you lose money, since 30% is all that "flows through" on each revenue dollar.

The above approaches are more likely followed by mature companies; they ignore the "land grab" mentality of the early dotcoms.  But it's what Wall Street means when they say "Return on Investment," and the way Wall Street thinks usually matters in the long run.

The Drilling Down book shows you how to predict the ROI of marketing campaigns based on simple financial models you can build with an Excel spreadsheet.