In part 1 of this article, I introduced a new report that shows some of the powerful analytics capabilities of Nextanalytics for Excel – in this case, a full series of tables exploring the relationships between the Recency, Loyalty, Duration and Depth segmentation. This Excel-based dashboard also provides the ability to filter by any of Google’s default segments or your advanced segments. The combination provides an unprecedented visibility into visitor behaviour, all in one screen.

Most Loyalty or Recency reports are restricted to a single dimension. Individual numbers are grouped into buckets to form a distribution chart where each bucket represents a segment with a range of values, with the number of visits aggregated for each segment. These reports let you answer a number of really good questions (e.g. How do I measure success), but the golden nugget in segmentation appears when you explore how one segmentation set interacts with another.

For example, in this table, we see the intersections of Duration of Visit segments and Depth of Visit segments (filtered for non-bounce visits only). It shows a nice spread of the data, and clearly indicates that, as the number of pages viewed goes up, so does the length of time spent. Now you might be thinking ‘duh…of course’, but if your site had a lot of navigation pages, it would be more likely that you’d be seeing a lot of page views in a short period of time. In this example, it looks like most visits top out around 3-5 pages, but they are hanging around for a minute or more; they are obviously reading something. Maybe that’s good, maybe it’s bad – it depends on the purpose of your site.

As another example, if we filter using the Returning Visitors segment, and look at when they were last on the site (Recency) versus how many pages they viewed (Depth), we see that a lot of people (29.4 + 7.7 + 8.3 = 45.3%) came back the same day for a handful of pages, but that a significant number (over 14%) have come back within the week, and some even a month later (10%).

The real power of the report, though, is when you combine observations from all the charts. For that Returning Visitors segment, we can look at the broader picture (see image). To read a page of segmentation data like this, you should start by looking at the biggest number in each table. We’re particularly interested in the tables with the largest and the smallest ‘big’ numbers because they will tell us the most.

The largest number on the page (lower-left table, Duration x Depth) tells us that over half of all return visits (54.9%) were single page views and under 10 seconds, so we know that bounces predominate the results. Because of that big rock in the pile, the default formatting doesn’t highlight the variation in the remaining cells, but hey! we’re looking at numbers in an Excel spreadsheet – just change the formatting rule for that table! [try that with any other reporting tool]

Now we can clearly see the same type of Duration-Depth relationship discussed earlier, dropping off after 5 pages and 600 seconds. Good or bad, we now have a measurement that we can track from week to week, month to month, to evaluate whether actions we take have a desirable effect.

The other table of interest contains the smallest ‘big’ number on the page (upper-right table, Recency x Loyalty). It reveals that less than 27% of the visits were from people new to the site (2-5 times) that came back the same day. Because this number is smaller, it means other numbers in the table are larger, and we are interested in which segment combinations are ‘different’ and form a pattern. Larger numbers mean bigger segments.

What we see here is that both the first column and the top row are higher than the rest and do not drop to zero, in fact they appear to peak at the end. [Note: exercise caution in comparing numbers from the various segments since each segment was defined by arbitrary limits and may be a different size.] We can also see that the values drop off quickly as we move away from that first row or column. So it appears that visitors with more than 5 visits tend to come back the same day, and that people that don’t come back the same day tend to have less than 5 previous visits.

That’s an interesting pattern, but it’s up to you to decide if it relates to your web site goals and what you want to do about it. Remember, analysis without consequences is a waste of time. Maybe you should use all that white space in the lower right corner of the worksheet to list your observations, gauge the business impact, and propose an action plan (gee, where did that idea come from?).

Extending the Concept

For eCommerce web sites, you could add a dimension based on the revenue per visit and produce what marketers have long called the RFM or Recency-Frequency-Monetary analysis.

Rather than tables of numbers, you could create a series of sparkline or micro charts to show the trends for each segment. That’s easy to do, and I’ve used that technique in many of the other reports on this site.

The tables could be extended to include a total row and column. Again, a simple task that has been demonstrated in many of the other reports.

The date could be included, and each dimension or a combination of segments of interest could be trended over time. This is a trade off decision — by trending by date, you lose the second dimension. Maybe you choose to take a slice of two or even three other dimensions, and trend that — e.g. where Loyalty = ‘2-5 times’, show how each of the Recency segments change over time.

Analytics is in our Name

With Nextanalytics, all of these ideas are easily within reach. They can be automated and the results delivered directly in Excel workbooks where you are free to format or further analyze the information to suit your needs. No other tool provides this kind of power, flexibility and integration.