In a number of previous articles we've talked about the problem of spam referrals appearing within Google Analytics data. This problem now appears to be growing exponentially and without taking steps to exclude this data from your reports (changing your tracking / configuration within Google Analytics or using advanced segments to remove this data) your reports and any insights derived from them could become meaningless - or worse still highly misleading.
We're hopeful that Google will roll out a solution soon as for some unfiltered views we are now sometimes seeing that well over 50% of the traffic is spam - in a few cases equating to over 50,000 visits within a relatively short time frame.
Spammers are now also getting smarter - solutions previously described to use your website hostname as a filter are now encountering problems and maintaining lists of referrals to exclude requires ongoing proactive management with new sources appearing regularly.
In the meantime we've started work on a list of spam referrals, automated traffic and other suspicious referrals that you may want to exclude. This list won't be fully comprehensive but we'll endeavour to keep it up to date based upon what we encounter. We're hoping to publish this later this week.
Solutions such as that proposed by LunaMetrics (see link in the following article) may become increasingly essential to maintain the integrity of your reporting until such a time as a solution is available via Google.
Changing spammer tactics include:
- Adopting legitimate source names (but perhaps adapting them via adding a subdomain)
- Changing the hostname
- Using minor misspellings so they appear legitimate e.g. theguardlan (rather than guardian) or Inked.in (rather than Linked)
- Using new referral sources
- Use variations of the top level domain (e.g. .org or .info)
- Dropping suspect metrics - e.g. a 100% bounce rate or 100% new sessions can raise a red flag and so spammers appear to have changed tactics to often avoid these
If your site uses a site search facility you will probably want to know what your users are looking for. This kind of information can be particularly helpful in two areas:
- Mining for keywords – it’s likely that people will use the same search terms on your website as they are using in a search engine so knowing what they are looking for can help you create keywords lists for better targeted paid advertising campaigns as well as SEO
- Optimizing landing pages – visitors might use the site search on a landing page if they are having issues in finding relevant content. This can flag potential navigational issues with your site.
By default site search is not enabled in Google Analytics and so you will have to do this in the Admin section of the account for every view you might have created.
To do this simply go to View Settings and switch Site Search Tracking on. You will need to specify the query parameter search terms are associated with.
Bounce rate is a key engagement level statistic that many website owners rightly pay a lot of attention to. If you are unfamiliar with the concept then a bounce occurs if someone lands on your website and then leaves without any further interaction (e.g. clicking through to another page). The bounce rate is simply the percentage of visitors that leave.
In practice a bounce can be a little more complex depending upon how you've configured Google Analytics. If you have a standard setup then a bounce occurs when someone leaves at a landing page without viewing any additional content.
With more advanced tracking however - for example if you've implemented virtual page views or event tracking - things get a little more complex. If a virtual page view is generated e.g. by someone downloading a file from a landing page then that additional interaction will adjust the bounce rate.
Google have announced a re-branding of Google Webmaster tools as "Google Search Console". The change follows an announcement earlier this month that the "search queries" report has been updated to become "search analytics". The new "search analytics" report makes it easier to segment data by device, pages, countries and search type.
For further information on the launch of Google Search Console please click through to the Google Webmaster blog
"Out of the box" Google Analytics captures a wide range of visitor activity on your site. By default data is collected for each page view generated.
Sometimes however you’ll need to track additional interactions – for example visitors playing videos, downloading files, progress through complex forms where URLs don’t change or clicking onto outbound links. To do so Google Analytics offers two tracking options – to either capture such interactions as “events” or “virtual page views”. Tracking can be configured programmatically on your site or using customisable ‘tags’ within Google Tag Manager. So which should you choose and what’s the difference between them?
Looking at events first – once configured these appear within the Behaviour / Top Events report. Events allow you to capture three levels of data – a category, action and label. The latter for example could allow you to create a category such as “Downloads”, an action associated with these (e.g. “click”) and a label – for example the file name. Data captured in this format allows you to aggregate data for particular interactions with your site and easily identify e.g. what the most popular downloads from your site were.
Virtual page views
Virtual page views however appear within your main content reports (e.g. Behaviour / All Pages) alongside other standard page view data. As such it’s considered best practice to always include some identifier for virtual page views within the page URI in order to be able to distinguish these from standard page views - for example page URIs could start with “/virtual/” or end with “-vpv”.
Effects of additional tracking
Bear in mind that virtual page views will inflate the average pages per session metric – as such you may want to create an additional view that excludes these. Both virtual pages views and event tracking can also influence your bounce rate – a virtual page view will be seen as an additional interaction and so if e.g. someone downloads a file from a landing page and you’ve set this to be tracked as virtual page view and the visitor then leaves this won’t count as a bounce.
With event tracking you can choose as to whether the event influences your bounce rate or not. By default a “non-interaction” option is set to ‘false’ meaning that events will influence bounce rate – set it to ‘true’ if you don’t want bounce rate to be influenced. For example if you had an event configured to capture scroll depth you may well decide that this should not influence your bounce rate.
Which should you choose?
So having understood the basic differences between virtual page views and event tracking which should you choose? This largely (but not entirely) comes down to how you wish to use the data and whether you want to understand navigation and visitor journeys or to aggregate data for particular visitor interactions.
For navigational purposes virtual page views have the advantage that you can configure them as funnel steps within Google Analytics goals – something not possible for events. So if for example you want to track progress through a complex form (where the URL doesn't change) with the end point configured as a goal having a series of virtual page views would allow you to configure funnel steps and allow you to analyse performance via the Goal / Funnel Visualisation report.
You’ll also be able to see these pages within the reverse goal path report – which shows up to the three page views that occurred prior to a goal being completed – and also be able to visualise progress through a funnel via the goal flow report (which will include loop backs between steps). In addition you’ll gain access to the page level “Navigation Summary” reports and the Next Page and Previous Page dimensions.
Tracking activities as events however will often make more sense. You can easily get to summary data that shows how many times a particular activity was triggered and if you’ve constructed the category / action and label well you’ll be able to drill down to additional levels of detail. Events, as mentioned above, can also be configured not to influence the bounce rate and certain activities – such as a click onto an outbound link wouldn't make sense as a virtual page view.