Across a 12 month sample of 300 websites and over 40 million sessions we found that in the UK, the average website conversion rate is 1.2%, while the best boast a conversion rate of over 30%
I wrote this piece because at seminars and events, I'm often asked "what's the average website conversion rate?".
Once I'd worked out the average conversion rate of UK websites, I realised that this information begs further questions: what is a better than average performance? And, what is the best website conversion rate I can aim for?
Time to drill further into the data to find out what are the "best in class" conversion rates for UK websites.
The best UK websites convert at 34%!
Looking at the top 10% of the websites analysed (I'm going to call these "best in class"), the vast majority are run by e-commerce, B2B or Professional Services businesses.
Before we dive headlong into the possible reasons for thier better performance, lets refresh our memories of that sales and marketing favourite, the marketing funnel.
Best in E-commerce
Approximately 50% of the best in class websites fall into the retail category. Of course, I don't have visibility of their wider marketing efforts but Google Analytics provides clear evidence that most of these are driving visitor traffic via search results using Google AdWords.
I think, therefore, I can reasonably hypothesise that the marketing brains behind these sites are targeting "bottom of funnel" keywords because, as everybody with an iota of sense knows, these will have the greatest chance of converting to a successful check out.
By choosing to bid on phrases where the search intent is in the decision phase of the marketing funnel, the best performing e-commerce sites are maximising their PPC spend by driving visitor traffic to highly optimised landing pages, which is why so many of them feature in the best in class section.
Best in B2B & Professional Services
Most of the other best in class websites improve conversion rates (and eventually, sales) by delivering useful information to the prospective client's email addresses as part of the sales process. Top perfroming B2B and Professional Services websites count content downloads, demo requests, webinars and newsletter sign ups as a conversion.
It stands to reason that because these conversions typically take place in the awareness and/or consideration stages of the marketing funnel, you'll see more of them. Sometimes it's hard to see why more of these top of funnel conversions generate more sales at the bottom but the effectiveness of this method is why Inbound Marketing has gained such a foothold in the marketing industry.
You see, one or two top of funnel conversions by a prospect during a considered buying process, will make it much more likely that they will subsequently return to the website during the decision making stage: your website helped them identify what their problem is and offered advice on some of the methods to fix it. In doing so you began to nurture their trust.
When visiting your website again during the decision making stage, should the prospect be interested in taking things further with your brand they'll likely convert again via a bottom of funnel conversion page such as your contact us page.
About the websites in this sample
I took a sample set of 300 UK websites, all of which were using Google Analytics.
For obvious reasons, I've kept all data anonymous. However, I chose these particular sites because the overwhelming majority are owned by organisations which are classed as SMEs or Micro Businesses (according to the government's definition in this PDF).
47.7% were not measuring conversions using Google Analytics goals or E-Commerce tracking, so I disregarded them. How the marketing / sales teams in charge of websites that don't measure conversions keep their jobs in the 21st century, I'll never know.
Calculating the average conversion rate
Outliers were identified and highlighted using Excel by calculating and evaluating Z scores as explained by the very clever Dr Todd L Grande here: https://www.youtube.com/watch?v=TtxBF4BZzl8
Standard Z outlier value: 2.68.
Mean (before removing outliers) = 2.9%
Mean (after removing outliers) = 1.2%
OK, so I got rid of nearly half the sites before analysis because they weren't tracking goals or e-commerce conversions. Should be easy now, right? Wrong.
A decent number of sites which are tracking goals in Google Analytics didn't have any conversions. For a whole year. While I was (and remain) sorely tempted to go shake some marketers roughly by the lapels, the more pressing issue was figuring out how I could standardise data where many of the values were zero.
This exercise was made harder because I didn't really concentrate in Maths at school and my only further education was a National Certificate in Agriculture (Animal Husbandry). Plunging shoulder deep into a sickly bovine is something for which I'm both qualified and gifted - identifying outliers using the Interquartile Range, not so much.
They chewed their pencils and scratched their heads for a weekend or so and decided that, as a result of there being no normal distribution (the majority of the data was zero, after all), we needed to set aside the basic Tukey's Test method (whereby an outlier is defined as any data point that is more than 1.5(IQR) above the upper quartile of data, or less than 1.5(IQR) below the lower quartile of the data).
By now I was totally baffled but remained unable to convince them that "just highlighting the cells and then reading the AVG of the bottom right hand corner of Excel" would stand up to scrutiny.
Because of our large proportion of zeros in the data set, Gertie (and Gertie's dad) liked the Z score method of identifying outliers.
In the final analysis, Gertie (and I presume Gertie's dad but by now, quite understandably, drink had been taken) decided that any Z score above 2.68, or below -2.68 were outliers.
Of course there's an argument for using the Z score of 2.5 but as I say, drink played a part.
This is the first time Noisy Little Monkey have dug quite this far into statistical analysis of this kind so if you have questions, concerns about our calculations or think you can do a better job with the core data, please get in touch - we love to improve!