A/B test for WARBY PARKER

A/Btest for WARBY PARKER

JiahaoYang

4/7

1.Elementsthat are going to be tested on are the main picture on the homepage of WARBYPARKER, the words beside this and the call to action button.

1)Asthe brand is selling glasses, which people put on their face, we may assumethat the homepage picture would be better with models wearing those glassesthan just the glasses.

2)Wordson the side doesn’t seem to be good enough. “all the hues” can be seen on thepicture, so this description is rather useless and redundant. “our best sellingsunglasses” sounds too weak in evoking people to realize how good thosesunglasses sell, as nowadays we can see products labeled “best selling”everywhere. Also, stating the price at first is suspected to be a good tactichere, as the pricing of WARBY PARKER is not their advantage.

3)Callto action button here says “shop now”, and it’s in a light color compared tothe whole tone of the website. I wonder if changing into other words and colormay have a better result.

2.Themethod of a/b testing will be adopted to test out if any of the above changesleads to a higher purchasing rate. Firstly the altered version should bedesigned and put up to the website. Then divide the traffic of people in thetest group randomly into two different version. In the end, analyze the dataand choose the best combination of all the versions. For pictures, there willbe original pictures of just sunglasses and the other version of models wearingsunglasses. For words, there will be the original one “All the hues of Haskell.Our best selling sunglasses are now available in six new flash lenses; eachstarting at $95” and the altered version saying: “Haskell: every hour one ofyour friends is getting a new pair of it, starting at $95”, or delete theprice, just say that “Haskell: every hour one of your friends are getting a newpair of it”. For call to action button, test between “shop now” and “buy”. Alsotest on some brighter colors like red or blue to see if they contribute tobetter result.

3.

According to semrush.com, latest datashows that the traffic in Feb. 2017 was 1 million. So the daily traffic mightbe around 34,000 people. I will need 15,000 users from their daily basis anddivide them randomly into 2 groups of 7,500 people or 3 groups of 5,000 peoplefor each version at a time. There are at least 5 different versions. Versionsregarding the same elements would be tested at the same time, and each of thethree variable elements would run for 3 days. So, 9 days would be needed to runthe whole test.

4.Toassess the effectiveness, two metrics play an important role here. First isaverage visit duration. Because we are changing the look of the homepage, whichis the most direct visual demonstration to grab the viewer’s attention in thefirst 5 seconds and determines if they don’t leave the page but start browsingit. Second metric is conversion rate, which identifies how much of the totaltraffic end up purchasing one of the glasses. As the homepage of WARBY PARKERis not about their brand but their best selling product, so data on the salesof this main hot product is really important.

5.Thistest is important because it may help WARBY PARKER get more consumers and earnmore profit. The goal of every business is to make profit. WARBY PARKER alreadyhave a large group of traffic on their website. And the traffic is so valuablethat it should not be wasted. How to leverage these group of people to improvetheir products and services so that they appeal to more new customers would bean important task.

With the outcome of the test, I wouldlearn that if commodities are more appealing when they got dressed on people orbeing separate just as a single object, if words like “all new hues” that youcan get from just looking at the pictures still necessary to be shown or not,if price is better to be shown or hidden at the homepage and if brighter colorsdoes better job in calling actions.

My prediction would be that things lookbetter when you see people actually using them. Concise webpage is morecomfortable so redundant words should be deleted. When pricing is not anadvantage, they should not be shown to customers at the first place. Brightercolors does better job in conversing into action, as we can see from theexample that bright traffic lights of red and yellow have already trained usinto stop or go action according to the colors.

6.Throughthis a/b test, the brand would get to know that how to converse more traffic ofpotential customers into real consumers. And the sales may rise due to the a/btest because better and more appealing version is shown to viewers. As thereare 1 million of traffic on the website every month, just an extra 3%conversion rate might lead to 30,000 more purchasing every month, and that’s arevenue of an extra $2,850,000 when we take the price $95 of each glasses foreach purchase. So just a little improvement may lead to great revenue inreturn. And the brand must care about this.

Appendix：

1.Originalversion

2.Change　of picture

3.change　of words

4.Change Of color