The previous post in this performance measurement and optimisation series involved mapping our identified business objectives to relevant metrics that can be measured and eventually optimised to improve online performance. As mentioned towards the end of the aforementioned post, not all visitors are equal, and it is of critical importance that the online audience is split into sensible segments.
Continuing with the same example, one of our hypothetical business objectives is to double revenue over the next two years. In the last post, we mapped out a list of metrics which relate to or influence our overall success against this objective. This consisted of a relatively large group of supporting metrics affecting two primary success metrics: cost and revenue.
Although the metrics map is intended to be quite generic (i.e. it should apply loosely to all segments), the actual ongoing analysis of performance would involve isolating different audience segments for individual attention. This provides greater context and allows us to compare the behaviour of different sources of traffic and different types of users. The reasons for users converting (or not) will differ depending on the campaign they came to the site through, their current stage in the buying cycle, their geographical location, whether they are a new or return visitor, etc. Looking at the overall conversion rate, for example, will include data for visits that bounce (visitors who potentially landed on the wrong site for their needs), visits from countries/cities not served by the business, visitors from mobile devices on a website that is non-transactional for mobile visitors and so on.
The exact segmentation required will differ for each business and industry. However the different segments should always be grouped in a sensible way, categorised by the needs of the department/personnel that will be looking at each segment. For example:
These segments will be most useful to the IT, design and UX teams as they all relate to the different technologies used to access the website. In terms of ensuring the user experience is optimised for the most common devices, browsers and screen sizes, it is essential that the analytics data can be segmented by these dimensions.
Examples of technology-related dimensions to segment by include:
- Browser and version
- Screen resolution
- Operating system
- Mobile device and operating system
These segments will be most useful to the marketing team as they are all related to inbound traffic sources for the website.
Examples of marketing channels (or traffic source) and their secondary dimensions to segment by include:
- Natural search
- Brand/non-brand keywords
- Generic/long-tail keywords
- Paid search
- Brand/non-brand keywords
- Generic/long-tail keywords
- Ad group
- Keyword/matched search query (i.e. bid term versus actual search query)
- Individual publisher
- Type of link (e.g. banner, email)
- Marketing list
- Time of day/day of week
- Ad type (e.g. MPU, leaderboard, skyscraper, etc.)
- Direct traffic
This segmentation category would be used to cover the different types of visitors that visit the website in question. Irrespective of the channel or device they visited the site through, this would relate purely to their visitor status and behaviour on the site.
Examples of segments that could be used based on user behaviour include:
- New/returning visits
- First time/repeat customers
- Converting /non-converting visits (macro and micro conversions)
- Non-bounce visits
This category would encompass any user-specific data and is likely to be less straightforward than the others depending on the choice of platform to collect and analyse the data.
Examples of user-specific dimensions for segmentation include:
- Geographical location
- Relationship status*
*These dimensions are typically provided by user input and consent, for example user registration, surveys, feedback, etc.
Many of the segmentation options detailed above will be possible by default within most analytics platforms. Some of the more specific segments may involve custom tracking which would need to be included during the implementation phase of the analytics/measurement solution. It is also important to note that certain segmentations above will not be possible via an analytics solution but may also rely on CRM data, social platform data, email database data, etc. Personally identifiable information cannot be stored in certain analytics platforms (e.g. Google Analytics) and privacy policies should be strictly adhered to when planning any analytics/measurement project.
As with the previous stage in this process – mapping objectives to metrics – it is advisable to ensure you are tracking in as much detail as possible for all scenarios. It is better to start with too much data and refine with dashboards and/or custom reports as opposed to realising six months after an extensive analytics implementation project that you do not have the segmentation options required.
At this point, we have a very clear idea of the key business objectives and the macro and micro conversion points on the site. We have also mapped the objectives to a large group of relevant metrics and have now identified the different audience segments required. This collective information forms the core part of the measurement brief and can naturally lead us to the next stage of the journey: measurement framework, implementation and QA.