As noted in Occam’s Razor by Avinash Kaushik, as well as his popular books Web Analytics 2.0 and Web Analytics: An Hour a Day, the most important aspect of any web analytics initiative is getting to the actionable insights that drive strategic decisions. He says that from raw data alone, it is impossible to deduce a good strategy. To get any insight into your information, you need to segment it into useful metrics. He also says that default reporting segments are never adequate; you need to draw your own conclusions about your customers. As Kaushik puts it, you have four steps to follow:

• discover what is important from a business point of view
• create a segment tailored for that,
• apply relevant reports to spot key performance indicators, and
• take action

We’d add another step because we are firm believers that nothing stays the same for very long. So don’t build a system that is static or difficult to change. The very nature of business today means that KPIs must evolve.

• iterate

In this article, we’re going to focus on the second step, segmentation, and the challenges of iterating the process.

What is Segmentation Analysis?

Segmentation splits up various dimensions of your web traffic so you can identify visitor groups that have unique behaviours or strong tendencies. The analysis hopefully will reveal groups that read more content, download more files, buy more products – whatever the goals of your website are. Google Analytics has a wide variety of reports that let you look for segments with desirable attributes and an Advanced Segmentation Wizard that allows you to request subsets of data using a combination of filters.

What if you wanted a metric filtered by east or west coast, sometimes, and day of week participation, sometimes, and a certain set of Landing Pages, sometimes, possibly in various combinations. It starts to become a factorial problem, and you are quickly overwhelmed with segments. If you are creating reports for others to view, you will undoubtedly have additional segments based on their needs as well. Add in the requirement to iterate over time, and you will eventually find yourself in a real mess.

When analysis starts to become tedious, people cut back on exploration of the data. They stop looking for new behaviours or changes in site usage and they become complacent (or overburdened) with the status quo. Business managers rely more and more on the web analyst to be clairvoyant as to what might impact the business or where opportunities may exist. The pressure is on…

On-The-Fly Segmentation

If you use Nextanalytics, there’s no need to define segments in advance. Nextanalytic’s Page Filters reduce the size of the data before it’s loaded into a Microsoft Excel worksheet. An Excel user can use Page Filters to create and re-use segments of their own choosing. The Page Filter commands are human readable, can be altered, and can be parameterized by cell values in a worksheet. Using simple cell references, you can create on-the worksheet checkboxes that enable or disable aspects of the filter (segmentation) commands.

But there is more to segmentation analysis than just filtering. It is supposed to be behavioural analysis. Nextanalytics includes a wide array of analytics capabilities that let you create trends, group items into segment categories, aggregate and compare, and even create custom time periods on the fly. By doing these things your data is reduced and re-shaped, revealing aspects not apparent from simple counts. And Nextanalytics has been proven to work with over 500 million input records, so there is no fear of exceeding an Excel limit, even if you are using Excel 2003.

In Web Analytics 2.0 and in his blog, Kaushik places a high value on Segmentation, especially if it is dynamic. “You don’t need to say up front, you don’t have to add tags, and don’t have to submit and wait”. That is the Nextanalytics environment. Ask more questions and get increasingly precise answers.

Examples of Segmentation

On-the-fly segmenting requires you to be able to ask multiple questions and blend them in various combinations. The combination of Nextanalytics with Microsoft Excel provides a number of ways to accomplish this, allowing individuals to use the method they are most comfortable with. Excel and Nextanalytics are empowering tools, so we give you as much flexibility as possible to get your job done as quickly and easily as possible. For example, Nextanalytics can be used to simply load the data in columnar form, and then use Excel’s AutoFilter feature to select subsets of the data.

If the dataset is really large, it may not fit into the version of Excel you are using, so the same type of filtering can be achieved interactively using Nextanalytics Page Filters.

Nextanalytics also offers the ability to script the action for inclusion in a report – automatically including the filters when the data is refreshed.

What Makes Us Different

With Nextanalytics, you work interactively in Excel. You are not fumbling with a database query language or editing filters as your understanding improves. Nextanalytics operates iteratively – the results of each step become the input to the next question (automatically). When you have a result that you’re satisfied with, you can save your actions in a human readable and editable worksheet, making it easy to distribute to others. You can edit, join, parameterize, and re-arrange the command scripts as your needs evolve.
It is easy to create worksheet templates (like the ones we provide in this blog) that can be re-used with multiple accounts and web properties.

Extending the Value of Segmentation

As previously mentioned, segmentation can and should involve more than simply filtering counts. Nextanalytics provides a number of features that add tremendous value, including:

Fix the Data. Let’s you combine multiple items to create a new arbitrary item. Good for creating product or geographical segments that summarize detail items, or combining a series of values into a distribution group. When multiple items are given the same name, their values are automatically aggregated into your choice of sum or average or count. Can also be used to rename jargon such as a web page paths into something more readable.

Trend (pivot by date). Let’s you put dates onto a column axis and see the segment values change over time. Let’s you choose the date-precision to suit your needs – days, months, years, even day of the week. When this is combined with Fix functionality, your dates can be converted to fiscal or operational time periods.

Pivot. Can show one segmented dimension against another; Recency versus Depth of Visit for example. Mix and match combinations to see which dimensions are better indicators of desirable visitors and really zero in on the best.

Calculations. Things like net-change, growth, rolling 3 month or 7 day average, percent of total, percent of row or column average. Multiply two columns together. Cumulative total for Pareto curves.. Convert to rank or split into deciles. Top 10 and average of the rest. All done with simple functions and no formulas or intermediate worksheets to maintain.

Multiple Data Sources. Load data from your CRM, use it to segment your web traffic (customer domains). Compare to target. Compare multiple web sites in a single report. Compare segment trends to sales journal entries. The possibilities are endless, really.

End Note

Yes, we hype Nextanalytics. It’s hard not to. The product packs a lot of capability into a little package. It’s like a Swiss Army knife in your pocket; every day you’ll discover something new you can use it for. Segmentation is critical to improving your web site’s contribution, and Nextanalytics lets you explore faster and deeper, then automates your reporting. Don’t be satisfied with yesterday’s segments. Iterate.

Tags: calculations, Excel, fix, Google Analytics, pivot, segmentation, trend