What Analytics Solution is Right for You?
Web measurement is a very hot topic today, with lots of solutions to choose from and even more buzzwords to pepper into presentations for stakeholders. Questions such as “when will we outgrow Google Analytics?” and “what do we need to track?” are common questions for companies.
Luckily, there are a few high-level considerations that can help get you on your way regarding this very important decision. Below are three of the most important items to think about.
Size (and Sampling) Matters
Very often Analytics providers such as Adobe or Webtrends ask the question of “what monthly traffic do you have?”. They ask this in order to assess whether or not your company is well-suited towards their solution (and to devise pricing). However, this a bit of a fallacy. The traffic to your website is not the most important per se. Instead you should be asking “how large are the traffic sizes based on the date ranges that matter most to your company?” to help guide your decision.
The reason for this is that ‘sampling’ will begin to prominently appear in Google Analytics if the amount of traffic being examined exceeds 500,000 visits (the full explanation by Google can be found here). The implication of sampling is that the number seen in the Google Analytics interface has been derived from a ‘sampled’ set of visits – meaning that Google is a extrapolating based on a sample and may not be expressing the pure reality.
Here is a real-world illustration of how sampling can impact your company: let’s say that your company’s website receives 1,000,000 visits per year at a relatively steady pace, say about 85,000 visits per month. If it’s October 1st and your company likes to view its data on a year-to-date (YTD) basis and prefers to view this data in a year-over-year (YoY) fashion, then the number of visits you’d be looking at is 1,530,000 (85,000 per month*9 months*2 years). This would mean that sampling would very likely occur. However, if your company is satisfied viewing data on a strictly month-over-month (MoM) basis, then you’d be looking at 170,000 visits (85,000 per month*2 months), which would not produce a sampled result as it is below the 500,000 visit threshold.
It is important to note that sampling is not necessarily the end of the world – meaning, it can be alright that some of the data reported is sampled. However, it is important to call this out in reporting commentary (or on graphs illustrating the data), as it will set the expectation that results might vary if people try to reproduce them. Also, it should be noted that even the enterprise-level analytics solutions use sampling – but this sampling can be better mitigated or eliminated.
Know Thy Company
The discussion of how important sampling is to your organization leads into the larger point of what is most important to your organization for its web measurement practice? What kinds of stats do you want to track? Are they fairly standard metrics, such as visits, transactions and revenue? Or are they more specific to your business, such as new client purchases by a specific customer segment? If the answer is the former, then it is more likely that a free Google Analytics implementation is right for your company. If it is the latter, then it is more likely that a paid solution will be more suitable (Google Analytics Standard has a limit of 5 custom variables, for example).
Another important question to ask is what other platforms do you want integrated into your analytics solution? If you don’t have any platforms you want to integrate with and it’s all about the website behaviour for you, then Google Analytics will be good enough for this purpose. However, if you want to bring in your Likes and paid/organic newsfeed posts from Facebook, then an enterprise platform is probably the solution for you (note that these features might entail an extra cost for Adobe and Webtrends).
What other specific needs are really important for companies? Of course, it depends. In order to figure this out, we recommend interviewing analytics users and stakeholders in order to perform a Needs Analysis. The goal of this analysis is not to select a solution that will satisfy everyone’s analytics needs in every way (quick tip, there is no such solution). The desired outcome is to properly identify, document, group and weight each of these needs in order to match them against the various vendor feature sets. This way, the decision of an analytics solution is arrived at in a thorough and, more importantly, transparent fashion that creates stakeholder-wide buy-in.
People > Technology
You can have the best web analytics technology in the world, but if people don’t know how to leverage it properly, then it’s all for not. Ensuring a proper roll-out and training program for your existing or new analytics solution is key to reaching your company’s full measurement practice potential. Also, making sure there are clear roles and responsibilities for everything from troubleshooting, to training, to tracking will help people get what they need faster and with less friction.
For Google Analytics, there is an out-of-the-box certification available for free. Note that you’ll want to supplement this training for ‘power-users’ at your organization with further in-house or external training. For enterprise solutions, there is often a training component included in the initial implementation (and most often online resources and training packages for ongoing support).
As you can see, there are lots of considerations to run through when selecting an analytics solution. However, if you are mindful of sampling, try hard to understand your company’s specific needs and have a view towards the future in terms of getting the right people and processes in place, then the challenge of selecting and implementing an analytics solution will be more manageable.
Need a little help in evaluating whether your analytics platform is right for you? Not sure if you are tracking everything you should be on your website? Vovia can help! Just shoot us over an email and we’ll get those metrics tracked and that KPI humming.