Editor note: this article was contributed by Avinash Kaushik, the Analytics Evangelist at Google, blogger at Occam’s Razor and Author of two great books: Web Analytics 2.0 and Web Analytics: An Hour A Day. He is also the he Co-Founder and Chief Education Officer for Market Motive.
There is a reason your analytics data does not make any sense. There is a reason you are unable to find an iota of insight that you can action. There is a reason you feel data is your enemy.
You see, you are not following one of the holiest of holy covenants when it comes to data analysis: Segmentation.
Perhaps because web analytics tools make it easy for us to access data about our customers and our websites we simply jump and start reporting as much data as we can (a process I lovingly call “data puking”). All data reported by Google Analytics, Omniture, WebTrends et al is aggregated data. Yes all that gorgeous beautiful data lumps everyone that comes to your website, everything they do into one ugly and messy glob.
Yet if you simply sit and ponder for a few minutes it will be clear that your website exists to do many different jobs and people come to your website to accomplish many different goals. So why analyze your data as one big ugly glob?
For example your website exists to do ecommerce, to accept job applications (for those that can actually find the hidden link!), to collect leads (for future “marketing promotions”), to help people find your stores offline, to connect visitors to your social media channels, to ….. so many things. How do you tease out where you are doing well and where you stink? Segmentation!
If you take the visitor view I might be there to buy something (so you wish!), I might just want to read something to help me buy your product on amazon.com, I might want to get tech support (also impossible to find by the way on your website!), I might want to forward a particular page on your site to my friend, I might want to subscribe to your world leading blog, I might want to apply for your loyalty blog, I might want to…. so many things. So why not try and understand all these different behaviors and intentions differently? Segmentation!!
The best ideas for taking action come from the process of segmentation. Put simply it is taking the entirety of the data on your website and breaking it down into meaningful chunks. Knowing your overall conversion rate is 2%, useless. Knowing the conversion rates of your main acquisition channels are: Paid Search: 9%, Direct: 6%, Affiliate: 0.58%, priceless.
So Why Don’t More People Use Segmentation?
I think there are two reasons:
All our offline existence manifests itself as a single purpose entity. You go to the supermarket to buy. You go to a movie theater to watch the latest disappointing “blockbuster”. You go to the bank to deposit money (or well after you have already decided exactly where to open an account). So on and so forth. Hence we have always been obsessed with analyzing a single outcome. We bring this to the web.
Problem is that people come to our website for many different purposes, driven there by many different motivations and invitations. If you can’t tease them apart you are not going to find anything of value in your data.
Many people thing segmentation is hard. Becoming really really good at anything is really really hard. But segmentation in web analytics is less difficult than you might imagine.
It is not that hard to figure out what your initial segmentation strategy should be (just spend time with your business decision makers and ask a few questions about what’s improtant) and most web analytics tools have created user experiences that make it simpler for you to start segmenting your data (without massive mastery of SQL – structured query language).
So if our online world serves multiple needs and it is not difficult to get started with simple segmentation in analytics tools, how can you really get a jump start?
Focus On Three Segmentation Strategies
This is perhaps the simplest places to start. Separate our people who come to your website from Google, email campaigns, banner ads, Twitter etc. Quite literally this is the matter of dragging a dimension (Source, Referring Site, Campaigns) on to segmentation profile and adding a condition like “contains google”. There you go. Done. Baby step taken!
Here are the kinds of questions you’ll want to answer. Is there a difference between the bounce rate for these segments? Why? What content / products do people who come via display banner ads care about more than those that come via paid search? Is there a difference between Visitor Loyalty for those that come via Twitter compared to those that come via our email campaigns?
I love this one. Different groups of people on your site exhibit different behaviors, because they had a different need / intent when they came to your website. So why not use that as your segmentation strategy?
On a news website separate people who read three pages or more? What do they like? What don’t they like? Where did they enter the site compared to people who saw only one page? On a price comparison website separate out people who visit more than ten times a month vs those that only visit twice. Any difference in products / places searched? Do they tend to be from particular regions / countries? What can you change about your business / site from their behavior?
I lied before. This is the one I love the most!
Separate people out by products they purchase. Separate people out by the size of the order (people who buy the big bling bling from people who buy trinkets!). Separate people who applied for store loyalty cards (offline buyers!) vs. people who signed up to be there for your protest against global warming (OMG!) vs. people who submitted a lead vs. people who placed a online order.
Don’t try to get down to individuals that analysis or any action you take from it, does not scale. Focus on groups of people who delivered similar outcomes to you and answer the types of questions mentioned above. Then go invest more in campaigns that delivered profitable outcomes, so create the type of content that works, promote the cross-sells and up-sells that worked the best, kill the products that are a drag, go buy better keywords for products that sell themselves, go… well I could keep going for another 5,000 words.
Pretty much all the segments you’ll apply to your data will fall into one of these three buckets. Not that hard right? Quick to get going. Just a little bit more work to away from useless glob reporting to focusing your effort in groups of visitors you can understand. And just a few more steps to take action based on your data. Action guaranteed to help improve your company’s bottom-line and earn you a promotion.
All the best!
PS: A bonus link: In case you use Google Analytics check out this post: Favourite Google Analytics Advanced Segments. It contains around 15 segments that you can start with, all you have to do is click on the link and all the definitions and meta data will be applied to your own Google Analytics tool and you’ll get started faster than I can say goodbye.