Conversion Rate Gremlins: Part 1

We work with a lot of large ecommerce clients, helping them with Google Analytics strategy, collecting marketing partner data for dashboard reporting, and deep analysis of their site (and business) performance. E-commerce retailers are in a position that comes with significant benefits because they’ve got a direct driver to sales. This just so happens to be their biggest risk as well, because if the website isn’t working, it’s bad news … really bad.

E-commerce conversion rate is a metric we spend a lot of time analyzing on behalf of our clients, and it’s not an easy number to figure out. A good way to think about it is as an efficiency metric. If the ratio of visits to transactions begins to fall, then the site is less efficient. It’s taking more visits to result in one transaction. The problem with conversion rate is that it will never tell you what the problem actually is, or if there even is a problem (not all reductions in conversion rate are bad).

It’s determining the cause of the change that is the challenging part, and there are two primary situations we see when evaluating fluctuations in conversion rate; (1) a sudden drop, and (2) a slow erosion over time. In both of these scenarios, Google Analytics makes it pretty simple to rule out a few major variables right off the bat; reducing your time investment and producing a more comprehensive and actionable result at the end. The big three I hit first are as follows:

1) Device Type

It’s really fast to determine if the decline is specific to desktop, mobile or tablet users with the Mobile Overview report in Google Analytics. Compare a time period of good conversion rate vs a time period of poor conversion rate and you’ll quickly see if one of these three line items is your high-level driver. If desktop conversion rate is down by a significant margin over the others, then you may want to build a custom segment to look at only desktop sessions and run the rest of your analysis on that. As written about in my last article many ecommerce retailers are seeing big time growth in mobile which generally runs at a much lower conversion rate, so hitting this report first can shave hours off your analysis.

2) Traffic Medium

Use the All Traffic report under the Acquisition menu item, and select Medium, to see a quick view of your main traffic acquisition buckets against conversion rate. Again, you’ll compare two time periods here and you’re looking for the medium(s) that may be driving the ecommerce conversion rate decline. Generally this look should be limited to your top 5-8 traffic sources since we’re just looking for the big impact items to see if there’s a single driver. If there is, then again a custom segment may be the ticket for deeper analysis, or simply drilling into that medium to see if there’s a specific source driving down conversion rate. If a medium/source is identified, then tracking down the cause just got a whole lot easier.

3) Landing Pages

This one can be a little more tricky if you’ve got a complex site, but it is a good one to consider up front especially if device and medium turned up dry. If there are specific pages or groups of pages that are not resulting in ecommerce conversions, and they’re a significant share of traffic, then focusing analysis on those pages will put you miles ahead of where you would have been otherwise. It’s possible that something is wrong with a page that visitors are landing on that could be pushing them away from the site, that’s what this report is great at identifying. If your site is more complex, use the filter box to search for big groups of pages (ex. content categories, product types, site sections, etc)…. Google Analytics will provide you a calculated ecommerce conversion rate for all pages that are returned from that search filter. This approach is especially good for identifying groups of content that aren’t working. You’ll find this report in the site content section under the behavior menu.

So, there you have it: Three quick-hit reports that can eliminate a number of variables for any conversion rate analysis. These are my three go-to reports when a question comes up around conversion rate, and it generally takes me less than 20 minutes to get through these. If results are down across each of these somewhat equally, then things just got a lot more involved and that’s when the fun begins. But we’ll save that for Part 2 of this series.

Leave a Reply