If only working with our web designers were so easy.

it seems like I arrived late at the party this time… Anyway, I liked your post.

In the light of your writings I decided to publish one of eSnips’ graphs in my Web Analytics Graph collection. It shows how I compare six months of data from the “Length of Visit” feature of Google Analytics using Excel.

I hope you like it (remember to click on ‘Original Size’): http://tinyurl.com/ysewz9

1) New Visitor Percentage: Ratio of new visitors vs. returning visitors.

Since you always want to be building your audience base, you’ll want to track what percentage of your visitors are new. As you increase the level of new visitors you’ll also want to know if these new visitors are part of your target audience and interested in your sites content. That leads to the second piece…

2) Returning Visitors: Ratio of visits to unique visitors.

3) Page/Content Depth: Ratio of pageviews to visitors at the page or content level.

This will allow you to analyze not only which page or content areas are your most popular, but also see them in comparison to the overall pageviews of your site. Trending this overtime will show you what content is building momentum and what content is getting stale.

4 metrics for websites that deliver content not commerce: 1) loyalty, 2) recency, 3) length of visit, and 4) depth of visit….

As you mentioned displaying ads/charging more for your ads as visitors tend to stay longer, etc. maybe it would be possible to only display ads to people who have already become loyal repeat visitors (and are thus unlikely to get put off by a couple of ads)?

Can the length of visit metrics be further segmented into “length of visit of first-time visitors” (with some tool)?

I think that would be very telling, because a lot of people (like myself) probably are repeat visitors who type in your domain name and then disappear quickly b/c they just wanted to check whether you had made a new post (and probably do this multiple times/new blog post).

That would probably account for most of the 0-11 seconds visits..and many of the 11-30 and 31-60 might be such people, too who decide to take a look at the comments section or read your old/last blog post again for a couple of seconds.

I’m wondering if the assumption that if you draw them in for 60seconds that they tend to stay a lot longer..and what you could do to make them stay longer is all that helpful, though? I might be wrong, but I would guess that 0-10, 11-30 and 31-60 seconds stays are mostly repeat visitors (any way to segment that??:-)) or visitors who are new and don’t really like the blog (b/c they had expected something else?)..

and that those who stay longer would be mostly visitors that are reading a new blog post…the decision you could make based on that being that you should write new blog posts :-). j/k

Does the length of stay peak around times when you make a new blog post?

I really think it would be so interesting to segment this data for new visitors. I think that would really help the webmaster take action and realize if he could do something with the site that would help make first time visitors stay longer than 60 seconds - and thus get them to continue reading for quite some time and eventually come back and turn into repeat visitors.

I think if I come to a site that Ive visited before (and visit it somewhat regularly) the length of stay-metric on the website isn’t all that revealing, but I know that a lot of times I’ll go to a new website and THEN either I like the content on it and I’ll continue to read..and start reading other contents on the website..or I dont like it and I’m often gone forever.

Actually that’s the reason why I thought the top-blog-posts-idea can help you draw in visitors for a long time. Often it’s not just about the value of the content, but also about the packaging of the content: Write about a topic and put the content on a website and people will think of it as another website, but package it as an e-book or a “real” book and people will associate more value with it despite the content being the same. “Repackaging” top blog posts and making them look appealing to first time visitors might have a similar effect as packaging of a book..and make (hopefully many of) them stay for longer than 60 seconds.

But again, I would find it so interesting to see a segmentation for time on site vs. time on site for first-time-visitors ;-)

P.S.: I hope my assumption that these are the stats for your blog is right..?

I will add that what I caution against is making a judgment call that simply becuase then people saw the page they were looking for means that they found it to be helpful. This is very often what happens, “we made a thousand people happy becuase they saw this valuable piece of content on our site”. It could be badly written, maybe it has a high bounce rate etc.

For that reason I recommend that people use surveys etc to get a feel for how the customers feel about the content itself (”Did you find this page easy to read.” “Were you able to find the answer you were looking for.” “Did you complete your task on our website today.”). You get the idea. :)

Nick : Completely agree with you on segmentation by keywords, or anything else for that matter. The insights will come tumbling down.

I am not sure that I would say that the number is “inflated” by the bounces. People come, see just one page and then they leave. I guess it would be better to table them as NA because for those sessions the number is unknown. I suppose the zero suffices. :)

JS : Here is an attempt at a cogent answer to your wonderful question, directly from a very important horse’s mouth!

If I had 1 visitor who on 1/3/2007 visited my site, then came back on 1/6/2007 (e.g., same visitor), and I was looking at GA with my date range set to 1/5-1/6…

The bar chart would show:

0 day ago: 1 visit
3 day ago: 1 visit

Note that this is EVEN WHEN the date range does not include 1/3/2007.

In summary, we do the above calculation for each visit. For every visit we determine when was the previous time we saw a visit from the same visitor so the count would add up to visits. The length calculation is based on the visitor cookie.

The formula for calculating # days ago is :

(previous visit time in seconds - current visit time in seconds)/ 86400

where 86400 = number of seconds in a day.

Does this help explain what this is?

Chuck : See Justin’s post on how to go about doing segmentation with GA (for other tools you can check their user manuals - do we still have user manuals for these things! :)). Link:

http://www.epikone.com/blog/2007/07/17/segmenting-visitor-loyalty-reports-in-ga/

Brian : Yes and no. Let me try to explain.

For blogs specifically I recommend not to get obsessed with bounce rate becuase blogs are essentially one page website. Most people will read the home page (where most of the fresh content will be) or they will look at the latest post page and leave. So even if they spend thirty minutes reading it in either definition (single page or number of seconds) they qualify as “bounced”.

For other normal websites that is not the case. Then no matter how you measure bounce rate (single page view or number of seconds) that is a great metric to measure quality of traffic that you get and your landing pages.

Does this make sense?

Andrew : Thanks for sharing your story, you are my new hero!!!

Stuart : It is not easy as I wish it were but using standard GA mechanisms you can. See the link about to Justin’s post for the how.

Everyone : My apologies for the late replies to the comments. Between the seminars and my other personal speaking and consulting engagements life has been a blur recently. I do get to meet a lot more of you, but I get to sleep a lot less! :)

Thanks so much for all the delightful comments.

-Avinash.

Thanks,
Anirudh

A real understanding is needed of how to provoke “deep” visits: retaining visitors beyond landing pages and moving the content consumption needle to 4, 5, and 6 pages.

Segmentation is crucial to really getting mileage out of this data, and I’m glad you hammered that point home as one of your recommendations!

Someone at Sports Illustrated is clearly paying attention to the latest trends in online surveying. They’ve come up with a survey front-end to tabulate the first ever (to my knowledge) Fan Value Index. They’re asking fans to rate elements of……

It’s about time I visited your blog often, if possible, whenever you post. I don’t know why web analytics so far have scared me. May be the reason is I have had no data of my own to analyze :-). I scarcely improve my own websites and whenever I do, it’s all about adding new content; very little promotion, and near-to-nil traffic analysis. I’m gradually going to change that, and this post has helped me a lot. Thank you.