Web Analytics Deep Dive #SESLON
Part 1: Optimising ecommerce using Google Analytics – Yehoshua Coren (Founder of Analytics Ninja)
Yehoshua initially started out doing SEM for an ecommerce network and then moved into doing PPC for B2B businesses. He then opened up his consultancy in 2010.
Engagement can be broken down into three things: acquisition, behavior and conversions. These can then be broken down into other metrics.
It’s important not to just think of conversions as orders, anything that can deliver financial gain or economic value is a conversion – eg: newsletter signup, enquiry etc.
A lot of people don’t take advantage of product category, which is a great way to categorise data.
Capturing behavioral data:
It’s important that you track all of the calls to actions on your pages, by adding event tracking to navigational refines you’re able to quickly identify what people are using. You can also test changing the order of navigational refines to generate more conversions.
Yehoshua recommends testing using different products as best sellers and then using event tracking to determine if different products get more clicks.
You should be using event tracking to track what people are doing on your checkout page, there might be things that are distracting people or maybe things are helping to upsell products.
Compare people who are using site search to those who aren’t using site search – it’s interesting to see the impact on conversion rate. You can break this down further by looking at different keywords.
You need to identify all of the different things on product pages that people can interact with, then you can use event tracking to track how people are engaging with them. You can also use custom variables to track how people are engaging with product pages.
Look to book ratio: the number of unique purchases divided by the number of times that product was viewed. This is useful for looking at which products could sell more dependent on how they’re positioned on category pages.
You can capture the products that are being upsold to users by using a custom variable.
Look at the number of transactions made and how much revenue is generated. You can find out how people respond to the website once they’ve already made a purchase in the past. Yehoshua found that the average order value goes up when someone returns to the site after making a purchase.
Affiliate report: You can’t get personally identifiable information, however you can map people by how they used the website. Creating groups for customer types can be very useful – Yehoshua recommends having a wholesale and consumer report. This can be tracked using event tracking or custom variables.
Yehoshua doesn’t rate 4qs as a survey tool, however he does like that it can be integrated into Google Analytics. You can create segments based on how they found the website. You can then use this within an advanced report and look at how groups of people interacted with the website. You can also look at comparative average order duration, average order values, per visit value etc. You can also then look at the landing page, which could be attributed to why they had a good or bad experience.
The key to analytics for ecommerce is in understanding your customers – How did they get there, how did they behave, did they do what you wanted and what type of visitor were they.
Part 2: Analytics for non-ecommerce sites - Russell Sutton (Managing Director of ConversionWorks and Webexpectations)
The problem with non-ecommerce websites is that nine times out of ten is that they’re not measuring value. It’s important that you define your business objectives before you speak to anyone about analytics – this is harder for non-ecommerce websites.
Web analytics is about actionable insights, you’re getting the insight and then you’re taking action. The key thing is to understand the value, which is about far more than just visits.
For Volkswagen, they would need to identify smaller outcomes as objectives, things like book a test drive, sign up to a newsletter etc. These are good proxies for success, however they need to understand the value of the different outcomes. It’s the client who needs to identify and attribute value to each of the outcomes.
Once you’ve identified the value for the outcomes, you can assign value to actions in Google Analytics – for example, an affiliate website could assign value to external clicks.
For Volkswagen, their main objective is probably to get someone to book a test drive, so they would assign a value to this (a goal value). They would then assign values to all of the other desired actions. Once you’ve assigned these values you can start to look at measuring the value of visitors and engagement etc. They can also look to attribute value to visits from specific sources, based on conversion rate and quality of the traffic from this source.
In order to understand how valuable a keyword is, you need to be able to assign value to them, by assigning value to outcomes you can look at conversion rates, per visit value etc.
Analysing different traffic channels is really important and looking at engagement metrics isn’t enough, even for non-ecommerce websites. Once you’ve assigned a value you can start to understand how your budget could be better assigned / used. Having the value also enables you to look at your content and understand which ones are generating conversions and are adding the most value to the website.
Once you have the value figures, you can start segmenting your customers based on how valuable they are to you. You can create groups or buckets for people based on their value. You can then break this data down by content and how they got to the site to optimize conversions.