An effective web analytics framework must take into account the type and purpose of the site it is designed to measure. Whilst the key principals are shared across all frameworks there are some distinctions. Frameworks normally diverge by the following six categories – ecommerce, ebusiness (e.g. banking, services etc.), media, lead generation, support and brand sites.
The emergence of social media and ubiquity of mobile Internet in recent years created two new types of analytics (though the latter applies to all other categories). But increasingly the top categories have seen further specialisation into sub categories. One in particular is the split between retail and non-retail ecommerce.
Sibel Akcekaya of NobHill Digital Consulting will be running the Non-Retail Ecommerce Analytics session at MeasureCamp. She has kindly shared her thoughts with us about the subject.
Sibel, can you tell us a little about yourself and the work you do as a web analytics and optimisation consultant.
I have been working in Internet and ecommerce for almost 13 years. I started my career in my hometown Istanbul, Turkey. Online was still very immature in Turkey so I moved to the US. There I discovered the beautiful and mysterious world of analytics.
I got my Omniture and Google Analytics certifications and now specialise in implementations and analysis. I tend to combine web analytics data with usability testing to better understand the user experience. I also implement exit survey to keep an eye on customer feedback.
I’m a firm believer in A/B and multivariate testing. Testing is absolutely essential for continuous optimisation and is a big part of the success of companies like Expedia, one of my clients.
Do retail and non-retail ecommerce analytics differ? And if so, how?
When it comes to more advanced analysis and customer segmentation, non-retail ecommerce sites have a significant advantage. Let’s take B2B ecommerce sites as an example. B2B sites often have a smaller customer base than retail ecommerce sites. However, that customer base tends to be more selective and easier to define. You end up with more distinct customer segments offering greater personalisation and targeting opportunities. Getting more personal with your customer is one of the most important factors in increasing conversion in my opinion.
So customer segmentation is a key differentiator. How do you go about developing a segmentation strategy from a web analyst point of view?
Your first step must be consulting the product managers, online marketers, interaction designers and especially the customer analytics people. Get to understand how they segment customers. You probably want to keep to their segmentation, at least initially, in order to get the business to buy into your online segmentation. You’re more likely to get management buy in if you use segmentation that is already well established in the organisation.
Then it is up to you to define those segments online. You’ll need a measure of creativity and rely on your business and analytics acumen. Always validate with the stakeholders I mentioned above.
Don’t forget you are not reporting, you are supposed to uncover business insight and make recommendations. If you cannot find valuable insight with a certain segment then thrash it and focus on another one.
Are there key segments that web analysts should be using but rarely do?
Key segments really differ per industry and even per company. But there is one vital one that every web analyst should be using. Always segment your shopping funnel customers. If you don’t then you’re missing a big trick.
Let me illustrate this point with an example. Start by analysing your sales funnel by marketing channel. Does Google AdWords traffic differ from display campaigns? If so go talk to your online marketing person and try to understand the difference between those campaigns.
Is there anything you can do to improve the sales funnel? Maybe your landing pages are insufficient or worse – have technical issues. You might find that visitors who come from Google ads are more likely to reach the checkout pages but have a higher exit rate on the checkout than people who come from display ads.
You must form hypotheses as to why you’re seeing certain customer behaviors. Then find a way to validate those hypotheses or discredit them. Share your hypotheses with your stakeholders. They can help provide you with the insight needed to validate your hypotheses. As a side effect you will increase your standings with them. As you include them in the analysis process they are more likely to accept your conclusions.
Finally, make your recommendations as specific as possible and based them on your analysis. Avoid general recommendations such as “you should increase spend on Google ads as this is a profitable channel”. But rather tie your recommendation to the data and, as much as possible, demonstrate the potential revenue uplift expected from your recommendation.
A recommendation such as “You should increase Google ads for generic terms on this product group as it is performing well, yet, we are not maximising the potential search volumes on Google for these terms. A moderate increase of 5% in sales would yield an extra £100,000 a month at a cost of £30,000.” is far more likely to get noticed by management and actioned by them. Isn’t that the purpose of optimization? Getting things changed to increase sales and profitability.
Your session at MeasureCamp is focused on non-retail web analytics. Can you tell us more about what you’d like to cover?
I’d like to focus on segmentation much as I discussed above. We will be exploring several more examples for non-retail ecommerce and I look forward to discussing my examples with the audience. I hope to learn something new in the process which is the main reason I’m attending MeasureCamp.
Are there any other MeasureCamp sessions you are looking forward to?
Yes. I’m very interested in any Mobile Analytics sessions. This is still a relatively new topic for me. We are trying to figure out how to capture data and how to make use of it so learning what others are doing is very important for me.
I also want to attend Multi Channel Attribution by Nicolas Malo. I find marketing analytics directly tied to web analytics and again multi-channel attribution is one of the hard concepts. I am always curious to see how others are managing multi-channel attribution.
Sibel’s Non-Retail Ecommerce Tips & Tricks session is at 11:00 at the XL room
Interviewer: Michael Feiner
Position: Director, AEP Convert
Interviewee: Sibel Akcekaya
Position: NobHill Digital Consulting