Update to sessions in Google Analytics

Posted on 19. Aug, 2011 by in Digital Marketing, Thoughts

On Thursday 11th of August 2011, Google released an important update to Analytics via its blog . The blog post describes the change in how a session is defined within Google Analytics.

Until 11th August, sessions ended when over 30 minutes elapsed between page views for a single user, at the end of the day or when a user closed his/her browser. With the recent change, the first two scenarios remain the same but closing the browser will no longer mark the end of a session – rather, that will be when the campaign information for that particular user changes. This means that Google Analytics will consider any changes to utm_source, utm_medium, utm_term, utm_content, utm_id, utm_campaign and auto-tagging from AdWords (gclid) as an indication that a session has ended.


How will this impact data?

Google communicated that this change only applies to visits from the 11th August and that historical data will not change. Regardless of whether the change is seen as a positive or negative, if any significant changes are witnessed in the data, historical comparison and benchmarking becomes difficult.

After analysing an aggregated view of bigmouthmedia’s client data, it appears that visits have increased significantly more than the 1% that Google’s initial data research indicated. Any percentage change in visits will have a similar impact on the metrics defined by a visit, eg Bounce Rate, Conversion Rate, Average Time on Site, % New vs Returning etc.

By continuing a session when a user closes his/her browser for a short time, sessions aim to more accurately model a user’s engagement with the website. Coupling this change with the fact that a session now ends when a user’s “campaign information changes” could have implications for a large proportion of digital marketing clients and online marketers in general.  For example, if mid-session research behaviour is common place, the number of visits could increase as well as the channel or keyword that a resulting sale is attributed to.

Breaking this down to the simplest form, if a potential customer landed on the same site through 5 different PPC keywords within a 30 minute period, this previously equated to 5 clicks and 1 visit recorded within Google Analytics. From the 11th of August, this scenario now equates to 5 clicks and 5 visits.  However, if a converting customer transitioned through more than one channel or keyword within a 30minute period, the visit was previously attributed to the keyword and channel that initiated the 30 minute session but the sale was attributed to the last click channel…. From the 11th of August, each individual search that generates different campaign information (e.g. utm_source) generates a new session.  This change now assigns visits & sales more accurately and brings the rest of Google Analytics in line with its Multi-Channel Funnel reports.


As with any change, there is almost always going to be resistance, and judging by the number of negative comments online it has ruffled quite a few feathers. However, apart from the lack of ability to measure like for like historical performance, the change is also welcomed by many, especially those with an active involvement in Multi-Channel Funnels and Marketing Attribution in general.

Tags: ,

  • http://twitter.com/pulseltd Pulse

    Hi Nick
    thanks for this – its not great news for GA users I guess but every cloud/silverlinings etc…
    I’d be happy to talk to you about how http://www.wiredminds.com could overcome this kind of shortcoming and offer a compelling upgrade for your clients to using GA only.  Ive put out a message to your office to get in touch.

  • http://twitter.com/pulseltd Pulse

    sorry – that post was mis-attributed! shouldve come from a different twitter account! @wiredmindsuk:twitter