Media Coverage

Confused by the link between optimization and ROI? Here are pointers on how to calculate results.

When optimizing wesite content or ad campaigns, we can talk about improving user experience, creating engagement or brand-building, but in the end we have failed if the exercise does not ultimately lead to more money. After all, driving ROI is the point -- whether working off direct response goals or the goals mentioned above.

Direct: the shortest distance between here and more money
Money has a time-related value, which tends to focus the mind on what can be done now with easily measured results. Hence, there is a bias towards optimization of the easily measured and immediate.

In listening to conversations about client interaction optimization, the language of online optimization starts to sound very similar to the language of direct marketing: A/B testing, lift over control, champion-challenger, etc. When first using optimization, typically you want to capture the low-hanging fruit: the boost in conversions you get from presenting the right current offer to the right current visitor.

For example, let's say you are optimizing a home page, have five distinct products to offer, and establish a goal of maximizing the number of conversions driven from this page.

Before optimizing, you collect data, rotating the five offers equally, and as a result you get the following mix of conversions:

After optimizing to maximize the conversions, the results might look more like this:

Your lift from optimization: (1,633 – 1,066)/1,066 = 53 percent.

If we know nothing else but how much we make on an average sale (let's say that is $87), then these results seem like a good thing from the "more money" standpoint. This means we are getting an extra $49,400 a month, right?

If all we know is the average NPV (net present value) per sale that would be correct. However, if we knew the amount we made on each individual product, we could take a smarter approach to understanding what is going on.

Let's say we know the NPVs of all the products:

That means the money is distributed like this:

In other words, we actually lose money (-$21,800/month), because the highest response rate products selected by our conversion-rate maximization had low NPVs while the low-response rate products had high NPVs.

Armed with this comprehensive product value information, we'd be much better off optimizing to maximize the expected value (estimated conversion rate times estimated value of a conversion). Then the optimized case looks something like this:

That gives us a much lower conversion rate lift: (1,120 – 1,066)/1,066 = 5.1 percent.

But before we get too concerned about the lower conversion rate lift, let's look at the money:

Wow -- now we're talking. We have a value lift of: (171,225 – 93,100)/93,100 = 84 percent.

With this approach, we are pulling in an incremental $78,125 per month. This is the kind of thing that we can use to get promoted, get a bigger bonus, and, in general, enjoy the good life.

But what if the problem we need to solve is different? What if we are new to the market, no one is familiar with our products, and we are trying to invest now in order to make sales later? Can we optimize to that end?

While optimization is excellent at driving to direct-response goals, it is also a fantastic tool for pursuing longer term goals like brand-building, engagement and improving customer experience. Though the link between optimization and money becomes more indirect and needs to be approximated in these situations, make no mistake about it -- engagement is still all about the money.

The really big money often comes from less cut-and-dried processes
The most important levers driving return on marketing investment are not always direct-response type goals. Sometimes the product or service is too new, too complex or too unknown for a pure direct-response approach to work well; you may need to educate and engage prospects before they are ready to make a purchase. Perhaps the purchase decision process for your product is too long term for a direct-response approach, or perhaps you have already squeezed the most out of your current site and creative, and you need to make big, fundamental changes to take your business to the next level.

User experience improvement, brand-building, customer engagement and positioning versus the competition are all very worthwhile investments, and optimization need not be excluded from such efforts. We just need to translate these goals so that they make sense as optimization targets, in terms of measures that are easily captured online.

Let's look at an example:

Say we have a new product to promote on our website. The product is Whezone, a newly approved drug to treat respiratory system inflammation. A significant amount of education and brand building will be required to make Whezone a drug that patients ask for, but we know from our experience that informed patients can influence their physicians' prescription choices -- each 1 percent increase in top two box score on brand recall among patients translates to $10 million in additional sales over the time the drug is still under patent. A similar increase among physicians will produce 10 times that impact -- about $100 million.

Now, our optimization goal will be expressed not in conversions or NPV, but in terms of engagement. If we are maximizing engagement, we need to optimize towards a goal that accounts for the following factors, and we could combine them using a point value scheme for each type of activity to be included:

• Visit frequency (visits per month) -- five points per visit
  
• Visit depth (distinct pages viewed) -- three points per page viewed beyond starting page for the visit
  
• Visit duration (minutes per visit) -- five points for every minute beyond the first minute
  
• Commitment actions: Registration -- 30 points; blog contribution -- 20 points; posting to discussion -- 20 points; requesting free sample coupons -- 50 points

Let's say we have five different message/creative combinations that we have built for our different types of visitors:

We'd again put them in equal, random rotation for a while to collect data about who responds to which offer. If the results looked something like this:

Then in the optimized case the results would look something like this:

The lift in our engagement score: (17,400,000 -- 10,680,000)/10,680,000 = 62.9 percent.

What about the money?

Being precise about it would require us to link our lift to an increase in brand recall -- which would best be done via an online survey of our visitors run before and after optimization was initiated. Again, if each point in brand recall is worth $10 million for patients, $100 million for physicians, and we are actually having an impact on those scores, then the money side of this equation will probably look very good indeed.

Conclusion
Optimization is ultimately successful only if it drives a positive ROI. This should not blind us to the fact that the most important levers driving return on investment in the longer term will continue to be user experience, branding, engagement and positioning vs. competitors. Optimization can be applied to these if we can translate these goals into optimization-friendly terms.

This article was originally featured in iMedia Connection on 4/24/08.
Click here to see the original article.

add to del.icio.us

Digg this

Technorati Tags: conversion optimization  |  xplusone