If you want to evaluate the effectiveness of your marketing channels, then you have to know how to calculate conversion rate and customer lifetime value. We’ve talked about these metrics before in our article “Estimating Trade Show ROI”. In this article, we’ll delve into how to calculate these numbers more generally (for use in any marketing channel) and look at the reasons why marketers often fail to make effective marketing models using these numbers.
What is Conversion Rate?
Conversion Rate is the likelihood that a lead will convert into a customer. Sounds simple enough, but before you can calculate conversion rate, it’s also important to define for yourself what you consider a lead and what you consider a customer. For example, your definition of a lead could be a website visitor or someone who filled out a form on your website. The former is less likely to make a purchase than someone who has actively supplied you with contact information. Similarly, a customer could be defined as someone who signed up for a demo, made a purchase, or actually was billed for that purchase (which may happen sometime after the purchase).
Mapping Out the Purchase Funnel
Instead of focusing on producing one metric, it is helpful to actually map out the stages of the buying cycle. When you have that, apply a drop-off rate for each stage. What percentage move on from one stage to the next? For a SaaS company, the calculation might look something like this:
Simply multiply those numbers together to get a conversion rate for this purchase funnel. With a good map, you can adjust your start and end points accordingly depending on the channel.
If you run software like Google Analytics, you can create these types of funnels fairly easily. Map the different screens that indicate someone is moving down the purchase process to your conversion funnel (e.g. the thank-you screen might mean that a sale has been closed).
Once you calculate conversion rate, then the next step is to validate it. Take a period of time where you have measurements available. Plugin those numbers into your model to see how well your conversion rate mirrors the real world. Your goal is to produce a simple model that is accurate most of the time. A useful simple model trumps an unwieldy model with a high level of accuracy.
What is Lifetime Value?
Lifetime Value is the total value that you would get from acquiring a new customer over that customer lifetime. Like conversion rate, you need to define, for yourself, what this means to you. For example, a customer might drive additional value by introducing other customers to your products or brand. You may also want to consider things like future upsell opportunities as your company releases new products or features. Similarly, customer lifetime is the average length that a customer engages with your business. This can be done through things like purchases, renewals, or subscriptions. Customer lifetime value is highly dependent on churn (the percentage of customers who don’t come back)
If you run an e-commerce company, you might calculate customer lifetime value by looking at the amount of an average sale and multiply that number by the number of sales that you might make over the course of a customer relationship. For a SaaS company, this might be the annual subscription multiplied by the average number of years a customer stays with you.
How to use Conversion Rate and Customer Lifetime Value
Let’s create a hypothetical scenario to illustrate how one might use conversion rate and customer lifetime value. Say I ran an online store that sold belts and one out of every two hundred site visitors would make a purchase of $35. Later on, those same customers, on average, would purchase two additional belts over the next two years at $35 each.
Conversion Rate = 1/200 = .5%
Customer Lifetime Value = $35+$35+$35 = $105
Here, the value of a site visitor to me would be ($105 * .005) = $.53
As a marketer, I should be willing to pay, at most, 53 cents, for each visitor to my site. Anything less than 53 cents is profit margin (yes, I am somewhat glossing over the fact that there are costs to running such an advertising program – but to an extent, those costs should be fixed unless your marketing program is really small…).
Not All Customers Are The Same
Some sources of customers are going to be inherently worth more than others. Some sources of traffic will have different conversion rates or customer lifetime values. Let’s assume that my belts were popular with cowboys and each cowboy was twice as likely to make a purchase. Additionally, the average cowboy would purchase ten(!) belts over the course of our relationship. In this case, my metrics would look more like:
Conversion Rate = 2/200 = 1%
Customer Lifetime Value = $35+$35+$35+$35+$35+$35+$35+$35+$35+$35 = $350
In this case, the value of a cowboy to me would be ($350 * .01) = $3.50
If I could focus my advertising on cowboys, I would be willing to pay almost seven times more for that traffic!
Hopefully, this illustrates that conversion rates and lifetime values aren’t static. These numbers are actually quite fluid and will vary depending on the channel. I recommend grouping your channels by customer type (if possible), marketing channel, and stage of the purchase cycle (is your campaign driving awareness or the final purchase). This does mean that you have to calculate conversion rate and customer lifetime value several times over – so break apart your biggest channels first (where you either spend the most money or get the most traffic) and continue to add segments over time.
Avoid the Fail – Pitfalls and Traps
As a marketer, there is only one question that truly matters – did you move the business forward? Here are a few ways people fail when building these models:
The Math is Hard
There’s a reason we didn’t become accounting or finance majors
Make simple, useful models that can be plugged into a spreadsheet and stay away from complex formulas. You need to understand every part of your model. You need to be able to adjust the model when you see differences between expected performance and reality. Simple is usually better.
Attribution Models Hide Performance
Attribution models primarily discount the value of marketing activities (to prevent things like double-counting. This happens when you give multiple marketing activities credit for a sale). More often than not, how attribution is distributed is usually influenced by political factors within the organization. That is, if you work at an organization that is heavy on brand-building activities, you’ll usually find that people want to give less credit toward more direct channels like Pay Per Click (PPC) advertising. They might argue that without those brand-building activities, that Pay Per Click campaign wouldn’t matter.
As a marketer, you need to recognize that your attribution model may be taking value away from your marketing activities on a somewhat arbitrary basis. You still need to find ways to experiment and determine whether or not your activities are driving the business forward.
Customers Behave Unpredictably
The path to purchase is rarely linear or predictable. The triggers that push someone toward making a purchase may not also work the same for everyone. For example, when some people view online advertisements, instead of clicking on the ad, they might perform a search for the product or company instead (a search that would have never happened without that online advertisement). This sort of counterintuitive behavior does not show up well when you are evaluating the performance of your online advertisements.
Another important factor is the time that it takes between awareness and purchase consideration. If you sell B2B products, then the time between awareness and purchase may extend for months or even years. This may be based on things like the approval chain necessary to procure that product. When you are calculating your conversion rates or customer value, you may need to build in a time delay to properly evaluate effectiveness. Assuming you run continuous campaigns, you may want to evaluate performance metrics against a period that may be several months prior with the understanding that: a) the buying cycle takes time and b) past performance might not be the best indicator of future performance, but current performance is probably worse.
Conversion Rates and Lifetime Values Change
Seasonality is a factor for almost everyone; things like weather patterns (not kidding), summer vacations (Europeans and the whole month of August!), and the presence or absence of competitors are among the many different factors can really move things around. How this seasonality actually appears in your marketing reports may be hard to determine. For example, if you sell a software service for posting jobs and managing HR, you might see a huge lull in December because the hiring managers have typically done all the hiring for that year, budget for new personnel isn’t quite available yet, and they are busy planning their winter vacation.
You can’t rule out the impact of seasonality, but making micro-adjustments to your marketing campaigns may cause more harm than good. Making small changes that are hard to measure tend to cause a drift in performance. One way to address seasonality is to smooth out your conversion metrics over a longer period of time. That is, take an average that spans several months instead of one particular week or month. A tactic like this has the added benefit of also supplying you with more data points. With enough data, you can effectively determine whether a channel is really working or not. The ideal time-frame might be different for you depending on your industry and how big the seasonality impact is. A three-month window is usually sufficient – too much more can make it difficult to detect poorly performing campaigns early.
Each Channel Has Different Metrics
We discussed this earlier, but let’s revisit it to see how this might work in practice using search marketing. In search, your conversion rates and lifetime values might change at the keyword level. If someone searched for “new cars” they might be at the beginning of the buying process. “New cars” is a relatively generic term with a lot of volume and competition (expensive). If someone searched for “mid-sized SUV”, they might have a stronger idea of what they want. They would be just a little closer to making a purchase. There will also be less competition (car manufacturers and dealerships that don’t sell mid-sized SUVs won’t bid on those terms). Finally, if someone searched for “Mazda CX-9” – they are probably closer to making a buying decision.
The reality is that it is often impractical to calculate conversion rate or lifetime value down to the keyword level for all keywords. Beyond a certain point, there isn’t enough data to determine or understand what is happening. Your sample size is too low to be useful. In these cases, you should group marketing campaigns or marketing channels into more manageable chunks based on the type of audience you are attempting to reach, how many other marketing channels they are likely to touch, and the relative stage in the buying process. As you might have guessed, distilling multiple channels into useful “groups” tends to be more art than science.
You Can Do It!
We’ve shown you how to calculate conversion rate and lifetime value. Hopefully, you can see that the math isn’t hard and you can make really useful marketing models with it. When you do it right, you can prioritize your marketing channels and adjust as needed. If your company released a new product that boosted customer lifetime value then, with the right roadmap in place, you would already know which channels warrant additional investment and would be able to take full advantage of the new opportunity.
Are you trying to measure the value of an upcoming trade show? Download our Trade Show planning guide to take these principles and apply them to your event program.