Your customers’ digital journey is a gauntlet. From the initial search to the ad click and all the way down to payment, each checkpoint whittles through all but a dedicated minority. In SEM, we often encapsulate these customer checkpoints discreetly into what we call the “Conversion Funnel”. Lots come in the top and few go out the bottom.
The Conversion Funnel is a useful conceptual model only in that it serves to maximize what goes out the bottom. But any purpose short of that is misguided – it doesn’t matter how many people are interested in your applesauce if no one buys your applesauce. As with any model, though, we have to operate under some educated assumptions. In SEM, we assume:
- If more people show interest in your applesauce, more people will buy it.
- If one person shows more interest in your applesauce than another person, the first person is more likely to buy it.
These beliefs are obvious (and generalizable to all sorts of fruit sauces, and beyond). And they’re the same ones, of course, that all marketing is founded upon. Nobody is going to give you money for anything if they don’t first have the intention to do so. Thus, marketing aims to create that intent in someone and nurture it until they actually give you money. There are, however, some pitfalls lurking in these assumptions. They can be expressed as questions of measurement.
- How do we know that somebody is interested in our applesauce?
- How do we know that one person is more interested in our applesauce than another?
Where everything is quantifiable, as in SEM, these complications double down. A number measured correctly by the wrong metric is tantamount to a number measured incorrectly. And since how you measure determines how you optimize, you might peddle insistently to someone who prefers mango
The Volume & Monetization Rate Tradeoff
The first step in effectively applying the Conversion Funnel to your SEM program is understanding the traits of the various points along it. The Upper Funnel is the wide-brimmed top half of the upside-down pyramid where potential customers enter. The Lower Funnel is the half towards the pointed tip at the bottom where your company ultimately monetizes those customers. In terms of their usefulness as points of measurement, the two halves of the funnel have inverse tradeoffs.
Upper Funnel Metrics:
· Higher volume
· Lower monetization rate
Lower Funnel Metrics:
· Lower volume
· Higher monetization rate (including 100%, if the metric is the “bottom-of-the-funnel” point of monetization i.e. “all Sales turn into money”)
*Note: “Monetization Rate” is defined as the percentage of the time a potential customer at a particular conversion point eventually monetizes
As you move from the top of the funnel downwards, the number of potential customers tapers while their presumed interest hones. For example, more people are going to fill out a web form to be contacted about applying for a loan
Think of the Average Monetization Rate as a measure of predictive power towards future revenue: the higher the rate, the more power packed. One ‘Program Enroll’, then, holds a lot more weight in this department than does one ‘Contact Form Fill’ (by, on average, a factor of 100). Similarly, one ‘Loan Application’ averages 20x the predictive power of one ‘Contact Form Fill’.
On the other hand, the absolute Volume numbers can be viewed as a barometer of the statistical significance of each conversion point: the bigger the sample size, the more likely that information is to tell us something about the future.
For example, let’s say a keyword has 100 clicks. On average, I expect 10% of clicks to become a ‘Contact Form Fill’ and 0.5% of clicks to mature to ‘Loan Applications’. If we see that said keyword has just 5 ‘Contact Form Fills’, it’s probably a below-average keyword. If, however, we instead see that this keyword has 0 ‘Loan Applications’, can we confidently say the same thing? We can only reasonably expect 1 ‘Loan Application’ for every 200 clicks. So, if we’re using this conversion point to measure the worth of that keyword, we’re going to have to wait until we get more clicks before we draw any conclusions. Responsibly untwining noise from signals like this is especially critical when we start to think about optimization, as significant data volume is a prerequisite to all but the most sophisticated bidding solutions.
What We Measure Influences Our Results
Now comes the predicament. Assume the below keywords both benefited from the same amount of ad spend input to achieve their historical results. Which of the below keywords would you invest more money in today? Which is more likely to bring you back more money in the future?
Don’t wrack your brains too much – there isn’t an obvious answer. Doing the simple math of applying their monetization rates would suggest that these two keywords are not terribly different in value. Distinguishing further between these two keywords with any significance would require machine-learning algorithms with access to deeper contextual data.
Now let’s see how our decision-making would change if we were only to look at one metric or the other. Bear in mind that these two keywords have already revealed themselves to be of roughly equal value.
If you were to size up the success of your program using only ‘Contact Form Fills’, you would invest more money in Keyword B. You might even invest twice as much in it, ignorant of the fact that it likely won’t result in any more revenue than Keyword A. When we instead evaluate our program against only ‘Loan Applications’, the story flips.
So, if we were looking only at ‘Contact Form Fills’, we’d divert our entire budget to Keyword B. And if we were only looking at ‘Loan Applications’, we’d drown hefty sums in Keyword A. How do we reconcile that?
Nothing Is Average
Well, it’s not quite the landslide that the ‘Contact Form Fills’ would have us believe. Only when we take both conversion points together do we realize that – even though they might average out to 1% or 20% in aggregate – the Monetization Rates can differ wildly across more granular program segments. Some keywords are naturally more conducive to generating higher-funnel conversions but nothing more, while other keywords might drive better advancement deeper into the funnel. In the end, the ‘Contact Form Fills’ and ‘Loan Applications’ tell a story of two similarly lucrative keywords.
To illustrate these keyword-level disparities, let’s return once more to our table, this time adding the bottom-funnel ‘Program Enrolls’ back into the picture. Remember that these are the point of monetization in this example. And since we’ve already established that these two keywords have similar chances at revenue, we shouldn’t be too surprised to see that manifest in their ‘Program Enrolls’ totals.
While the average monetization rates for the entire program are 1% and 20% for ‘Contact Form Fills’ and ‘Loan Applications’ respectively, these keywords are just two of many that go into that average. In actuality, Keyword B is great at getting potential customers just over the doorjamb while Keyword A caters to a more committed bunch.
While this may seem an extreme example, it’s hardly far-fetched when you consider a program with thousands or even millions of keywords. In the most simplistic case, consider two keywords with the same number of ‘Contact Form Fills’. Should we treat them equally? Does our answer change if we happen to know that one has a few more ‘Loan Applications’ than the other?
Each and every keyword is unique: they will all defy the average. The goal of any measurement is to understand those differences, using all the signals available and discarding all the noise. The goal of any optimization is to react to that.
Learn how to create Hybrid Conversions for a more revenue-oriented optimization in “Hybrid Conversions, Part 2: How You Can Utilize the Whole Conversion Funnel”. Coming soon.