Upcoming LunaMetrics Seminars
Los Angeles - Anaheim, Sep 8-12 Washington DC, Sep 22-26 Boston, Oct 6-10 Chicago, Oct 20-24

Reading Reports in Google Analytics: Loyalty





I really, really had no idea how to read the GA Loyalty report.

So, once again, I did some testing (including just myself in a profile) and learned, this time, how Loyalty works. (Want to learn how to read the recency chart? I wrote about that one last week.) I learned a few important issues wrt Loyalty:

Like the Recency Chart, Loyalty is about visits, not unique visitors.

Grab your calculator again, and add up all the visits in the loyalty chart; you will see that they match the number of visits in Visitors > Visitor trending > Visits. For every visit in the period, there is a visit on the chart.

The Loyalty chart will measure the visitor’s visits as far back as there are cookies on the visitor’s browser.

It doesn’t matter that old visits are outside of the period. This one is hard to understand immediately, so I will show you a picture. Those of you who read my post on understanding the Recency chart know that I had 16 visits between November 13 and November 22, and I was the only visitor in the profile (and if you didn’t read the post, you can glance at the chart really quickly.) I didn’t clear my cookies before creating this new profile, and of course, I was the only person in the profile. Imagine my surprise when I looked at the loyalty for that period:

loyalty-google-analytics-report

In retrospect, this makes perfect sense. I did this analysis on the LunaMetrics site, and I visit it all the time, especially when I am blogging. So I had cookies on my computer that showed the hundreds of times I had been there, and GA put all of my visits into the top category, 201+.

GA does not aggregate an individual visitor’s visits.

This might seem counterintuitive. After all, we just looked at my sixteen visits getting aggregated into the 201+ category, right?

Well, that is an illusion. For each visit, GA computes how many times the visitor has already been to the site. I have been to the LunaMetrics site many hundreds of times, so these might be visits 3001,3002,3003… etc. But the top category is 201+, and that’s the only place GA can put them.

So let’s look at an experiment where I did clear my cookies first. Here, I had six visits between my start date, 11/18 and 12/4: Loyalty with cookies cleared

Eye opening, isn’t it? Notice that every time I visit, the Loyalty chart puts my visit into a new “bucket.” (There are two visits with a loyalty of 1 time, because GA thinks that I am two different visitors and doesn’t know how to show that I visited six times. The visitor dashboard even says: six visits, two unique visitors. Web analytics are not perfect. Sorry.)

The Loyalty Chart is quite useful.

Unlike Recency (which you can make a little useful, but is still a really, really hard chart to work with), Loyalty is really meaningful. If you are an e-commerce site with a ton of one-time visits (which is what I see a lot), then you need to look at your e-commerce menu and see if your “visits to purchase” number is very high. If it is, congratulations – maybe they come only once, but they buy. On the other hand, if your visits-to-purchase is low, and you have a *lot* of one-time visits, you need to start digging deeper.

And if you are a content site, like Facebook, oh wow! A high number of one-time visits really does show low engagement with the site (or maybe it’s about growth – you’ve got all sorts of new members, but you’ll know that from your back end.) Facebook and their sisters should be looking for a graph with lots of those 100-249 and 201+ visits.

Note: GA has updated their “about this report” descriptions (found in the lower left of every report). Go check that out, and congratulations, GA, on a great usability fix.

In closing, let me just get down on my knees and beg the Google Analytics Product Manager to rename this chart: Can’t we call it “Frequency”?

Robbin Steif

About Robbin Steif

Our owner and CEO, Robbin Steif, is an analyst herself. She is a graduate of Harvard College and the Harvard Business School, and has served on the Board of Directors for the Digital Analytics Association (formerly the Web Analytics Association.)

http://www.lunametrics.com/blog/2007/12/08/reading-reports-in-ga-loyalty/

12 Responses to “Reading Reports in Google Analytics: Loyalty”

That’s really interesting Robbin!

This series explains a lot about the essence of Google Analytics reports (what about a book called Google Analytics Demystified? Would it be considered copyright infringement ;-) If your parents read this they are probably saying to themselves: “We knew she should have become a professor!”

And if you decide to start a petition to Google for changing the report’s name you can count on me.

Thanks for sharing the experiment.

Robbin says:

Oh Daniel, I know that I can *always* count on you.

steve says:

@Daniel: Trademark Infringement surely? ;-)

@Robbin. To echo Daniel’s praise. Extremely useful!

If I can rephrase to confirm my understanding?
The graph shows the bucket that a given “person” falls into over the lifetime of the site (ignoring pedantry for the moment).
The number of visits, ie the X axis, vs Y; shows the current period.

Correct?

If so. Would not this graph be made *more* useful by being able to see the bucket deltas over time?
The idea being to easily see if one increases the # of 2 or more buckets vs single visit visitors. Just thinking of being able to track if one can increase or decrease by making changes – actionability. :-)

MS Excel calls that graph style “100% Stacked Area” – to help clarify my meaning.

FWIW, I personally use this style of graph a lot with various system monitoring details. CPU usage being the most obvious and easily understood example.

???

Cheers! & Thanks!
- Steve

Robbin says:

Correct on the x and y, and even though I can visualize the stacked graph (this is an old concept, MS doesn’t get credit for it), I can’t see how it shows you the delta. Am just braindead at this time of day, maybe. You, of course, just woke up…. and it is summer there, you lucky dog.

Steve says:

How did you find out I was a dog???? Who told!
Ah well. If the secret’s out… this is me at 6months old, ~ 7 years ago:
http://www.stedee.id.au/gallery/Doggies/ben-01.jpg.html

“Ben” is an alias for “Steve”. Obvious if you think about it. ;-)

re: Morning? You also forgot that I had the benefit of a large, mostly drunk, coffee beside me at the time.

The graph: By having a stacked graph – one band for each bucket, we can use the X-Axis to show time, with Y giving the bucket percentages. Apologies for obviousness – going for clarity.
Such that with changes in time, daily for example, we can see the changes or deltas in the various buckets as a percentage.
So suppose we do some effort to encourage folk to come back to our site. Would we not want to see how successful that was? Is the increase due to our efforts, or simply natural and expected growth. Or seasonal etc. These stacked graphs can make these subtleties, IMHO, very clear.
Hence why I, and others, use them for tracking CPU use over time. Is the server running out of capacity, if so when? type of thing.

Does that make sense?

As it currently stands it is useful, but I look at it and think “Cool! so how do I easily see if *I* have improved things, or not?”. Hence the suggestion.

As for summer. Well it was 19degC here today, so that *could* be defined as summer in some parts of the world. :-D
http://www.bom.gov.au/products/IDN60902.shtml

Still, give me 19 over 38 degC any day! At least you don’t have to worry about bushfires when it’s a struggle to get to 19….

Cheers!
- Steve

I looked at the following:

Unique Visitors (39811 for the last month on my traffic court site))
minus
# of 1 times visits (37037)

Wouldn’t this mean that about 2800 different people visited my traffic court site more than once?

But no …

For this year, I had 306,810 unique visitors and 305,025 # of 1 times visits. Following my above logic, only 1800 visited my site more than once for the year, but 2800 did so for the month.

Any explanations? Please e-mail me at wredlich – at – gmail.com if you can explain that one.

Warren

I have to echo all the praise on this one…spectacular post!

[...] información | Lunametrics (loyalty), Lunametrics (recency) y Avinash Kaushik (visitor [...]

Carl says:

Hi,

Great post and explanation. If you run a loyalty report with segmented data by traffic source, does it only take into account visits from that traffic source?

e.g. when you want to see the impact of a display advertising campaign on visitor loyalty, you would segment so only traffic from that medium is included. The problem is, people using a banner to initially find a site, probably wont use the banner again to get back to the site (they’d probably have bookmarked / remembered the URL).

Carl

Aviva B says:

Hi, Robbin!
Thanks for the post. Three years later, but the issue came up for me with a client now. I saw that you recommended this article about a year ago on the Google help forums, so I assume it was still accurate then, but have things changed in the interim? (as with the recency report – which I’m sorry you didn’t put more info in the comment about how it changed – but I guess that’s for another blog post…)
In any event, if you’re still answering questions over here, I wanted to check to make sure that I understood: since the total number of visits listed in the “visits that were the visitors nth visit” column does add up to the total number of visits for the time period (still does – just checked), does that mean that if a visitor visited 3 times in that time period, then their visit will be noted in three rows in that column (“1, 2, 3″ or “5, 6, 7″ or “8 and twice in the 9-14 row”)?

Marta says:

Hi Aviva,

Yes, your this is right. Depending on this vistors history he would be noted in three rows “in a row” as you mentioned. Let’s assume that until the period that you are interested some visited your website 4 times and 3 times in the given period. Them he will be counted under rows: 5,6,7 (each time he enters your website).

Hope this helps!
Marta

road says:

Today, while I was at work, my cousin stole my iPad and tested to see if it can survive a forty foot
drop, just so she can be a youtube sensation.
My iPad is now destroyed and she has 83 views. I know this is completely off topic but I had to share
it with someone!