When pulling segmentation data through the Google Analytics API for April 2011, I was getting 0 every time. After several head-scratching hours I could find nothing wrong with my software, so I ran a report on the web and got the same result.

It seems that source segmentation is unavailable for April 2011. I’ve tried across several accounts, UK and US, large and small.

Most other segments seem to work ok.

What’s surprising is that I can find no mention of this anywhere.

Has it affected your accounts?

Update 10th May 2011

Google have acknowledged the issue here

This is all according to a simple but effective infographic from FFunction. Notice how they use the ‘Sleep loss is a killer’ line to draw your attention and keep reading – if only all subjects were that hard hitting your audience.

I can only assume the owner has either gone bust or something unfortunate has happened to them.

After a lot of work I have finally managed to get my domains released from Webmania and transferred over to a reputable host.

If you are in a similar position, read on.

How to transfer your domains and hosting away from Webmania

Webmania is a reseller of OnlineNIC. If you have tried and failed to resolve your problem with Webmania directly, go to OnlineNIC.com and register within their support section before opening a new ticket, explaining the issue.  They are getting pretty sympathetic to Webmania customers now, and usually help.

You need to specifically ask them to do the following:

1. UNLOCK each domain. This is a setting Webmania enabled to stop unauthorised transfer of domains away from them.
2. Change the administrator email address to your own email. Without this, all domain transfer requests will go to Webmania, they wont respond and your transfers will fail. Can you tell I tried this several times before realising?
3. Provide the EPP codes. Your new host will ask for these to initiate transfer.

Note: for .co.uk domains, you should bypass OnlineNIC and go straight to nominet.org.uk who will help you out in a similar way.
Now, register with a new host. I highly recommend Netmotivated. They now host all of my sites (this blog is about to transfer as I write this), and I know the operator so can vouch for their business standards. All support request are dealt with quickly and by UK staff.

Had a similar experience? Or do you know where the Webmania owner has gone?!

Drop a comment and let me know

Dan

Update 15th May 2011

Webmania have been appointed a liquidator (source). More interestingly, the owner Nathan Barnes set up a new hosting company the very next day – http://www.yorhost.net/ Needless to say I wouldn’t recommend rushing to sign up to this one

Saw this and loved it.

This isn’t a map. Each line represents Facebook friendships between two people, to the nearest city. I was looking for the source and it was actually made by Facebook intern Paul Butler. What an innovative use of data.

Here is the full resolution.

A couple of people have asked me about this so here it is.

Step 1: Cpen the project you want to copy from, and open the specific report you need.

Step 2: Now open the project you want to copy to, or start a new project. Don’t close Discover at any point.

Step 3: Go to File -> Recent workspaces -> and select the tab from the previous project you want to paste. Rather than opening the old report, it pastes it into current the project.

Done.

I’m not sure if this is a feature or just a useful bug, but it works all the same.

The content isn’t what I’m showing here, but the effectiveness of presentation.

Here are some of my recent favourites.

Undersea cables of the world

Click image for larger version

A chap by the name of Michael Anderson liked infographics so much he made his CV into one. I’m not sure this’ll catch on, but it looks cool.

Click image for larger version

The famous Billion-Dollar-O-Gram by David McCandless.

Click image for larger version

Both David’s recent TED talk and his book Information is Beautifulare well worth checking out:

The magic bean shop and the fries that bind us

Click image for larger version

How long will it last?

Click image for larger version

The key difference between instant and the old search is that the visitor does not need to type their whole search term as Google will make intelligent suggestions based on the first few key strokes.

The good news is that this doesn’t affect the keyword field in Google Analytics – the whole keyword is still collected rather than just the initial characters.

What we can see though is how much of that search term was typed by the visitor. This will be particularly useful in diagnosing bounce rate if certain terms are frequently misinterpreted by Google.

There will probably be a new metric in Google Analytics soon to show this variable, but in the meantime you can capture it yourself and keep ahead of the game by using a custom filter to push the part-keyword into one of the custom variables available to you.

This works because Google inserts the partial search string into the referring URL. So for example if I search for ‘analytics eva’ but never finish typing, the URL would be caught under the variable ‘oq’, shown here in red:

http://www.google.co.uk/#hl=en&xhr=t&q=analytics+evangelist&cp=13&pf=p&sclient=psy&aq=f&aqi=g4g-o1&aql=&oq=analytics+eva&gs_rfai=&pbx=1

To capture this automatically, set up a new profile in Google Analytics, eg ‘Source: Google Instant search’ and implement the following two filters:

Click to see a larger version of this image

Click to see a larger version of this image

Sit back, wait a few hours, and then you should be able to see these partial search terms under the user defined value metric in reports. Results with “(not set)” in the field indicate that the visitor did not use Google Instant on this visit. These are early days but I think we can expect these numbers to grow substantially.

Click to see a larger version of this image

How do you get to see them? The easiest way is to visit the site in Firefox, open FireCookie (F12), navigate to the Cookies tab. You’ll now see all the active cookies for that session.

I usually put “__” (that’s two underscores) in the search box on the right and hit return, this cleans out most of the non-GA tags.

What you’re left with is something like this:

The first thing to notice in this example is that it’s not just my site that’s tracked your presence here – my twitter and backtype plug-ins have each dropped a cheeky tracking cookie in too. Of course we are just looking at Google Analytics cookies here so there could be more from other vendors (I’ll cover that in a future post).

What does this all mean then? Google Analytics cookies are made up of six key variables:

• __utma =<domain>.<visitorID>.<first visit>.<last visit>.<current visit>.<session>
  This defines you as a site visitor. It’s a ‘persistent’ cookie which means it’ll remain on your device and identifies you as the same visitor on each visit (unless you delete your cookies of course).
  <domain>: a hash of the domain where the cookie was set
  <visitorID>: a random number which identifies you as an individual
  <first visit>: the unix timestamp of your first visit to the site
  <last visit>: the unix timestamp of your last visit to the site
  <current visit>: the unix timestamp of your current visit to the site
  <session>: the count of your sessions (visits)
• __utmb
  This is the session identifier. It is non-persistent so it dies after 30 minutes of non-activity on the site.
  
• __utmc
  Similar to utmb, this is the second session identifier. It is non-persistent and dies when the browser window is closed. By using both utmb and utmc together, Google knows when your session has come to an end.
  
• __utmz =<domain>.<current visit>.<session>.1.utmcsr=<source> |utmccn=<campaign>|utmcmd=<medium>|utmctr=<keyword>
  I generally find this is the most useful cookie to look for. It tracks the source from which you entered the site, known as the referral. It is persistent, but will update to the latest source method, unless you come to the site direct in which case it retains the previous source data.
  <domain>: a hash of the domain where the cookie was set
  <current visit>: the unix timestamp
  <session>: the count of your sessions (visits)
  <source>: the source (eg google)
  <campaign>: the campaign (eg organic)
  <keyword>: the keyword (eg pies)
• __utmv =<domain>.<custom var>
  This is used to store custom variables, so it will not be set unless the site is using this functionality.
• __utmx
  This is used to store the variation of the page viewed if the site is undergoing a Google Website Optimiser programme, so it will not be set unless the site is using this functionality. It ensures that the same version of the page is used if the visitor views the page over multiple visits.

I find in practice, the most common use of this information is to check referrals are registering correctly.

The most common problems you may need to look out for are javascript or flash redirects which do not pass referral information correctly, or internal site links which overwrite the referral information.

‘Ah, filters’ I thought. So I did the same thing on a completely unfiltered profile. Same result.

‘Non Javascript enabled users!’ I thought. Nope, as that could only account for <5%, probably under 3% now (source: w3schools).

Browser incompatibility with our JS then maybe? Doesn’t seem to be the answer either, as the browser visit to sale conversion is pretty consistent.

What’s odd is that this is happening across both of our major sites, with completely separate infrastructures, one with asynchronous tags and one without.

And I’m not alone, there are a stack of forum posts and group threads about this subject from online retails around the world. Similarly to other users who have reported this issue, I can find no common denominator in these missing transactions.

So far in my research I have looked at the following hypotheses, mostly sourced from this Google thread:

• Users have Javascript disabled. Disproven: all research indicates the % of users with JS disabled is sub 5%.
• Google is intermittently failing to process transaction data. Seems unlikely as not all retailers are seeing this phenomenon. Secondly I have tried to count sales through pagetracking alone and still have a 20% gap, meaning the problem is unlikely to be solely linked to commerce data processing.
• Conflict with Sarafi 5. Seems unlikely as I can see several version of Sarafi posting transactions in the logs, and Sarafi’s market share isn’t large enough to drive this gap.

Does it matter?

In order to recognise trends and optimise journeys for all users, we need a data set which encompasses all users. That said, in most circumstances capturing data including 80% of users gives a sufficient trend to identify drop off points and optimise for your key users.

Web analytics is not the tool to use if your requirement is to explicitly count and record large volumes of sales, that job should lie with the system which processes your payments or stores order details. Analytics tools are generally designed to amalgamate large quantities of user data into actionable insight, and in that it seems, Google is doing a great job.

That said, I really want to know where this 20% is hiding.

Do you have this problem, and have you solved it?

The tools below are my most used add-ons. They increase productivity by lowering the amount of time it takes to obtain data and increase the time you can spend on insight and action.

Personally I’m not a fan of companies who offer to generate reports and dashboards from analytics for you (free or paid) – they are missing the point. It’s not about massive reports and polished dashboards, it’s about finding functionality that can be optimised, testing improvements and measuring impact, and squeezing the best possible user experience from your online offering.

This is an open and ongoing post which will be updated. If you have a tool you use to increase Google Analytics productivity, please let me know.

• Firebug for Firefox
  This has got to be the most powerful Firefox add-on for web developers, but it’s also perfect for checking analytics tags.

  Download Firebug

  

  Those who do this a lot may also want to check out the paid solution Wasp.

• Greasemonkey & GARE Dimensions add-on
  Greasemonkey is a Firefox add in to allow custom scripts to run on certain websites.

  Once installed there are shed loads of Google Analytics scripts out there, but I’d recommend this new GARE script because it allows much greater definition in tables, particularly useful when pivoting.

  Normal Google Analytics selection menu:
  

  Google Analytics menu with GARE installed:
  

  Download Greasemonkey for Firefox and the GARE Dimensions add-on

• Excellent Analytics Excel Plug-in
  So those familiar with Omniture’s Excel Datablock plug-in will recognise this as a similar solution for Google Analytics. Essentially this is a plugin for Microsoft Excel which will automated the process of downloading a table of data.

  The beauty here for me is not that you can make loads of tables in Excel (you can do just by exporting in a CSV, and why would you want to?) but that you can automate the generation of reports or enable others to quickly update their own reports with little GA knowledge, freeing up time.

  Excellent Analytics report design menu:
  

  Data is automatically fetched and launched in Excel, and can be updated in table:
  

  Download the Excellent Analytics Excel plugin

• Campaign URL Builder from Savio.no
  Users of Google Analytics will be familiar with the standard URL builder from Google which is a quick way of slapping campaign variables onto the end of your marketing URLs and enable easy tracking and measurement.

  This tool takes that one step further by building in first/last referral preferences, the ability to use a hash parameter, and automate URL shorteners.

  

  Launch the tool from the Savio website