For the past 5+ years, Google Analytics has been my go-to web analytics tool. While it isn’t as robust as Omniture or WebTrends, the pricing model (Free!) and easy installation & simple configuration processes have resulted in widespread adoption. According to the technology adoption and trends tool BuiltWith, Google Analytics is used by more than 44% of the top 1 Million website. In comparison, Omniture is used on only 1.35% of the top 1 Million websites.
I’ve been spending significant time with Google Analytics over the past few months and I seem to be running into more limitations each and every day. Here are the 3 things that I would expect to be able to do with Google Analytics (but I can’t):
- Advanced segmentation isn’t really advanced. After I wrote my post about about the robustness of advanced segmentation, a colleague asked me to help her with creating a report on user behavior using internal and external visitor segments. Knowing that I’ve set up Views where I’ve excluded traffic using an internal IP address, I began searching for the “Filter by IP” field but could not locate it under any of the available options. Once I realized that this was not possible, I resolved to creating the 6 essential Views that we set up with all client accounts so that she could report on this behavior in the future.
- Custom Dashboards aren’t really customizable. I met with another colleague before the Thanksgiving holiday break to discuss the buildout of a custom dashboard. He sought my help with creating one since I had built a custom dashboard for my blog. We attempted to create a basic widget that displayed on a daily basis the total visitors vs. new visitors vs. returning visitors — a graph that a marketer would commonly look at. While the visitor metric could be added to the graph, we couldn’t add the visitor type dimension. To work around this limitation, we added the % new visitor metric. Furthermore, we could only add a single metric!
- API’s aren’t really open. I add an annotation every time that I publish a new post on this blog. Since my blog posts get automatically pushed to social media networks when they are published, I began investigating whether I automate the process and cut the manual step of adding these annotations. I looked into the Google Analytics API’s and while Google has multiple API’s — API’s for data collection, account management, and data reporting – none could be used to create annotations on the fly.
Google does have a support forum where new features can be requested so I posted my suggestion for automating annotations. While the features that I want are not currently available in either the standard or enterprise version, I’ve noted that Google has invested significantly in the platform this year. I fully expect the product development team to continue to release more valuable functionality in mid to late 2014.
What features do you think are missing in Google Analytics?
The American Marketing Association – Atlanta chapter held a special panel on Big Data yesterday. Big Data is a hot, albeit overused, term at the moment. The panelists, which represented big data leaders (they didn’t want to be called experts!) from well-known brands such as The Coca-Cola Company, Georgia-Pacific, and Google, and Definition6 (a local, independent ad agency) were tasked with defining Big Data and explaining how it can be used in today’s marketing organization and how it will shape future marketing efforts. The speakers at the event included:
- Julie Bowerman – VP eCommerce, The Coca-Cola Company
- Tom Lowry – Director, Google
- Douwe Bergsma – CMO, Georgia-Pacific
- Michael Kogon, CEO, Definition 6
Here are the key highlights from the event:
- Big data is the insights obtained from the analysis of disparate sources of data — a sea of data!
- Big Data is still in the infancy stages. Marketers should abandon their belief that it will drastically change their business in the near term.
- Marketers shouldn’t expect to uncover big insights without first crafting a strategy. As part of their strategy, they will want to define assumptions that they want test via data — data will either support or disprove their assumption (this was a key takeaway that I previously discussed).
- To get started with big data, you need to prioritize which data sources you’re going to look at first. You’re also going to need a marketing technologist — part scientist, part General and part storyteller — to inform the next big marketing initiative!
- The brand, not a third party, should always own the data!
Ken Bernhardt did a great job moderating the event and I’m looking forward to seeing how this conversation changes in the next 12 months.
What insights have you come across with Big Data?
This week I received an email from a client that discovered their website traffic suddenly plunged. Their Google Analytics visitors dropped by more than 50% in August and the situation never self-corrected. My first response was: DON’T PANIC! Since the issue was only 60 days old, I knew that we could identify the problem and suggest an appropriate solution.
I used a simple strategy (3 questions) to diagnose the situation:
- Has anyone changed/modified the Google Analytics tracking code? These kind of sudden changes are typically due to a configuration change. The client reported that their developers inadvertently reused the tracker code from the blog on their corporate website. Since this error was corrected in the end of August, I first investigated this potential cause. I applied a custom advanced segment to exclude visits from blog pages but the sudden downward shift was still present. This meant that the problem was unrelated to a tracking code change.
- Looking 60 days prior to and after the change, what sources referred traffic to the site? Oftentimes, traffic will fade away when an ad campaign ends so I looked at the Acquisition > All Referrals section of Google Analytics. The report showed that the majority of traffic was direct, yet the remaining referral sources didn’t dramatically shift from one source or site to another. This meant that I was potentially getting closer!
- Looking 60 days prior to and after the change, what content received the most traffic on the site? An alternate reason for a shift is that an important page is removed or modified. I looked at the Behavior > Site Content > All Pages report and discovered that a careers page dropped from the top 3 position to the bottom 20 position. This change occurred right around the time of the sudden traffic drop. Since a careers section typically draws 30%+ of the visitors to a corporate site, I contacted the client about this change and they explained that their recruiter recently left the company. They added that she was instrumental in referring potential candidates to the site. BINGO!
- When it comes to analytics, there’s no substitute for formulating a strategy and then digging into the data. Insights are rarely immediately apparent!
- The best approach to this kind of analysis is the scientific method: develop a assumption, analyze the data, and determine if the data supports or disproves your thinking.
- Don’t be afraid to be wrong. You may not get the answer that you were hoping for the first time!
What other factors could contribute to a traffic drop? What reports do you review to understand the cause of a traffic drop?