Are Google Analytics Motion Charts just Eye Candy?

It is certainly eye-candy, but it is in point of fact more than just that. I agree is it is not great science as such, but it is unquestionably a statistically very valuable visualization methodology that presents insights that I personally find very difficult to spot otherwise.

Example

We all agree that trending a metric over time provides good insight. So traditionally (incl. Yahoo! Web Analytics) one would e.g. visualize a visit to sale conversion rate in a 2 dimensional bar chart, with conversion rate on a Y axis and time on a X axis.

We also agree that looking at your data without segmenting it, is non-optimal, to the degree of useless. So for you to get true insight from the above example, I might choose to segment my visit to sale conversion rate by products, thus adding more values (we could call them separate segmented metrics). Let’s say we have a 1000 products.

How would we visualize this then?

We could choose the 50 best selling products for the period and perhaps do a trended line graph. This is messy, with far too many lines in one chart, but still doable. However as you probably noticed, we lost the trending on the product level, as we might have had a product spike within the period, but this is lost in the averaging of the 50 best selling products. Beyond this, imagine that we would apply revenue as a 3rd dimension. A third dimension could be applied as perhaps a color code to the bar chart or as the size of the circle in a bubble chart. But as you noticed again, this is a non-trendable visualization technique unless you add TIME (motion) as a dimension.

Conclusion

The fact is, that it becomes very difficult to trend many metrics (values derived from e.g. segmenting a metric) and or 3 or more dimensions without using techniques like Motion Charts. So I believe it is fair to say that motion charts are indeed very valuable and not just eye-candy – and they can potentially provide insights that one would not spot by shuffling through hundreds of e.g. bar charts in Yahoo Web Analytics or Omniture Site Catalyst for that matter.

And valuable even for the happy amateur analyst (like me) who can get answers to curious questions like: “how did thousands of organic search phrases trend over the last 6 months while having an eye for how many pages the visitors look at”.

This does however not imply that this is your average visualization methodology and the new communication standard and choice for your upcoming management meetings. Because it is not!

Cheers :-)
Dennis

Link

Efrontier has good commentary about this visualization method as well:
http://blog.efrontier.com/insights/2008/09/visualization-o.html