Google Analytics: Don’t tell me who visits, tell me who came back!!!!

If you love your visitors, should you set them free?

In a word – yes. Because one of the most valuable pieces of information you can get is understanding what fraction of them return.

Most folks are familiar with the concept of bounce rate – what % of the visitors to a site bolt after looking at the first page. If they bolt in under 30 seconds, you failed to engage…

This other metric goes a level deeper. Think about the process of selling a house. A certain % of the visitors take a polite (hopefully) 5min walk through the building and leave. BOUNCE!

But just because they didn’t flee the facility on sight doesn’t mean you have a sale…..

To get a sale contract, you need them linger….and come back, probably several times.  Finally, after walking through the place with several friends and family members in tow, you will get an offer. So the real metric of pending success is – “who comes back”…

Content websites are similar. I’m glad you came to read my keyword article about dealing with migraine headaches. I wrote it just for you 🙂 But what I REALLY want to see you do is come back to my niche health topic site on a regular basis. So I can flip advertising at you and otherwise turn your visit into a profit generating relationship. Particularly if I’ve got to shell out some cash to get your click via PPC or advertising.

Running The Numbers

Thus… I need to understand the odds of a visitor returning and, if possible, the likely volume of additional visits I can expect from visitors I convert into relationships. The second metric is particularly critical for an ad-supported site – if you multiply this by the average # pageviews / visit, you can get the volume of ad impressions you can serve. Combine that with your average % CTR and $CPC, you can estimate the average lifetime value of an “inbound click” from different sources. This lets you focus on the most profitable keywords to spend your time pursuing.

You can calculate these metrics using three bits of data from Google Analytic’s custom reporting module; at present, Google Analytics doesn’t support calculated fields, so you will need some scratch paper to manually calculate a few extra items. The approach I’m recommending here works for new / growing sites and sites where the expected length of an “active relationship” isn’t more than a few weeks (eg. visitor is looking for help with a specific issue/purchase/event). If you are building a site with long term memberships, you will need to take additional steps to properly analyze attrition within your user base.

Going back to the math, the three required data elements are:

  • number of visits
  • number of unique visitors
  • segment this by visitor type (new/returning)

I’ve found this approach to work best if you’re looking at 30 day period, particularly if you can compare the results of the most recent 30 days with a period 60 – 90 days prior. This is sufficient time for the average visitor to return; having the ability to look back to a prior period can give you some additional help in triangulating on the right answer.

Now we calculate two statistics:

Repeat visit rate : # unique repeat visits / # new visits

Repeat Visits / Visitor: # repeat visitor visits / # Unique repeat visitors

Repeat visit rate tells you what fraction of your traffic returned for a second visit; Google Analytics can track visitors across multiple visits, so you can understand what clicks (from SEO or PPC) turned out to be “dates” vs. “relationships”.

Repeat Visits / Visitor gives you some insight into the intensity of the relationship. When you compare this between different subsets of your audience (source, keyword, etc.) or aspects of your site’s user experience (eg. landing page, device type), you start to get a feel for how well you are meeting their needs.

So what does this stuff mean?

Here are some examples of putting this concept into action. Most of the traffic for our word game solver website is originated via SEO, so we  frequently use this type of analysis to understand the fit between a Google keyword and specific page. The better the fit, the higher the revisit rate and visits / repeat visitor metrics will be.

A comparison of the two primary keyword sets for our word solver site was interesting. Product A “converted” clicks into repeat visitors at about 2x the rate of Product B. And the repeat visitors for Product A came back about 2x as often. It also showed us some of the new keywords we were pursuing had low conversion rates… indicating we had some deep thinking to do about the design and strategy behind our new word solvers.

Incidently, when doing this type of analysis, you should remember that you are looking at product-customer fit, not just product or customer performance. Be ready to spend some time digging into the details since the root cause of a performance issue might be:

  • market driven (eg. the traffic from a source / keyword stinks)
  • product driven (landing page / value prop doesn’t work)
  • results from the combination of the too (I’m ok, you’re ok, but we don’t click).

The next area of insight came from comparing the experience of our mobile vs. desktop users. When we drilled down to this by product space, we learned we were doing a pretty good job for our mobile users with Product A and a fairly lackluster one with Product B. As a general rule, remember that mobile traffic is more “intimate” (less distractions) than desktop although they can also be clumsy (accidental clicks). Once you win them over, however, they can be extremely loyal (higher repeat visit rate).

I’ve found these insights helpful in framing up product enhancements and SEO objectives, since it tells you where your strategy/offering is actually producing results.

Potential Refinement (Added 7/27/2012) – Filtering By Initial Bounce:

A potential improvement became apparent when I was going over some of our analytics this past week – you may want to look at new visitor conversion as a three step process:

  • Of my total search engine impressions (Google Analytics-Webmaster integration), what % of them clicked on my link. Differences / Changes in this metric can be indicative of “search relevance” and quality of your title & description tags…
  • Of the new arrivals, what % of them bounced immediately? Note that for an AJAX based website where a user doesn’t leave the landing page, you will want to use Google Analytic events to track activity (effectively “tagging” who didn’t bounce).
  • From that remainder  – # New Visitors x (1 – Bounce Rate) – calculate what % of them wind up visiting a second time.

This is significant because populations with a high initial bounce rate will appear to have low % revisit rates – when this may, in fact, be an issue with search engine misdirection. One challenge we have to manage in my SEO space is the fact that many word games are based on similar principles (scramble, jumble, etc.) and game manufacturers have adopted similar sounding brand names. So the searchers aren’t necessarily very clear in what they are asking for and Google has difficulty sorting out different sites.

Which is why – to my general amusement – I’m ranked behind a scrabble site on my best performing Boggle Search…. and getting an awesome CTR for a #2 position…

2 Trackbacks / Pingbacks

  1. Google Analytics – Don't tell me who visits, tell me who came back … |
  2. Google Analytics- Don't tell me who visited, tell me who came back! -

Leave a Reply

Your email address will not be published.