Excellent Analytics Tip #8: Measure the Real Conversion Rate & "Opportunity Pie"
NOVEMBER 12, 2006
The topic of my speech at the E-consultancy Online Marketing Masterclasses 2006 in London ( sign up here ) is "Conversion Rate Optimization: What, Why, How" While working on one of the slides (Tip # 9) the realization dawned that we measure conversion rate rather sub optimally and in a way that grossly overestimates the improvement possibilities. This post will cover how you can improve your Conversion Rate in ten minutes by doing nothing more than applying simple math or by doing some really amazing investigative work you can figure out what your Real Conversion Rate (TM) is and help your company figure out what the actual size of the opportunity you have on your website to convert Visitors. Even though we should not obsess about conversion rate we do. Mostly because short term goals drive a lot of what we do and if you are selling something on your website then it only seems to make logical sense that we measure conversion rate and get it up as high as we can as fast as we can. Another reason is that we over estimate what is possible. Stepping back for a moment here is the definition of conversion rate that is recommended: Conversion Rate = Outcomes / Unique Visitors. [For exhaustive details click here. And no that is not Visits or Total Visitors in the denominator, for my perspective on why that is sub-optimal click here and read the part about Why use Unique Visitors.]. So a random ecommerce website gets 1.075 million Visits from 768,000 Unique Visitors. That results in 16,500 orders giving the website a nice little conversion rate of 2.1%. Seems really low (not really because last quarters shop.org’s study pegged the industry conversion rate at 2.2%). You show that to a Director / VP / CMO and their first thought typically is “ how the heck is it possible that our website is so bad that we can’t get more of 751,500 (768,000 – 16,500) people to convert on our website ” Unique Visitors to Website: 768,000. # of Outcomes (Orders/Leads etc): 16,500. Conversion Rate: 2.1 %. Opportunity Pie : 768,000 minus 16,500 equals 751,500. From there it is just a nanosecond computation that will be given to the person responsible for the website: “If you can just get another One percent of that 751,500 converted, just a measly One percent, then it would increase our orders by a humongous 46%!!!” [(751,500 * 0.01) / 16,500]. It seems very logical (very reminiscent of many entrepreneurs who head off to China dreaming of massive success by converting just one percent of the Chinese population). The problem with the logic is that is presumes that the size of the left over pie is that big (751,500). The reality is very different. The fastest way for you to improve your conversion rate is to figure out what is the number of people who are in play for even remotely being converted. Here are three suggestions on how to do that (and figure out what the opportunity pie looks like): # 1: Use Bounce Rate. Bounce rate helps identifies the % of traffic to your website that “leave instantly” It happens for any number of reasons: wrong clicks in search results or reaction to the website or missing promotion on the page or any other reason. Bounce can be defined as anyone who was on the website for less than five seconds or ten seconds or anyone who has viewed only one page. After analysis of different websites I have recommended using ten seconds as the criteria. That’s the minimum time that someone has to commit to your website, just ten seconds , for you to even have a chance of convincing that visitor of anything. So measure Bounce Rate of your website. A “normal” bounce rate is around 30% ( YMMV ). Here is the impact on Conversion Rate: Unique Visitors to Website: 768,000. Bounce Rate: 20 %. Unique Visitors “in the play”: 614,400. # of Outcomes (Orders/Leads etc): 16,500. Real conversion Rate: 2.7 %. Opportunity Pie : 614,000 minus 16,500 equals 597,900. Using traditional computations you would overestimate the size of pie (folks you could possibly convert) by 153,600! # 2 (If you use Web Logs) Filter out search bots, image requests, 404 errors, website monitoring software "visits" etc. Some analytics professionals are still using weblogs as the source of data for doing Web Analytics. This might be sub optimal for many reasons (more on this one of these days) but one of the primary reasons is that web logs will natively incorporate the visits by search robots (and even if we try to filter these out it is hard because new ones pop up every other day). This inflates visitor counts, if not filtered. Web logs can also contain “visits” by various site monitoring software and inflated number of page views (404’s and spurious requests you might have on your website like dll’s and css and on and on) which will not allow you to compute bounce rate accurately. So do your best to filter all the above stuff out. Normally this can take out between 10% to 30% of your visitor data (depends on many factors so YMMV ). Here is the impact on Conversion Rate: Unique Visitors to Website: 768,000. Crud Filtered Out: 25 %. Unique Visitors “in the play”: 576,000. # of Outcomes (Orders/Leads etc): 16,500. Real conversion Rate: 2.9 %. Opportunity Pie : 576,000 minus 16,500 equals 559,500. Using traditional computations you would overestimate the size of pie (folks you could possibly convert) by 192,000! # 3 Use Customer Intent. One of the biggest mistakes business make is thinking that every visitor to the website is fair game, conversion fodder. Close your eyes and imagine walking into a car dealership. You are greeted by a car sales man whose only objective is to do whatever he/she can to sell you a car today (not even tomorrow, today), mostly because they are paid on commission. The problem is that you are there just to look at the car, maybe take it for a test drive. You have not yet saved up enough to buy a new car. You really don’t want to be sold. To the guy in a suit that does not matter, he is going to bring it on rather than focusing on what you want (and it is not his fault, incentives drive weird behavior). Similarly not every Visitor to your website is there to buy and not very visit by a visitor is a opportunity to convert. Yet we do Path Analysis and measure conversion rate they way we do because we behave like that car salesman. Using Market Research or Website Surveys or other methods attempt to compute why Visitors come to your website (I call this Visitor Primary Purpose). Then segment out Visitors who say they are there to 1) Buy and 2) Research (/learn about your products and services). You’ll find other segments of people who come to your site looking for support or company information or jobs or register their products or update their contract or check status of their orders etc etc. You should be converting 100% of #1 (Purchasers). You’ll never hit 100% but work hard to create a frictionless process for them. You should convert a good amount of #2’s (Researchers / Shoppers), not all want to buy but they are fair game. Normally you’ll find that around 15% of your site traffic is there to buy (Purchasers) and around 20% is there to Research / Shop. Giving you a total of 35% (again YMMV , but I promise you this number is half of what you think it is for your site). Here is the impact on Conversion Rate: Unique Visitors to Website: 768,000. Potential Visitors who can be Converted: 35 %. Unique Visitors “in the play”: 268,800. # of Outcomes (Orders/Leads etc): 16,500. Real conversion Rate: 6.1 %. Opportunity Pie : 268,800 minus 16,500 equals 252,300. Using traditional computations you would overestimate the size of pie (folks you could possibly convert) by 499,200! Let me express that another way: When your Director / VP / CMO asked you to improve conversion rate you were overestimating the amount of Visitors you could possibly convert by half a million ! Something that to the CMO looked “so easy” for you to accomplish is a nothing short of walking on water for you. I am exaggerating a bit of course, but not by that much. You can now imagine that reality of the world is a little bit worse. # 1, # 2, # 3 above are not silos. So some Purchasers and Researchers will Bounce off you site because they landed deep into your site and on the wrong page. You can imagine more scenarios like this. Which means the real number of visitors who had the right intent and who stayed long enough to give you a chance to convert them is small. Now you can easily see why we all work hard on analytics and marketers work so hard on content and copy and offers but we can’t seem to move the conversion rate by all that much. The reason is that our denominator is incorrect (in our standard formula conversion rate equals outcomes divided by unique visitors). Recommended Actions: Measure conversion rate as everyone else does, just so you have it and then ignore that number because it is deeply misleading. Measure your Real Conversion Rate (TM) by applying the above recommendations (or use your own because you'll have unique things that apply to your business). Undertake a repetitive education program in your company to educate your decision makers the size of the real opportunity on your company’s website. Segment the Visitors in the Opportunity Pie to identify what their true levers are (in getting them to buy). Focus on the Why (use Surveys or Lab Usability or Experimentation & Testing for example). Sit back, relax and enjoy the ride because you have achieved the pinnacle of Web Analysis! : ) [Oh ask for a bonus, Jim this one's for you!]. What do you think? Is Real Conversion Rate metric a good one? Do you already use it and this is old news? Do you have suggestions on other ways in which to compute the Real Conversion Rate? Any flaws in the above analysis / recommendations? Please share your feedback via comments. [Like this post? For more posts like this please click here.]. Excellent Analytics Tip #8: Measure the Real Conversion Rate & "Opportunity Pie" is a post from: Occam's Razor by Avinash Kaushik. Advanced Analytics Analytics Marketing Tips Search Engine Marketing Web Analytics Web Insights Web Metrics