The Issue

Businesses need to know their eCommerce Conversion Rate, this is obvious, and so it makes complete sense that they would look to their analytics for it. The problem however is that the eCommerce Conversion Rate shown in Google Analytics is not always the best for a particular eCommerce site, and many a company is mislead to believe their site is doing better or worse than it really is.

Below: Google Analytics showing Conversion Rate = Orders/Sessions

 

eCommerce Conversion Rate Visits

Sessions vs. Users

There are two common ways to calculate Conversion Rate:

  1. By Session (aka Visits)
  2. By User (aka Visitor)

Google Analytics calculates Conversion Rate Session, which means that if one person comes to a site and returns 4 times before purchasing, their personal conversion rate after their 5 Sessions is 20%.

ConversionIQ, in most instances, calculates eCommerce rate by User so when someone visits the site and returns 4 times before purchasing, their Conversion Rate is 100%.

Client X.com Example

1.28% = 14,404 Transaction / 1,125,339 Sessions

2.01% = 14,404 Transaction / 717,994 Users

In the case of ClientX.com (A Big Box eCommerce Site), because the majority of purchases on eCommerce sites are made after more than one visit to the site, calculating Conversion Rate based on Users (Unique Visitors) is the recommended approach.

Problems with using Transactions / Session

Below are several examples that expose significant problems in using Sessions as the base for measuring Conversion Rates.

Example 1: more session = lower Session Conversion Rate

When an increase in Sessions outpaces an increase in transactions (due to less qualified traffic increases, loss of qualified traffic etc.), it looks like the eCommerce rate has decreased.

i.e. (Adding Display Re-targeting to the mix – hypothetical).

BEFORE Display Traffic

1.28% = 14,404 Transaction / 1,125,339 Sessions

2.01% = 14,404 Transaction / 717,994 Users   

AFTER Display Traffic

1.24% (-3%) = 16,101 Transactions / 1, 298,339 Sessions

2.24% (+11%) = 16,101 Transaction / 717,994 Users

Here we see the site has increased orders significantly, but due to a Display Retargeting only Sessions go up, while the # of user’s do not.

Clearly, the Conversion Rate by Session and Conversion Rate by Users go in separate directions, making a win, look like a loss. Likewise, the opposite can be true, and Session based Conversion Rates can increase when the site received less Sessions (i.e. Display ads are turned off).

Example 2: Seasonal Intention

The higher the intention of customers, the more likely Sessions per User will change, either growing or shrinking, depending on the site and season: When this happens, Conversion Rate by Session is artificially deflated or inflated.

  • i.e. Christmas 2013 – Comparing November to the previous July
    • Session Conversion Rate is up 68% from July to November
    • User Conversion Rate is up 98% from July to November
    • Difference due to Holiday shoppers on the site making more return visits prior to their purchase, thus masking the degree of intention present during the holiday season.

Example 3: Tracking Issues

Often, Google Analytics will suffer from tracking issues for a multitude of reasons. One common reason is when internal traffic is accidentally identified as external.

Below: A Homepage Carousel has all its slides tags and tracked as incoming traffic.

carousel

Note: This is not a ClientX.com example.

The above example, and several other common issues can negatively impact Session Conversion Rate as Users are counted multiple times, whereas looking at Conversion Rate by User is not.

Example 4: Site converts consumers types disproportionately over time

This example is the most critical when doing YoY comparisons or trending of an eCommerce rate over time.

The easiest type of consumer to convert online is the Spontaneous. This group of people often just need certain cues and simple answers to their questions in order to get them to purchase. A good continuous improvement program will often begin with optimizing for Spontaneous types of people because that is most-likely the “lowest hanging fruit.” As a continuous improvement program matures however, efforts typically shift focus on the types of online customers that are harder to convert, such as the Humanistic, Competitive and Methodical.

Note: A good reference on the subject of online consumer types is “Call to Action” by Bryan Eisenberg. For a quick read on consumer types, visit here.

As a site is optimized for the types of online consumers that are harder to convert, Buyer Sessions (the Sessions of those that purchase) tend to increase due to fact that these type of consumers are less likely to purchase on their first visit to a website. Rather than purchasing right away, they will come back a second, third or more times while they consider the purchase, confer with others, and compare their other options (online & off).

A look at Sessions on ClientX.com

Over the course of the past year on ClientX.com, both Sessions and Users have increased by a similar ratio:

sessions_01

Even when Sessions and Visitors are segmented by Visit Count (the Session # in a user’s experience), the breakdown has not changed much as all either.

eCommerce Conversion Rate by Visit Counts Example

Above we see that YoY, we more people coming site, but their Intra-visit behavior has not changed much, so it appears to be “more of the same people.”

So, what has changed?

What’s changed is that the site is now converting more visitors thanks to converting more multi-session buyers.

Multi-session buyers are typically harder to convert, but as a site improves, these types of buyers are more likely to be “won over.”

On ClientX.com, the Conversion Rate of each Visit Count has shifted dramatically. The site is now better at converting those people who come back to the site again and again.

eCommerce Conversion Rate By Visit Count YoY Example

From above, we see those who visited more than once saw the largest increase in conversion rate, with those who visit more than 5 times now most improved at 127% higher than a year ago. This success contrast with that of the 1st time visit conversion rate which improved at a respectable, but comparatively lower level of just 31%

Note: The 1st year of the site’s Continuous Improvement Program, 1st time visitor conversion rate exceeded all others, due to “Low Hanging Fruit” optimizations.

YoY Analysis

When taking a “snap shot” in time, the importance of how Conversion Rate is calculated is not apparent, however when we start tracking conversion rates over time, it becomes obvious that the two metrics are not consistently aligned.

Below: Comparing Year over Year (YoY) Session Conversion Rates.

eCommerce Conversion Rate Overview YoY

 

The above graph taken from Google Analytics eCommerce Overview report shows a 42.64% increase in Conversion Rate according to Session-based Conversion Rate comparison calculations. While this is nice, we can’t know from this report what the Conversion Rate for Users is.

When we compare ClientX.com YoY (Sep 22, 2014-Oct 22, 2014 Compare to: Sep 22, 2013-Oct 22, 2013), we see: YoY User CR is 63.50% (2.24% up from 1.37%)which is 20% more than Google Analytics ‘s Session based Conversion rate showed.

Below: User & Session Conversion Rate differences month my month

Month

User CR

Session CR

Difference

October ’13 1.80% 0.93% 93.55%
November ’13 2.87% 1.46% 96.58%
Dec ’13 5.19% 2.68% 93.66%
Jan ’13 3.14% 1.61% 95.03%
Feb ’13 2.45% 1.32% 85.61%
March ’14 2.33% 1.16% 100.86%
April ’14 2.21% 1.22% 81.15%
May ’14 2.81% 1.38% 103.62%
June ’14 2.59% 1.22% 112.30%
July ’14 2.89% 1.21% 138.84%
Aug ’14 3.38% 1.41% 139.72%
Sept ’14 2.69% 1.33% 102.26%

Above we see that Session Conversion Rate over the past year fluctuates compared to User Conversion Rate when looked at a monthly basis.

Better Conversion Rate

While using Users over Sessions is a more stable and accurate way to look at most eCommerce Conversion Rates, there is still a lot of room for “noise” to enter the calculation.

Noise includes:

  1. Flash Traffic i.e. StumbleUpon sending a lot of visitors over a few days that do not convert).
  2. Affiliate Sites
  3. Email Blasts
  4. Social Media Campaigns (low converting)
  5. Site Sales & Promotions
  6. Personalization (showing 1st time visitors something returning visitor don’t see)

In order to obtain a stable Conversion Rate, it is often necessary to filter out volatile traffic sources, and arrive at a Conversion Rate based on Qualified Visitors (those visitors that appear to be reasonably interested in purchasing online).

To this end, Inflow maintains two different Conversion Rates, one called “SCOPE Filtered”, and the other called SWAY (short for Persuasion). The “SCOPE Filtered” conversion rate is based on filtered traffic as described above where as SWAY tracks the conversion rate of “Qualified” visitors. For more information on SWAY rate, visit the article published here.

Between “SCOPE Filtered” and SWAY Conversion Rates, true eCommerce Conversion metrics can be obtained and used to better understand an eCommerce site’s success.

Note: This document focuses on Conversion Rate however with eCommerce, the more important metric is Revenue per Visitor (RPV), which factors in Average Order Value into Conversion Rate. For ClientX.com, RPV is seen by Inflow as just as much a success factor as Conversion Rate.