CRO is Not a Marketing Channel

We reached out to one of our colleagues and friends, Francis Teo, eCommerce Conversions Strategist at Conversions Guaranteed to ask him about his thoughts and tips on the subject at hand.

Q: Francis, I asked you briefly on Twitter whether you thought CRO should be considered a marketing channel and you immediately answered a definitive NO. What can you add to that?

Francis: Conversion Rate Optimization improves upon the performance of any marketing channel, but is not a marketing channel by itself. The confusion comes about I guess since a lot of these agencies and “experts” throw around buzzwords in order to impress potential clients, but don’t really understand what it is about. Other buzzwords are “big data” and “growth hacking”. All these are good things, but we need to understand what they are and whether they are suitable for our needs.

Q: What types and kinds of conversions are there and what priority should they be given by ecommerce owners?

Francis: Generally for eCommerce we have the purchase event by a customer defined as being a conversion. Some other metrics that can be measured are the Average Order Value (AOV), Revenue Per Visitor (RPV) and also the click through rates of the intermediate pages in the funnel.

If you want “one metric to rule them all” in order to determine whether eCommerce CRO is working, then RPV is it. A higher conversion rate might still result in less revenue and profits for your eCommerce store.

Take for example if you slash all your prices by 50%. Your conversion rate will likely increase significantly, but your revenue will drop. Or perhaps if you do some testing with the increasing the threshold for free shipping and discover that people are ordering more (increasing AOV), but your conversion rate falls because less people are completing the purchase process.

Mathematically, RPV = AOV x CR

RPV measures the performance of your website, and is the one metric all eCommerce store owners should be looking to optimize for.

Q: Are the kinds of conversions you can track with Google Analytics sufficient to improve CRO?

Francis: Yes, the current version of Google Universal Analytics is extremely powerful and should be suitable for the vast majority of eCommerce stores. In terms of metrics it can track sales and revenue, and you can create funnels to show the drop offs at each part of the funnel.

Two quick notes here:

  1. RPV is visible by default in GA. You can simply multiply CR and AOV to get the RPV number. Alternatively the metric known as “Ecommerce per Session Value” is available as a custom report.

  2. If you are using A/B testing software, you want to be sure to send the information on each variation into GA in order to do segmentation on the data.

Q: What is the standard approach to CRO? What is yours?

Francis: There is no standard approach, each CRO agency uses their own framework, and we’re continually improving ours.

The basic framework that any good CRO should use is:

  1. Quantitative and qualitative research
  2. Create hypothesis based on this research
  3. Create a test based on this research
  4. Run the test
  5. Analyze the test for statistical and test validity
  6. Implement the treatment (if we have a winning treatment)

While this is a great general framework, I now believe that beyond frameworks, it is more important to have a mindset of continuous and never-ending improvement when it comes to CRO. Way too many people get into A/B testing and CRO thinking it’s a “quick fix” or there are “hacks” and “tricks” they can use to double their conversions or something.

Sure, this might work over the short term, but over the long term, having the right mindset to guide your decisions is the most important. CRO is not something that you should stop doing, it’s something you need to continuously improve on as your customers and market change, and your competition changes as well.

Q: What time frames and number of visitors do you need to get valid insights into testing and CRO?

Francis: We use a ballpark of 10,000 unique visitors a month and 12 months to get a treatment that performs better. This might involve several tests. Remember that not every test is a “winner”, but as long as the data is good, the test is statistically valid and otherwise free of validity threats, we can always learn something from the test and do better next time.

Mathematically there are test duration calculators available freely online that can tell you how long each test will take to run.

Q: From your professional experience, what kinds of improvements are realistic? How much of a role do the market you’re in play into CRO?

A realistic goal is a 15% increase in conversions over a year. However, this is a generalization. Some stores will be able to improve on their conversion rates significantly (50% or more) in just a few tests, some stores might take much longer.

The type of market you are in does affect the improvements that you can expect. We believe that improving the value proposition (and the communication of the value proposition) as one of the big drivers in improving conversion rates. In markets where you are selling mostly a commodity, it might be hard to find that differentiating factor. However, if you’re selling a unique product or type or product, this is something that can be more easily highlighted (in general).

There are always exceptions. For example, Rackspace competes in the very competitive and commoditized web hosting market and they have differentiated themselves in having superior customer service.

In general, it is possible to improve conversion in every market, on every website. It might just take a longer time for some. Having the mindset of continuous and never-ending improvement to website conversions is the way to long term CRO success.

Well, thanks, Francis. This not only answered our initial question but also gave us a valid basic understanding of what can be expected from working continually on CRO in the eCommerce space.

Francis Teo can be reached on Twitter (@TheFrancisTeo) or if you’re looking for eCommerce Conversion Rate Optimization help, find out more at