I was merrily using Time on Page and Time on Site metrics for quite some time before I actually realized how they were being measured.
It was a real Doh (!) moment.
Turns out we have not rfid'ed every visitor and they don't rub their head against their monitor before starting their session on my website (and of course another head rub when they decide they have had enough and exit). This would allow us to capture the time stamps accurately and have exact measures.
What a disappointment! Joking, just joking!! :)
I find that few people understand correctly how the Average Time on Site calculation is made.
That's regardless of source: weather they use the religious truth from a Competitive intelligence tool or from their website web analytics solution. For the latter it does not matter what data capture methodology you use on your site, WebLogs or JavaScript Tags.
This post is my humble attempt at explaining how Time on Page and Time on Site are computed.
In order to make life easy I am going to assume the following session happening on a website:
Someone requests your home page, your web analytics tool starts a session for the visitor, two more pages are requested before the visitor decides to leave your website (close browser, type in a url of a different site, click on a link on your site to go to different site….).
What we want to compute is…..
Tp = Time spent on a page.
Ts = Time spend on the website.
Someone visits your website at 10:00……
There is a entry in your log file (weblog or javascript tag, does not matter) that says, in English, "someone has requested the website homepage file at 10:00".
[
It actually looks more like this….
111.111.111.111 – – [08/Oct/2007:11:17:55 -0400] "GET / index.html HTTP/1.1" 200 10801 "http://www.google.com/search?q=avinash+kaushik&ie=utf-8&oe=utf-8 &aq=t&rls=org.mozilla:en-US:official&client=firefox-a" "Mozilla/5.0 (Windows; U; Windows NT 5.2; en-US; rv:1.8.1.7) Gecko/20070914 Firefox/2.0.0.7"
Notice the time stamp there?
Why spoil the fun with technical stuff! But if you want more the tech stuff is explained very nicely here: Log file sample explained.
]
So far all your analytics program knows is when a page was requested, hence:
Tp = N/A [Not Available]
Ts = N/A
Next more fun happens on your site: someone clicks on a link to Page 2 from your home page. Hurray no bounce! :)
Now there is a new entry in your log file that says: "The same Visitor requested page two at 10:01"
Finally your web analytics program can compute some time metrics!
It knows how long the Visitor spent on the home page. It subtracts 10:01 from 10:00 and gets one minute. Hence:
Tp (Home Page) = 1 minute.
Notice something important: the only way your program knows how long someone spent on one page is by looking at the two time stamps. One from the request for the first page and one from request of second page.
Next the blinking "get $200 rebate on a $210 product" link on Page 2 entices the person to click on to Page 3 to buy the product. More sweet success! "Engagement!!"
The magical math outlined above happens (10:05 minus 10:01) and for Page 2:
Tp (Page 2) = 4 minutes.
The Visitor reaches Page 3 and notices that the rebate offer only applies to people who live in Antarctica who can show they currently own refrigerators (!!). As you can reasonably imagine this happens on the site on Page 3…..
Exit!
How long did it take to find and read the rebate fine print? You could reasonably guess if you knew how long the Visitor spent on Page 3.
The problem is that your logfile is missing one time stamp to do the magic math.
Tp (Page 3) = Time of page request (10:05) minus Time of next page request (N/A).
Hence:
Tp (Page 3) = 0 minutes (Since N/A!)
The program has no idea how long the Visitor spent on the last page on your site.
This is true for pretty much all web analytics programs in terms of default behavior.
Let's wrap this puppy up….
Tp (Home Page) = 1 minute.
Tp (Page 2) = 4 minutes.
Tp (Page 3) = 0 minutes.
Ts = 5 minutes. (Time on Site, also known as Session Length)
Makes sense?
I mean it probably does not make sense that you don't know how long Visitors spend on the last page in the session, but the explanation of how it works makes sense?
The Case of Bounce / Single Page Sessions:
Now I am sure your are asking yourself: "I wonder what happens in cases of sessions where there is only one page?"
Good question. This is what you are wondering about….
Bounce!
This is what is computed….
Tp = 0 minutes.
Ts = 0 minutes.
For bounced sessions (single page view sessions) your web metrics program can't compute how long people have spent on the page or on your website. It certainly records the page request (10:00) and it records start of the session but it does not know how long someone was there.
In case the Visitor came at 10:00 and leaves the browser open and rushes as the Visitors Wife / Husband / Boss yells at them to do the dishes and cleaning the dishes takes and hour…. the session is terminated at the end of 29 mins (default setting in most session based web analytics tools). It is important to note that the web metrics program will still say:
Tp = 0 minutes.
Ts = 0 minutes.
The Case of Tabbed Browsing:
Firefox gets credit for popularizing tabbed browsing – I love it and I have no idea how we survived for years without it! The latest versions of IE also support tab browsing, so the masses are also now using this delightful feature.
But what happens to the Time on Page and Time on Site computations when people open a link on a site in another tab and browse the site via two tabs at the same time? I do this all the time! :)
It messes up the time computations.
Here is a common scenario that we'll use to understand the impact…..
A Visitor comes to your home page. From there opens the first link in a new tab, but continues to scan the home page. The clicks onto a link to Page 2 from the home page, then onto Page 3 and then closes the tab (or moves away and forgets about it).
The Visitor goes to the tab opened from the home page to Page 4 of your site, spends time there, goes on to Page 5 in that tab. Then exits.
How is time on site computed? There are two ways in which I have seen web analytics tools report on this customer behavior.
Dealing with tab browsing # 1:
The web analytics tool takes the above picture literally and this happens…..
Outcomes: Two Sessions, one for each tab in the browser.
Session One (top): [referrer -> Google]
Tp (Home Page) = 2 minutes
Tp (Page 2) = 3 minutes
Tp (Page 3) = 0 minutes
Ts (session duration) = 5 minutes
Session Two (bottom): [referrer -> your site/homepage]
Tp (Page 4) = 6 minutes
Tp (Page 5) = 0 minutes
Ts (session duration) = 6 minutes
Net net: 2 Visits. 1 Unique Visitor. Also notice the impact on referrers (for the second one you'll see your site referring to itself).
Really interesting outcome!
Dealing with tab browsing # 2:
Some web analytics tools "collect" all the "hits" (entries in the log files) and they will "linearize" the hits and construct one session from all the tabbed browsing Visitor behavior.
So keeping our use case exactly as above, visually this is what happens when data is processed……
[Pretty high resolution image: Click: Time on site impact tabbed browsing -linearized]
Outcomes: One Sessions, visit “reorganized” by time stamps.
Session One: [referrer -> Google]
Tp (Home Page) = 1 minute
Tp (Page 4) = 1 minute
Tp (Page 2) = 3 minutes
Tp (Page 3) = 2 minutes
Tp (Page 5) = 0 minutes
Ts (session duration) = 7 minutes
Net net: 1 Visit. 1 Unique Visitor.
Neither one perfectly captures exactly what the Visitor is actually doing on your site.
Which one do you prefer?
Be sure to ask your web analytics vendor which of the above two (or a different method) do they use to deal with tabbed browsing when it comes to computing time one site.
Given the increasing popularity of tabbed browsing the impact on your numbers could be big.
Google Analytics uses the second method ("linearization").
Take a deep breath now!
Extra Credit Section:
As always there are hacks that might allow you to capture the time spent on the last page (or even on the last event if you are using event logging).
One of the most common (and when I say most common I mean used by the 0.001%) is to add extra script / code that would capture the fact that the page was "unloaded" in the browser. Technically it is often called "onbeforeunload event".
In this case for each page you have a page request time stamp as well as a page unload time stamp. Hence you can do clean magic math required.
You will have to script this yourself or ask your vendor to help you create such code. You would then also have to ask your Vendor to modify how time on page (and time on site) are computed in the web analytics tool to take advantage of the additional time stamp.
If you are doing your own logfile parsing then you can more easily modify the log file to parse and data and then compute the metric for you.
There are other hacks as well.
Some people have also ventured that if your analytics tool has outbound link tracking (also often called exit tracking) then the web analytics tool has the time stamp of that click and can use it to compute time spent on the last page.
I am not a big fan of this because most people on your site will not exit the site by clicking on a link on your site (because most links on your site don't link to other sites, how inconvenient!). So if you use the outbound/exit link time stamps you'll end up computing Time on Site / Page for some Visitors to your site one way and everyone else another way.
It muddies waters, it becomes mixing apples with watermelons.
I am a fan of consistency, even if you consistently measure something imprecisely.
I hope that this post has helped you understand how time on website and time on a page is computed. Next time you see Average Time on Site or Average Time on Page you'll know what data is being included and what is not.
Also next time you compare numbers between a external tool, one of the competitive intelligence tools, and those reported by your internal web analytics tool, you'll know the questions to ask before you jump to conclusions!
Closing Thought : The purpose of this post is not to imply, overtly or covertly, that Time on Site is not a good metric. Far from it. For many types of businesses it can be a critical metric. My hope here is to educate you about how it is computed so that you can make more informed decisions. Computation of no web analytics metric is without flaws (look no further than Unique Visitors!!), time on site is perhaps one of the less flawed ones. :)
Ok now its your turn.
Please share your perspectives, critique, bouquets and brickbats via comments. For the Analytics Gods out there, did I get it right? Please comment.
[Like this post? For more posts like this please click here, if it might be of interest please check out my book: Web Analytics: An Hour A Day.]
PS:
Igor was upset that I was not doing a good job of cross posting. Since customer feedback is important… here are earlier posts in this series:
- Standard Metrics Revisited: #1: Visitors
- Standard Metrics Revisited: #2: Top Exit Pages
- Standard Metrics Revisited: #3: Bounce Rate
And here is a related post that outlines four attributes each great metric should possess:
Via 
Interesting read Avinash, thanks – as Google Analytics uses Javascript tags to measure the data, is it one of the tools which uses a script to measure time spent on the site?
Really?! I didn't know that. I assumed (oh dear, slapped wrists) that it 'knew' when the browser had moved on. Now I think about it, how could it? 'Doh' to me too then.
Avinash,
Thanks for this post. It's really good to see this confusing concept explained simply. However, I do have one quibble! I think in your second example, most programs will actually report Ts = zero, not Ts = N/A. Google Analytics started using N/A a few months ago, but quickly changed it back to zero.
Zero makes more sense. In your first example, Ts is really "5 minutes + N/A but we'll pretend it's 5 minutes". In the second example, it's "0 minutes + N/A but we'll pretend it's 0 minutes". If you call it N/A instead, then the average time on site will take the average of everyone who saw more than one page. That might be an interesting KPI for some people, but it's more complicated to explain and will significantly inflate the average compared with standard average time on site.
I'm no Analytics God but afaik you got it right. :)
However, I think that "Time on Site" is really pretty useless metric because of the limitations outlined by you.
And there's more which, in my opinion, renders this metric very unreliable:
Tabbed browsing
I use to browse a lot of web sites by not opening links in the same window but by opening them in a tab, so I can read the content of the current page in the first tab, maybe follow another link in this same tab, while the other pages opened in tabs "sit" there waiting to be read.
This does not only mean that the time spent on that first tab is reported as being much shorter than it actually was but also that the time on page for the pages in the background tabs will appear to be much higher than it actually is.
I haven't actually been on one of those background, after all. They were just waiting there to be read by me.
Well, but maybe it's just me browsing a site this way… :P
Great post, Avinash!
@Rob Lewis — page tagging or log files…are both log file based (I recently wrote a post titled "Web Data Capture Methods — Two Methods that Suck" that explains how both work — the "Suck" part was an homage to a post from Avinash back in 2006): http://tinyurl.com/2yo4o6
Using Javascript to capture the onunload would work with either page tagging or log files…but it does then require some processing modifications to interpret whatever "request" that script fires.
One of the "other hacks" is to use scripting to fire off a request at a regular interval. I was playing around with Fiddler a while back and could swear I found a couple of sites that were doing this. This, though, seems like a really good way to tee off the user!
I hope you plan on continuing this standard metrics series for all 28 WAA standard metrics. It's a lot of fun to follow along (for us metrics nerds)!
– CJ :)
If a website is designed entirely in Flash, how does the analytics software capture time stamps if there is only "one" page to the entire site? Just curious…
Avinash-
I always enjoy the way you break down concepts. Somewhere deep within you there lies an academic:-). This was an enlightening read as I did not think about "The Case of the Missing Timestamp".
One quick question though. If page 1, 2, and 3 above were to be part of a funnel (say a 3 level) would the total time for the linear path still be Tp(page1)+Tp(page2)?
On the time spent on Page 3, I was thinking that the calculation should be the other way around
(Time Next Page – Time of request for page 3). Did I get this wrong?
-Ned
Great article as always – i too was wondering about the whole tabbed browsing issue?
I think a large percentage of (primarily) FF users split the site onto different tabs – especially if it's a site with a main page or index, then flip between the different tabs. Does analytics pick this up?
I'm not thinking this will be a big problem with the new website i'm working on – but it's an interesting question for sure.
Also – i don't want to make the site stats of regular sites i visit seem worse than they should be if analytics doesn't like tabbed browsing :)
Interesting post Avinash. I always figured the 'time on site' metric was a tough one to understand, just considering my own browsing habits of having multiple browsers and tabs open.
It's also one of these metrics that hard to interpret when you do have it. IMO most pages have an 'optimal' time based on the goals of the content. Index type pages want you to find what you are looking for fast and move on, heavy content pages want you to stop and read. In isolation of the goal of the page, the metric is rather useless – it seems.
The best use is probably as a tracking unit – time-on-page moving up or down compared to its historical average needs looking into.
I'd be interested to hear from people who actually use this metric for a KPI.
I remember reading up on this in your book, b/c somebody asked about it on an SEO forum. Great to have another post about it. Especially one that's so visual :-).
Everytime I hear about this what comes to my mind is the fact that click density reports/heat maps (not sure which is the correct WA expression) exist.
If you know where users click isn't it possible to find out also when those clicks occurred? I often spend tons of time just reading the same page, however I usually have to scroll, etc.. I assume for some reason it's not possible/too complicated to pull this off?
This is a fantastic post series! These posts make very complex ideas so simple that even internet novices can understand.
Please keep this series going ;)
AH
Great post and I learned a lot. I'll have to go back and dig through your previous articles.
One question – you mention "the session is terminated at the end of 29 mins". Is that the case for Google Analytics? I understand that analytics packages take different approaches but I've been curious about some of the disparities I've seen as I've compared Google Analytics to PMetrics, Statcounter and several local analytics apps. In many cases, I've found Google Analytics reporting significantly fewer visitors and page views as compared to other packages. Thoughts?
Great article! I have started to get involved with web analytics. I am in the process of reading the book which my boss sent me for Christmas :). I am sure I'll be referring back to it too. Love the site.
I thought that Google Analytics injects some JavaScript that catches "onbeforeunload event" and reports back to Google Analytics.
Why standard urchin.js script doesn't do that?
Can you explain a bit how Google Analytics handles this using their "old" method (urchin.js) and how they handle it with their new event tracking (and ga.js)?
The reason I ask is because most of the sites I work with are in Flash and with urchin.js I've been creating fake pageviews to track "events" (which may have had a side benefit of creating lots of little time stamps – good for time on site, bad for time on "page").
However, with the transfer to ga.js and event tracking, I'll have less pageviews and more events tracked in the real sense. And since I compare performance across sites (using time on site as a performance metric – for better or worse), I'm afraid this might be unfair to the newer sites using ga.js.
Thanks for the post Avinash – makes me realize how much I don't know.
Rob : You asked since tag based solutions were already using "script / code" to collect data if they were currently automatically using the "unload" code in measuring Time.
The answer is no for pretty much all analytics tool in the market, including Google Analytics and others.
My hypothesis has always been that it is probably because by doing that the vendors will instantly get twice the amount of data (rather than one "hit" per page it will now be two). Impacting storage, compute power, query times etc etc (especially in a ASP implementation).
I want to stress that this is my personal hypothesis. For the actual best answer please ask your vendor.
Dr. Turner : You are right about 0 instead of N/A in the case of bounce / single page view session. I have corrected the post above. Thank you so much!
In terms of what GA did: It was not so much that they started storing N/A where they used to store Zero.
It was more the case that for that short duration the query that computed time on site would simply ignore single page view sessions since in reality there was no actual data for how long the Visitor was on the site. So technically improving the Average Time on Site. As you note this was reversed quickly (though it would make a very interesting computation of how long the actual Average Time on Site is for sessions were data is available).
Lars : I can't believe I forgot tabbed browsing!
For you and others who asked about it I have added a whole new section in the post, please scroll up and read the "Case of Tabbed Browsing" section above.
Kristine : What I described in the second scenario happens. It appears as a single page session.
So zero time on site and zero time on page.
But if you use features like Event Logging from Google Analytics (and couple other web analytics tools that also do true event logging) then you can capture how long they spend interacting with your cool flash website. What buttons they pressed, how much they forwarded or rewinded etc.
Everything except from the last click to exit from the site. There you are still stumped and for that last click you don't know how long (unless you do a hack).
[
Bonus Info: These two articles from Justin are a bit technical but a excellent read for understanding exactly how event logging works and how you can implement it:
http://www.epikone.com/blog/2007/10/16/event-tracking-pt-1-overview-data-model/
http://www.epikone.com/blog/2007/10/16/event-tracking-pt-2-implementations/
I also highly recommend Justin's excellent book: Google Analytics Short Cuts: http://www.oreilly.com/catalog/9780596514969/
]
Ned : You are right, it would be Tp(page1)+Tp(page2).
For the latter question you had: bigger number minus the smaller number (if the bigger number is present!).
Alex -S : Please see the new section in the post above on tabbed browsing! :)
Paul : I don't know if I'll agree that it is not a useful metric (think of content sites, think of 98% of the visitors on ecommerce sites that will never convert, think of non-profit or non-ecommerce sites).
The number is important but: Few people understand how it is computed (then they average it!) and fewer still then can make decent decisions off it.
Patrick : If there is a click (what you are seeing in click density / heat maps) then there is a time stamp.
Remember the article mostly covers the problem of what happens on the last page, in that case there is no click. The Visitor simply leaves. This means you have no idea how long they spent on that page.
Robert Irizarry : The standard for pretty much the same for all web analytics tools now in terms of how the sessions are terminated when there is inactivity.
Many tools will allow you to change the value from 29 mins of inactivity. So net net you should check your tool.
Also just to quickly note that this will impact the Visits number, not Unique Visitors.
Dennis : Please see the Tabbed Browsing computation section in the post above. I think it answers your question.
Latham : You are mixing a bunch of things together. Let me try to see if I understand your question and try to explain….
Time computations between urchin.js and ga.js have not changed one bit. So absolutely no impact on stuff you read in this post. It is how it used to be.
You were creating "fake page views" to measure rich media in the past. This had a detrimental impact on your page counts, as well as time spent. You were just mixing real page events with fake page events. (Don't feel sad, this is still how most, but not all, web analytics tools still measure rich media.)
If you use Event Logging which is available via ga.js you will capture true events, this data is not fake page views, it is stored separately in a clean way, and hence you can get real time on page and time on site and real time spent interacting with your rich media.
Hope this helps.
Everyone : This post was a lot of work and then it was several hours of more work to update it for tabbed browsing and cleaning things up.
But your comments and fantastic questions make it all worth it. Thank you and please keep your comments / feedback coming!!
-Avinash.
Now that it is a few years after you wrote this, I am curious to know how Google Analytics currently measures Time on Page/Site with Bounces.
Are Bounces still seen as 0:00 and worked into the Avg?
Or are they again N/A and not included in the Avg?
I have checked the Google Analytics Help section and they state that they do not include the exit page in the average duration, but they make no reference to bounce rate.
Any help would be appreciated!
Kirk
Kirk: All web analytics work the same way when it comes to time and bounces.
Bounces are sessions with just one page view (technically a "hit") during their visit. Bounce rates have no connection to time.
Time in all sessions, including where there is only one page view and hence no estimation of time on page, is included in the time on site calculations. Nothing is excluded.
I hope this helps.
Avinash.
"For Average Time on Page, bounces are excluded from the calculation. In other words, any Time on Page of 0 is excluded from the calculation.
For Average Time on Site, bounces remain a part of the calculation.
To calculate Average Time on Site, Google Analytics divides the total time for all visits by the number of visits." – From http://www.google.com/analytics/iq.html
Thank you for writing this! Now if we can only get it on the front page of the WSJ so the rest of a business will understand it!
Hi Avinash – I think you touched on this at Blog World and after reading this post I really understand how the time is computed.
My question are –
Even though the data isn't completely correct, it's consistent, so as a blogger, should I be paying attention to this metric?
Should I be watching for trends?
Thanks!
Heh. Tabbed Browsing. The amusing side of that for evil-me, is the mindset(?) that no-one ever had multiple browsers open to the same site before tabbed browsing came along.
The entire web is so unstructured, so non-linear, yet the common viewpoint from the analysis side only wants to see the linear view.
Recognising that no-one else is me. :-) Buying Stuff sites are the ones where I have always had multiple windows or tabs open. It's just easier to bounce around and compare "stuff" that way vs. back and forth.
I'd disagree with your phrasing :-) of "linearise" for the 2nd tabbed option. As the first option of splitting the two streams is somehow according that single person a split personality – that the two streams are somehow divisible. They are not.
So it's not "linearising", it's showing exactly what happened. *Hopefully* the referral side will allow for understanding and actions to take to fix the possible problem of … hmmmm…. "forcing" end users to split ones site up across multiple tabs.
The assumption I feel you've made here Avinash, for eg, Page 5 is *only* accessible via Page 4; is the key issue. The user may be using the 2nd tab as a scratch pad. Lookup style of Thing. Page 5 was simply more convenient for them off the 2nd tab than the first.
I guess what I'm saying (eventually :-) ) is that we are trying to complicate and understand something that is all but impossible to understand. We have no mind-reader tags yet. ;-)
Cheers!
– Steve
I'm probably guilty of promoting time-on-site reporting as much as Avinash. I'm puzzled to read the criticism of it and all the inaccuracies, when ROI / ROAS / Revenue reporting is even more error prone.
People feel confident about ROI reporting because it seems to solid, and time on site seems vague.
I assure you, if you knew the process by which ROI gets assigned, you'd run into the safe arms of time-on-site.
Maybe we should dissect ROI calculation?
On some sites i have a bounce rate of 40% for returning visitors. So tab browsing could be a part of the explanation?
For average time on site GA takes the average of all times including the 00'00 of the bouncers =A.
Then they take the average of the time on site without the bouncers =B.
And then they take the average of A and B.
That result is your "average time on site".
Thanks for this summary.
I was just looking into google analytics with this time on site metric in mind and wondered how they calculated avg time on page.
Do they still average A and B? A couple of years have passed since this comment, which is why I am asking…
Justin: The core of how web analytics tools, pretty much all of them, compute time on page and time on site remains the same. Perhaps primarily because there aren't too many pretty options for alternative paths.
It is possible to try and get a more "accurate" time recorded for the page and the user session. For example, I'd shared this example for more accurate bounce rates for Google Analytics: http://goo.gl/x2HpNW There are other options, and they all mean you will customize the code (which is not hard, as you'll see I even share the code there). Reach out to a GACP if you need help (www.bit.ly/gaac).
Do remember that if you do what is mentioned above, your calculations will be different and if you look at benchmarking reports in GA then you would not really be able to compare your data to that of others. This is not a terrible thing, but something to be aware of.
Avinash.
Once again, great post.
Are you sure GA will report 2 visit for the tab-browsing case even though it happened in the same time frame?
Just a question/idea.
What if in the GA code tracking we add
var pageEventTracker = pageTracker._createEventTracker('Exit');
and then in the onbeforeunload event we put
pageEventTracker._trackEvent('Exit', 'Exit');
Would GA be able to compute the correct time on page for this page? Has anyone tried it? I mean, GA may not have a following request to compute the time on page, but it is certainly receiving a hit and somehow knows the user is still in the page.
Just another idea. If GA doesn't use this events to compute the time on page at least we could pass the time on page – calculated via JS – to the event via the value parameter. Then in the event tracking at least we would see how many people spent how many seconds in their exit page.
Thanks for updating the post with tabbed browsing!
And also great that you explained how GA handles it, I didn't know that – yet another thing learned from your blog… ;)
Sheila : As a blogger I think Time on Site it is less useful for blogs. Mostly because people will come, they will read the latest post and they will leave. Especially your regulars. So I focus on other blog metrics.
From time to time I'll look at the Time on Site number.
One of the tools I use on my blog is ClickTracks and it allows me to analyze New Visitors separately in a very easy way. So what I'll do is from time to time I'll see what new visitors are doing, what pages interest them, how long they tend to stay on the site etc.
John : Hurray! I completely agree with you, and you are on!!
Peter : To your comment about 40% bounce potentially cased by tabbed browsing, if you were referring to Google Analytics then the answer is no. GA uses Method #2 described above.
In terms of Average Time on Site Goggle Analytics does what all other tools do. Take the average of all Visits to your site and report that (and that "average taking" includes bounce and no bounces etc). That would apply if you are looking at data in aggregate or if you segment it. For this, and other reasons, in the past I have recommended using reports that show distributions.
Open your web analytics tool and find the report that looks like this, every tool should have it (if you can't find it reach out to your web analytics vendor help center):
You can easily see how the attached image is so much more helpful in understanding what is happening on your site.
Zvika : GA will report one visit, see this link (or post above, Method #2)…
https://www.kaushik.net/avinash/wp-content/uploads/2008/01/time_on_site_impact-tabbed_browsing-linearized_2.png
Sergi : Great suggestions. I am afraid the Google Analytics help team might be best placed to answer the deeply technical questions. You can find them by clicking on the Contact Us link in the footer of any GA report – they will reply back in 24 hours or less.
Thanks everyone for your kind contributions and delightful comments.
-Avinash.
Sorry to have missed you at the eMetrics. but I hear you spoken to Ian.
The actual page view time based on the onUnLoad event is something that we have been doing for some time with our performance tag code. You're absolutely correct in that this results in addition log entries (three in total: Page Access Time, Page Loaded Time, and Page UnLoaded Time) for each actual page impression, but the results can be extremely insightful. On a basic level, this allows us to show slow loading pages (Page Load Time – Page Access Time), as well as individual page view times (Page UnLoaded Time – Page Loaded Time). In additional we often find that people have had a web page open for many days, which may be a result of power saving techniques such as hibernation or stand-by!
Additionally, we have been analysing tabbed browsing by comparing a pages stated referrer to the previous page in the context of the visit. Using your above example, the 'linear' sequence would result in the following:
Page
Home (Referrer = Google, Sequence = N/A)
Page 4 (Referrer = HP, Sequence = Home)
Page 2 (Referrer = HP, Sequence = Page 4)*
Page 3 (Referrer = Page 2, Sequence = Page 2)
Page 5 (Referrer = Page 4, Sequence = Page 3)*
The * represents the results of tabbed based browsing. Reporting on this, can help enrich the understanding of user behaviour.
Anyway, another great post, which I'll be pointing many people towards.
Sean.
I have posted on this great post on my blog as I find it extremely helpful.
That being said, Now after I've also read the great comments I am about to write a much deeper post on the whole concept of these metrics, their meanings and how to filter out the garbage.
Basically, once we realize the zero seconds concept, we must get rid of all these users while analyzing and counting. If we know that a zero seconds visit means that we couldn't actually measure the visit (and that's most visits!), it's best to discard these users and focus on the ones we CAN measure.
A second thing is with regards to John Marshall's comment(Hi John :-) ) about ROI/ROAS calculations – I think we all need to get awy from the solid numbers and get much deeper into the trends and movements – It does not really matter if my conversion rate is 1.8% or 2.6% as both numbers will probably be inaccurate to some degree. It is CRUCIAL to know when the number is MOVING and changing up and down.
I've just added a post with my thoughtsd and examples on why we should eliminate the zero time on site from the analysis, following the explanations and comments to this post.
Thanks for the very informative post, Avinash.
Hi Avinash,
Great post! The knowledge you provide us by this blog is extremely helpful for us.
And i have a question Avinash, In many organizations for some security reasons I.T dept. implement websense for blocking some websites of specific categories. So a lot of websites get blocked by this, Now if a visitor type the URL of such kind of websites he'll get blocked message on his PC but just before that he redirected to that website for a short interval which might not opened although.
Now i want to know that it'll count as a hit or bounce coz time interval is very less? or visitors are not counted at all by this activity?And if traffic are coming thro' Google or other search engines will it be display in referrals or not?
Regards,
Suchet Vatish
Suchet :It is hard to say definitively. I see two scenarios….
Usually the javascript tag for all web analytics tool sits at the end of the page and so all of the page has to render before the tag gets executed and sends data.
This means that if the redirect is happening quickly then the javascript might not have been executed and no data – no impact on time on site.
If the redirect you describe happens after all the elements on the page load (including the web analytics javascript tag) then data is sent – impact on time on site will be the same as described for bounce above.
Hope this helps.
-Avinash.
Suchet: The situation you are describing is like banning some websites using your organizations' proxy server filtering the websites to specific users or all users. If we type a wesite name ( or go through SERPs, or links on website, email, chat window etc. etc.), then the request will be first processed by proxy server and if it is banned then the request will end here only. So, there is no question of request reaching at actual website's web server.
Avinash, in response to this and to your earlier post at https://www.kaushik.net/avinash/2006/05/excellent-analytics-tip2-segment-absolutely-everything.html is it possible using Google Analytics to isolate visitors by their time on site? To be able to understand the site browsing difference betweens the long-duration visitors and the brief? And to completely eliminate the under-10 second visitors?
Thanks.
Hi Pat,
It is rather difficult to create a profile in Google Analytics based on the length of a visit. To do so you would need to customize the JavaScript page tag. Unfortunately there is no way to use a filter to create the data you're looking for.
What you can do is create profiles for important visitor segments based on attributes of the visitor (geo location, keyword, marketing campaign, etc.) This can provide insight into which segments may be affecting your time on site metric.
Here are instructions on how to segment visitor loyalty reports in Google Analytics. I hope you find it useful.
Justin
Hi Justin and Pat,
I just wrote another post here: http://www.ophircohen.com/2008/01/22/segmenting-visitors-by-average-time-on-site/ which explain exactly how to accomplish the segmentation you mentioned.
Only thing is, I don't use Google Analytics but ClickTracks. Still I think you should give it a try – use the trial and setup some data and see some insights which literaly blow your mind :-)
Oohir
Thank you for this site.
This is just what I needed.
I plan on returning here frequently.
Let me ask something about tab browsing you mentioned in this post.
I am confused the number of sessions for "Dealing with tab browsing # 2". One or two?
* Outcomes: One Sessions, visit “reorganized” by time stamps.
* Net net: 2 Visit. 1 Unique Visitor.
These two are conflicting each other?
If I am false, please let me know.
p.s. One Sessions -> One Session(misspeled)
The purpose I comment like this, just complete a great post if it needs. ;-)
– – –
NOTE: Minjae: My apologies, it should read 1 Visit and 1 Unique Visitor. I have fixed it now in the post. Thanks for pointing it out to me! -Avinash.
Hi Avinash,
I'm trying to get to the bottom of a problem we're having about the 'time on page' stat for pdf files. We have onclick GA JavaScript set up for pdf downloads, but how on earth is Google Analytics recording a time on page stat for these pdfs?
It's been flagged up by a client because the time on page for these pdfs is showing up as considerably higher (up to 11 mins) than all the HTML pages being tracked. If, as I suspect, it's not possible for GA to track how long someone has spent with a pdf open, where on earth is the stat coming from?
I'd really appreciate your help on this.
Thanks
Hi Jammit,
I have a feeling your tracking code referrs to counting clicks on links to pdf files rather than actually analizing time on pdf files.
Unless the GA code was embedded somehow inside the pdf (which I am not sure is possible – but will be happy to learn is it is) – It only counts the click on the link.
Ophir
My thoughts exactly – so my question is, where is GA getting the "time on page" data for PDF downloads from??
Jaamit – probably the same way all other "time on page" calcs are done. Delta between click/open time; and the next html click. 11 mins doesn't sound too unreasonable.
Also: Check the lunametrics blog. Robbin S did an experiment with some other funky GA stuff a few months back. You could easily duplicate and cross verify??
HTH?
– Steve
OK, assuming most people get PDFs opened in a browser window, most people wouldn't navigate away from a PDF via a link click – I'd say it would either be with the back button or closing the new window / tab. From reading this post it seems GA would not be able to measure the time from a close window action, so is it measuring time between the initial click and 'back' button click? :|
True, but with the HTML content on this site has an average time on page of 1 minute! Which leads me to think that there's something strange going on with the way it's measuring this stat. What I would like to know is, can I trust the 'time on page' metric for PDF files?
Jaamit: Let's dissect things a bit to get a richer understanding.
Time on site is computed using this "formula":
Time Stamp (last link hit) Minus Time Stamp (first link hit).
(Or look at the picture in the above post).
Let's talk a couple scenarios:
1)
+ Requested home page: 0900
+ Requested page two: 0901
+ Leave
Time on site: One minute.
Time on page two: Zero minutes.
2)
+ Requested home page: 0900
+ Requested page two: 0901
+ Click on link to download pdf on page two: 0902
+ Leave
Time on site: Two minutes.
Time on page two: One minute.
3)
+ Requested home page: 0900
+ Requested page two: 0901
+ Click on link to download pdf on page two: 0902
+ Dance for 13 mins to Ini Kamoze's Here Comes The Hotstepper
+ Go back to the site and click on a link to page four: 0915
+ Leave
Time on site: Fifteen minutes.
Time on page two: Thirteen minutes.
That should help you understand what is happening in your case.
Couple points of note:
# Pretty much every single web analytics tool will behave as above, GA or otherwise.
# Javascript based tools (pretty much all of 'em right now) don't track the download time of the pdf (even if they say that they do, push 'em and ask them exactly what they do).
# In your case, or in scenario #3 above, the whole pdf / download thing is a "distraction" to understanding. The pdf click was captured and reported, but what might be impacting time on site is the fact that the person came back to the site before the session was terminated (in less than 29 minutes).
# Above, #3, would happen to time on site even in this case: click page one, click page two, go to google.com and search for dresses, come back 28 mins later, click on page three.
Web Analytics is fun! [Seriously!!]
-Avinash.
Excellent, thanks for a very clear reply Avinash. Love the Ini Kamoze reference too (13 minutes…do you know of some secret extended version? lol)
Great Explanation, BUT (!!) I have an important question:
On google analytics, when "Time on site" is, for example, 00:03:42, does it mean 3 min's and 42 sec's? Or is it 3 sec's and 42… smaller parts??
I never see a topic is clear like this ! I'm clear about TS & TP mean now. GA shows all TS are zero for me, I wonder if my site can't open or too slowly. finally I know it's because high bounce rate.
thank you very much.
Thanks for this. Huge help.
It also begs the question(s)… as this turns to a KPI for monitoring those who travel to +1 pages, how does that change the landscape? Is there a benefit in knowing that the time on page is directly tied to those who visit a page and decide they want more?
Cheerio
@Trontastic
Just a comment to those interested in getting the real time on page. It's possible to get it two ways, triggering a fake pageview with the tracker inside the event window.onUnload (better in another profile as we would get a lot of fake pageviews) or via the new GA event tracking (now still in beta but available if you requested it) triggering a event when the user leaves the page, that passes as a parameter the time on page.
Hi Avinash, I enjoyed the post thoroughly and enjoyed the extra credit section even more. This post actually was the strongest when doing a site search on google when i was searching for an answer on how I can get Analytics to do ROAS reports. Currently I'm using a commercial technology but I would prefer to use Google Analytics – where I can parse the product revenue into the javascript action tag – can analytics do this??
Andreas: I am not sure I understand your question completely, but. . . . .
In this post check out item #4 and the links provided:
Google Analytics Maximized: Deeper Analysis, Higher ROI & You
GA provides a very flexible javascript for ecommerce tracking and it has the ability to capture lots of different kinds of information (this kind of robust and flexible functionality is also provided by pretty much all good WA tools provide).
-Avinash.
Thanks! Excellent info – you spent a lot of time – great value – had to tweet it –
I am setting up a segment and I can't find the answer to this.
What is the unit of value that Google uses for 'time spent on…' ?
David,
I believe time is measure in seconds in Google Analytics.
Best,
Justin
Many thanks for this explanation. I have been working with my analytics provider the last few days trying to understand how time on site is computed. As a non-technical person, I was having a difficult time understanding how the data was computed. This article is very clear. Thanks again.
Time on Page and Time on Site metrics are important, but they don't tell the whole picture. Also it depends on what analytics package you are using. Using Omniture SiteCatalyst or Google Analytics will show different results.
My apologies if I missed this among the many replies above, but I have a question: when a visitor has my site open in one tab and another site open in another tab and is looking at the other tab, does Google Analytics count that time in the time on site in my site? Does that change if they go to a new page in my site during that session or not?
Thanks!
Alex
Hi,
Everyday I compare site avg time on site through my alexa.com and find that it is higher than facebook.com, youtube.com, myspace.com.
What does it mean?
Alex
Having found this article, I've learned a lot today. Thanks for this valuable share.
Hi Avinash,
Its a wonderful post. Nowher have I seen the above topic explained soo clearly and simply.
Thanks dude. :-)
Could "Time on page" be considered a worthy metric for pages which are purely educational in nature?
Audrey: It could be, but as you would have noted in that post it is not possible to accurately capture that piece of data (i.e. without extra specialized coding or without special tools that add a lot of weight).
But the time someone spending on a page might not allow you to infer as much intent / success as you might believe.
Here is a blog post that has my point of view on how to measure pure content site, in this case measuring Government websites (can't get any more content only than the government!)…
Web Analytics Success Measurement For Government Websites
-Avinash.
I appreciate the information. I'm on my way to finding the answers to my questions. Thanks and God Bless!
How does GA define "activity" in the website?
Is it possible to have pages with 30+ minutes avg time on page registered in GA?
I understand that after 29 minutes of "inactivity" the session is terminated…
Thanks!
Gustavo: Consider these two scenarios:
1. You come to the page. You do nothing and you leave (or leave the browser open) after 35 mins. In this case the time on page can never be more than 30 mins. You did nothing on this page for 29 mins and at that point your session was terminated.
2. You come to a page. You do nothing for 26 mins. You reload the page (or do some action) at min 27. Then you read the page. Hit reload again (or do some action) at min 40. In this case it is possible for the page to have more than 30 mins of time on page.
This applies to pretty much any web analytics tool's behavior on standard settings and not just Google Analytics.
Hope this helps.
Avinash.
Thanks for your reply Avinash.
That's what I thought…
I just still can't understand why do I have several unique visitors (with repeated visits) to my home page with an average time on page of ~1 hour or more…
Is there any other activity (or action) besides clicking and reloading that can be considered part of an active user? (scrolling?, clicking non-items?, moving the mouse?)
Thanks again and greetings from Sydney!
Gustavo
Using Google Analytics:
This may seem a little novice, but here is my question.
What if the same scenario happened and Page 1 and 2 were coded with the GA code but page 3 was not. Would GA pick up the time on site for Page 2 after going to Page 3 (the non-coded page)? What if they left Page 3 to go to Page 4 (a coded page) would the time on site start over? Would the original referrer still be given credit?
Thank you in advance.
Jim: Assume this scenario, based on your comment:
Page One viewed at 1000 hrs
Page Two viewed at 1010 hrs
Page Three viewed at 1020 hrs
Page Four viewed at 1030 hrs
If page three is not tracked then that won't be in GA, or Omniture or WebTrends or anywhere. No javascript code, no data.
In this case:
Time on Page Two = 20 mins (1010 – 1030)
Time on Page Three = no data
Time on Site = 30 mins (1000 – 1030)
I hope this helps.
Oh and in this case since the the "hits" were in the same session all the referrer info is the same (as it was for Page One). If Page 4 was viewed 30 mins after Page 2 then the original referrer will be "lost" as the Page Four hit would start a new session (since after 29 mins the original session expired). This is how it is in every tool, GA does nothing different or extraordinary.
Avinash.
Thank you Avinash! The issue I am facing is that the Bounce Rate I am seeing for a specific referrer (search engine x)on GA is relatively low (between 5-10%) but the average time on site for that source is very low as well ~50 secs. The average time for the website overall is ~ 6min. Every page is coded properly. How is it that this referrer could send so much traffic that moves past the homepage but only stay less than a minute. Some of the "hits" even show as a 1 second visit??? What could be the issue with this referrer? Any input would be helpful.
Again, thank you in advance.
Jim: It is hard to identify the issue with the context in your comment. But when I run into issues like this I use surveys or other such mechanisms to collect VOC.
For example at the moment I am running a survey to collect task completion rate for my site. It has amazing insights into what people are doing. You can review my results at http://zqi.me/wakiss
I would recommend using qualitative analysis to get insight when quantitative fails. Better than guessing. :)
Avinash.
I had a follow up question about this post:
"Oh and in this case since the the "hits" were in the same session all the referrer info is the same (as it was for Page One). If Page 4 was viewed 30 mins after Page 2 then the original referrer will be "lost" as the Page Four hit would start a new session (since after 29 mins the original session expired). This is how it is in every tool, GA does nothing different or extraordinary."
Let's say the visitor in this scenario left after visiting all coded pages. Closed the browser and everything and 20 minutes later decided to run a search for the website on Search Engine X. Who would show as the referrer for this second visit? The seach engine/term used on the initial visit or would this second visit be seen as a different referral path with a different time on site, page, etc.?
Thanks in advance.
Avinash,
Nice post, Wondering if you know if the type of cookie effects the time calculation.
On one of our clients sites we changed all 3 cookies to be session (timeout set to 0). Now our time metrics are 2 to 3 times higher.
Any ideas???
Thanks, matt
Matt: When you set the timeout to zero what you are essentially doing is telling the analytics system to blow all the cookies away when the browser closes.
Additionally you are also setting the session timing cookie to ignore the 29 minute timeout signal, which will mean sessions that otherwise would have ended now could last all day (technically from 0001 hrs to 2359 hrs)!
By doing what you are doing you are ensuring at least two problems: 1. Every single visitor to your site will be counted as a new visitor to your site (so wrong Unique Visitor data) and 2. Average session duration (time on site) will also be messed up.
You can well imagine that #2 will also mess up your time on page for at least some pages.
Typically unless there is a magnificently profoundly good reason (which you might well possess) it is inadvisable to mess with how your web analytics provider has set up the cookies to work.
Avinash.
What is confusing to me is how to measure how one's site is doing. Each common measure besides bounce rate has its flaws.
Time on Site: What is the ideal time on site? You want it to be long enough that the visitor was actually able to accomplish something, but a lower time on site means your site is streamlined and intuitive enough that visitors go from A to B quickly (ideally).
Pages/Visit: seems like it would pose the some interaction/ease of customer navigation duality as time on site. What is too high and what is too low?
Do you look at these two measures, and if so what do you look for?
If you don't, what else should I be looking at (ecommerce site)?
Natasha: You should not be confused that you are confused. That is very common given the field we occupy. :)
There are a couple of metrics that are almost always useless (say % Exit Rate, unless you are looking at a structured funnel). But most other metrics fall in this category "this is a great metric for us because it helps us measure specific goal x for our website".
Time on site can be a good metric, Pages/Visit can be too, so can Bounce Rate and Visitor Loyalty and Task Completion Rate and …. so many more. But not all of them are right for you.
Here is a blog post with a framework you can use to pick the right metrics for your business:
~ Web Metrics Demystified
I also encourage you to embrace the Web Analytics 2.0 mental model, simply because it will help you answer your questions, like in your comment, in a much more sophisticated way than just using Clickstream. See the first part of this post:
~ Best Web Analytics 2.0 Tools: Quantitative, Qualitative, Life Saving!
All the best!
Avinash.
Dear Avinash,
I can not figure out the calculation of time on page(Page 2) of Sample Dealing with tab browsing # 2:
Tp (Page 2) = 4 minutes? (10:05-10:01)?
I think the time on page of page 2 should be:
Tp (Page 2) = Time of next page request (10:05) minus Time of page request (10:02) = 3 minutes.
right?
Nick: The most important thing to take away is that it will "linearize" the hits (put them in the order the links were clicked, and then compute the times.
Here is what happens in our specific example…
1000 visit starts
1001 click to new tab (Page 4), this tab is in the background
1002 click (in the original tab) to page 2
1005 click (in the original tab) to page 3
The time spent on page 2 will be 1002 minus 1005 and so three minutes (and not four as I had stated it).
Thanks for helping me fix it.
Avinash.
Thanks for explaining this. Someone has probably mentioned this elsewhere on here but I got some much more realistic Avg Time on Page data when I segmented to exclude average time on page greater than 5 minutes.
It'll vary from site to site what is a reasonable limit (since I'm in retail page dwell times are not particularly long and 5 mins is much more than enough to view any page on site unless perhaps it's account creation pages) but a bit of common sense can do wonders here!
Charles: Averages stink, as we both well know! :)
I typically look at the distribution of the time spent on the site and use that to determine "typical" behavior or "desirable" behavior or, in your scenario, choose the number to exclude.
In GA the report is called Length of Visit (in the Visitors -> Visitor Loyalty section) in the old version, and Engagement (in Visitors -> Behavior section) in the new version.
All other tools also have this metric.
Thanks for sharing your tip!
Avinash.
Hi Avinash
Great post as always! Wondering if there are any standard measures of these metrics? How do we know what's the time spent on a good website. Or is that time spent below or above category average or industry average?
Suchi: I personally believe that there is no good "time spent on site / page". Simply because the only good time is the shortest amount of time it takes someone to complete the task they are there for.
So I would stress an understanding of time on the site in context of task completion rate (using a survey based mechanism). Or if you are able to use something like the new integration in http://www.4qsurvey.com which allows you to see survey data with analytics data. You can see the time for visitors who are most satisfied. That can be your guide.
-Avinash.
Hi Avinash, I hope you're still keeping an eye on this post.
As you're one of the leading Google Analytics experts, I wonder whether you know exactly how bounced visits affect the avg. time on site and time on page on Google Analytics?
I'm asking because for me it doesn't make much sense to include bounced visits (exits) in, especially, time on page calculations as if everyone spends a minute reading a page on your site but half of those people then leave your site, you'd be reporting on 30s time on page..
Thank you
Tomas: If you look at this part of the blog post "The Case of Bounce / Single Page Sessions" it clearly outlines what happens to time computation for a bounced visit in any web analytics tool.
As to if they should or should not be included in the computation of overall metrics… all tools (Omniture, WebTrends, GA etc) include bounced visits in the computation.
But if your tool has an Advanced Segmentation option then it is very easy to create a segment for "Time on Site Greater than 1 second" or something similar to exclude all the bounced visits. When you apply that segment to any report (say Visitor Overview) then you should be able to see "real" time on site / page or just the visits or keywords for non-bounced sessions.
Avinash.
Hi Avinash-
When I'm looking at the data for a time on site goal, I also see a set of goal completion URLs associated with the time on site goal. What does that mean? Are these the pages that visitors were looking at when the time on site goal was reached?
Thanks!
Liz
I was wondering if there was a report in Google Analytics that will allow you to see the distribution of the time spent on a specific page by page views, similar to the Engagement reports by visit in the Audience section.
I know we currently get average time on page, but it would be great to see what % of the pageviews are less than 10 seconds, between 11 and 59 seconds, between 1-3 min, and 3 min plus (you get the idea).
Can you let me know if there is a way to create a custom report for this?
Eric: I'm sure there is a sexy way to do this using the Google Analytics API, but assuming that you don't want to go down that route (at least initially)…
You can simply create advanced segments for the time buckets you want. Then apply those segments to the report of that page and, boom (!), you have what you are looking for.
To get you going here's a segment you can import into your GA account, it will segment everyone who spent more than 3 mins on a page: http://goo.gl/R1Vx1
But remember one of the core lessons of this blog post. If that page was a landing page, and people exited then their "time on page" will be zero. So when you apply my recommendation above you are only looking at people who came to that page and saw at least one other page on the site.
Just remember that and you're all set.
Avinash.
Avinash, this is a great post. There's something even more obvious you touched on that I apparently always misunderstood…
I always thought bounces were simply very SHORT (ie, < 10 seconds) 1-page visits. Are you suggesting that ALL 1-page visits are considered a bounce? Since they have no way of measuring how long was spent on the first page?
So this means the person with a lot of info on a single landing page would be likely to have a 100% bounce rate.
Fascinating. Thank you for sharing.
Jeremy: You are right, the standard definition of Bounce Rate in pretty much all web analytics tools does not have anything to do with the time spent on the first page. If a session (visit) has only one page view recorded in it then it will be considered a bounce.
Time can be tricky to accommodate in a single page view session (for reasons outlined in this post). There are timers you can use, but they are often reading too much into the behavior.
Recently I'd shared a piece of code that you can use to record "engagement" with the first page. What the code does is record that someone scrolled on the page. We can use that as a proxy for the fact that they read the page. You can learn more about how to use this here:
~ More Accurate Measurement of Bounce Rates for Content Sites
-Avinash.
Hi Avinash,
Again another good read about time on site measurement in detail.
However, my question was still not answered from reading this.
It's regarding Browser caching of 12h for ga.js file by default.
Based Google Page speed performance guidance, we should have at least 1 week browser caching expiry period but why ga.js is not following this guidelines set by Google Website performance (Page speed)?
I tried sourcing answer for this question but unable to find any reputable answer that really answers my question.
Can we set expiry for ga.js for more than a week? Does it effect's its data collection such as time on site, unique visits, visits, etc? If this is not possible, then I would strongly suggest to the developer of Google page speed to exclude ga.js in their browser caching evaluation.
Hope to hear your point of view.
Thank you so much again for your continuous support.
Ha, finally found an article that clearly explained these stuff. I have always wondered how Google calculates some of the metrics, now I know.
Hi Avinash,
All of these calculations revolve around average time spent per visit.
Is there a way in Google Analytics to get the average total time spent on site per unique user? IE If a user comes to the site and spends 2min and then comes back three days later and spends 3 min, the average total time on site is 5min over that timeframe.
Is there a metric like this in GA, or can we only get per visit? Basically, if people are spending shorts amount of time per visit, but are coming often, avg time per visit doesn't really tell the right story.
Thank you,
Lindsay
Lindsay: No there is not. Or in other tools, unless you buy the expensive data warehouse versions though even then this might be difficult to accomplish.
Regardless of the tool you use, I have to admit that what your are postulating might not be a valid hypothesis. If you are worried about your short visits, run a Task Completion Rate survey, collect a scientifically valid sample, figure out why there are so many short visits. Then you don't guess, you are enlightened by the users.
Avinash.
Hi Avinash!
I am 50 pages into your "Web Analytics 2.0" and I must said I am very impressed. I like the way you present the material, I like how easy and witty it is. Great Job!
I came across this web page when I decided to check if things had changed since publication of the book. Is it still impossible to get precise "time on last page" with GA?
Actually, Is there a resource on your site where I can track all or some of the things that by now outdated in the book? I am sure a lot of things changed in the past 4 years in Web Analytics world. Also, is there a "Web Analitycs 3.0" on the horizon? :)
Thanks in advance for your time reading and answering all these questions here
Alexander: You can use the blog as a companion to the book.
My vision with WA 2.0 is for it to teach now people think and the processes they use to approach analytics and driving results. The book is not supposed to teach now to teach which buttons to press in a tool. Hence the book is very current, even if on some elements I have no models and framework that will eventually make it into the next book.
To your specific question, no web analytics tool can measure time on last page. It has simply to do with how all of them collect that data. With all tools, including Google Analytics, you can implement hacks to approximate the time on last page. It is a bit of an extra effort but you can hire a consultant to help you. For GA you'll find a list here: http://www.bit.ly/gaac For other tools, please check the vendor's website.
-Avinash.
Hi Avinash!
I just wanted to know that I found a free web analytics tool that can in fact track all single-page visits (as long as they are 15 sec or longer).
The name of the tool is Yandex.Metrika (http://metrica.yandex.com/) and it is provided by Yandex – russia's largest search engine company. Among other features that stood out are heat click map (similar to paid CrazyEgg) and Webwisor (recording of visitor browsing your pages). I don't think there are any other free tools that offer these premium features free of charge.
Alexander: Thanks for the link, Metrika looks quite good.
You should be able to get any tool to measure time for single page visits, including SiteCatalyst, Google Analytics etc. You just have to implement some custom code.
For Google Analytics, you can copy the code that is outlined here: http://goo.gl/TDKDaN
-Avinash.
Dear Avinash,
Thanks for this post explaining and solving some of the mysteries. It's really hard to find valuable information about the way some Google software work. Probably by purpose.
I love to read your articles they always contain valued information to improve my skills.
Hi Avinash,
This is just the explanation I was looking for. Seriously, google to document what they do in a human language :).
Thanks for the information.
Thanks so much for this awesome post.
I love how it's still relevant after five full years.
Hi All,
I just checked with Adobe Analytics (prev known as SiteCatalyst.)
Unlike GA, Adobe Analytics excludes bounces from its Time on Site / Visit Duration numbers.
Thanks Avinash!
Realise you wrote this a while ago, but it's still very useful :o)
This has got me thinking, does Google Analytics use linearization for any other reports?
For example the navigation summary in the Behaviour > Site Content > All Pages report?
The reason I ask is because I've noticed in the navigation summary for my site that people have arrived at Page_2 from Page_1 but Page_1 doesn't contain any links to Page_2. How would that be possible? I wonder whether linearizing the hits could be to blame or is there something else happening?
Thanks, Katie!
Katie: I hesitate to provide a definitive answer about your website and user behavior on your site and other such small contextual things.
But in general you would be correct in assuming that user data is "linearized" and that would cause you to see some of the behavior you see in your data.
If you want to dig into it more please work with a GACP, here's a list: http://www.bit.ly/gaac
Avinash.
Many thanks Avinash for your reply – it's very useful to know that this linearized data could apply to other things in Google Analytics as well.
Katie
Hey…Suppose I am visiting my site in tab1 and at the same time I opened some NEW website in tab 2, I keep switching on the tabs from 1 to 2 and 2 to 1, Is this beneficial for my site?
For eg. I visited my site at 10.00 and go to a new tab and visit some other site at 10.5 but I return beck to my site in the first tab at 10.9, is this beneficial? Will Google take it to be 9 minutes on my site or just 5 minutes?
Thanks and Regards,
Saloni
Saloni: You have two questions:
1. What behavior is beneficial or not, that really depends on what your site is and what behavior you value. There are no standard answers sadly.
2. As to what GA will capture…. GA will collect all the hits that happen and "linearize" them (see the post for more on this). It will put hits that are made until 29 minutes of inactivity into one session. In your case, it will be nine minutes for visit duration.
Avinash.
Hello Avinash, thanks a lot for the explanation. I am currently reading your first book and needed more insight into this subject. Is this still valid as of today?
I'm quite confused with G Analytics. There are metrics such as "Avg. Time on Screen", "Avg. Session Duration" or "Avg. Time on Page". What is the difference between those?
And how can you distinguish users that "really" bounced vs. those that simply visited a single page? (In your old book, you define a bounce as a user that visits a website for less than 5 seconds. G Analytics defines a bounce as a user that visit a single page (avg. time on page for those users is therefore always 0:00).
Is there a way to determine the % of users that… "visited a single page AND stayed on the site for less than 5 seconds"? That would be a real bounce in my terms and provide real insight (thinking about landing pages for example.. in order to distinguish paid keywords that generated interest vs. those that didn't).
Thank you for your support
Pablo: Let's unpack a couple of different threads in your comment.
Every leading web analytics tool just uses one way to measure bounce rates: Visits with just one page view recorded. (Technically speaking, just one hit recorded.)
The amount of time you spend on that page is irrelevant.
You can create different segments, like for example using five seconds on a page, to explore the behavior, it can be valuable, but that is not what the tools are using. You would do this separately from the standard reports and standard metrics.
If you want to play with other ways to measure bounce rate, please see this post where I share code you can use to measure bounced visits "more accurately" (notice the quotes, that is just my point of view): More Accurate Measurement of Bounce Rates for Content Sites
Finally… average time on screen is a time metric for your mobile applications, average session duration is the new name for average visit duration (how long someone spends during their visit to your website), and average time on page is a metric represents how long someone spent on a page (unless it was the only page viewed in a visit or the last page viewed in the visit).
I hope this helps.
Avinash.
Hi Avinash, thanks a lot for your prompt reply.
I think Yandex does this in fact. (Yandex Metrica). They measure "accurate bounce rates". It counts people who visited only 1 page AND left the site in less than 15 seconds. If they stay for more than 15 sec but see only 1 page they still don't get counted as bounced visits. Your script is also helpful, thank you. However, it would only work if the page was long, and even then, people could be scrolling just to "glance" the content and then leaving right away. But I guess accuracy is hard to get, even with web analytics.
That's hell of an interesting thing Kaushik!
Never really thought like this. Congratulations on discovering this metrics.
I definitely want to share this with my fellow webmasters. This can make a lot of differences I see. Thanks a lot man!
This is indeed the best explanation to all my answers related to time calculations for google analytics.
Related observation, Google analytics is giving me
Tp (Avg Time on Page) = 2min, and
Ts (Avg Session Duration) = 50sec
Your explanation does not quite cover this. What else is happening?
Paul: It is difficult to pin-point what is going on here. You've not mentioned where you are seeing this (which report, for which dimension, segmented data or not, sampled, etc.).
But my initial thought is that something is amiss somewhere in your analytics implementation/reporting because those two numbers look odd.
If you need specific help in figuring out what might be up, please engage a GACP who can dig deeper and provide assistance. Here's a list: http://www.bit.ly/gaac
Avinash.
Thanks for replying. After some searching, it comes down to the fact than "Time on Page" avg does not include bounce sessions; only sessions with 2+ pages are included in the sample. "Session Duration" avg does count bounce sessions. Not comparing like for like.
Hi Avinash
Thanks, it's a great post.
I am wondering about how analytic tools set cookies with separate tabs in the same browser. E.g. when User see product listing page and he clicks on each add in separate tab so analytic tool might set cookie in order to track conversion (In each add product id is different).
But different tabs come under the same session. How Google(doubleclick.net) solves this problem or we cannot track conversion with same browser with multiple tabs?
Thanks!
Prashant: Here's a blog post that you might find to be of value:
~ A Primer On Web Analytics Visitor Tracking Cookies.
To your query… Cookies, by any tool, are set at a browser level. As the person experiences a website, the hits that go back and forth, regardless of from which tab, reference that same cookie set at a browser level.
-Avinash.
This post helped me to understand how time on website and time on a page is computed.
Thanks for Posting.
What is confusing to me is how to measure how one's site is doing. Each common measure besides bounce rate has its flaws.
Time on Site: What is the ideal time on site? You want it to be long enough that the visitor was actually able to accomplish something, but a lower time on site means your site is streamlined and intuitive enough that visitors go from A to B quickly (ideally).
Pages/Visit: seems like it would pose the some interaction/ease of customer navigation duality as time on site. What is too high and what is too low?
Do you look at these two measures, and if so what do you look for?
Donovan: Here's a post that will help you pick the best metrics for your business:
~ Best Web Metrics / KPIs for a Small, Medium or Large Sized Business
And, here's a post that will help you benchmark your performance so you'll get a stronger sense of what's too high or too low:
~ Benchmarking Performance: Your Options, Dos, Don'ts and To-Die-Fors!
Avinash.
Thanks for this post explaining and solving some of the mysteries. It's really hard to find valuable information about the way some Google software work. Probably by purpose.
This is just the explanation I was looking for. Seriously, google to document what they do in a human language :).
Thanks for the information.
Hi, I'm trying to understand how this works when users click a link and it opens a new window.
I understand case #2 in this article and tabbed browsing within the same window, but what about when a link opens a new window? Would that start a new session?
Some of the comments and articles I'm reading interchange browser tabs (within same window) and new window, when I'm thinking the analytics would behave differently with a new window.
Can someone please shed some light on this? Thanks for your help and the amazing article.
Jennifer: I am so sorry to reply to you so very late. For some reason your comment was trapped in the spam folder.
It does not matter if it is a new window or a new tab, as long as the person is on website A. But. If during this time when they have tabs with pages from website A open, and windows with website A open, they visit website B which has a link to website A and the person clicks on that link then the session will be closed and a new session will be opened on website A.
All the best!
Avinash.
Avinash:
You mentioned that "for some types of businesses, avg time on site (now avg. session duration) is a critical metric."
What are some examples of this type of business?
Eric: Entities that primarily deal with content would fall in that category. Think of the New York Times, the numerous sites where video is the primary purpose (think of twitch as an example) or even music (think of spotify), and others.
This opposed to for example transactional sites, ecommerce obviously but even support websites, corporate blogs, or sites where you'll make offline decisions (say buy a car). In these cases there are much better metrics than time (because you never know if more or less time in these cases is a good thing).
Avinash.
Thanks Avinash. Our company website alvarezandmarsal.com is in the consulting category.
We rely heavily on content production, and it is not a transactional site. However, sales come from offline negotiations, often many months down the road.
What would you say is a good metric for us to focus on? (Right now we include Avg. Session Duration and Time on Page in our report)
Thanks,
Eric
Eric: I recommend creating a Digital Marketing & Measurement Model to identify what you should track.
There will be two very important elements you'll need. 1. An understanding of purpose that is clear. 2. An identification of the micro outcomes (this post will help with that: The Biggest Mistake Web Analysts Make… And How To Avoid It!)
Avinash.
Hello Avanish, I hope you’re still watching for questions and comments on this topic.
I’ve been wondering about the accuracy of time spent on site for two years. I have an online business that’s very local to the city I live in. When I see time on site from other locations, time spent is very low. When I look at time on site for my city, time is very high, especially since I have very few products and pages.
My main product is unique and the first of its kind. It’s a pair of marble Replicas of iconic and beloved statues we have in town. Average time spent is about 8 minutes per Google Analytics, even when I filter out my own IP address. Yet, sales of the product are low online. They do much better in a gift shop in town that carries them.
The price is the same at my site and the gift shop. What could the problem be at my website that prevents conversions? I’d appreciate your opinion very much. Thank you, Gail
Gail: The lions are beautiful, congratulations.
I have an hypothesis. Because the lions are so unique, people want to learn more about them. They are also expensive – comparatively. Hence, for people without a connection with the city there might be less of a tendency to buy. Not an issue when people walk into your gift shop.
There could be other reasons. If you would like to use an onexit survey on your site, that might be a good way to ask the visitors directly that question (or ask people who buy on the site why they bought). You can work with a consultant who can help you look at the data and draw out more insights, you'll find a list here: http://bit.ly/gaac
Avinash.
google analytics confuses me but ive seen many comprehensive guides lately. thanks for the post