How to measure the ROI of personalisationby
The rise of personalisation has been discussed far and wide. As technologies, agencies and clients become increasingly sophisticated in their attempts to deliver better digital solutions, the ability to create personalised experiences for users whilst obtaining better insights into their habits has become more important. Greater insights deliver greater return on investment.
From the 1990s, research from various parties (e.g. Broadvision) looked into different means of enhancing a user’s web experience. Since then several factors have contributed to the rise of personalisation, both in its effectiveness and also correspondingly in its popularity:
- Fall in the cost of deployment.
- Personalisation tools used in many mid-level content management systems to gain a competitive edge over rivals.
- More content, so corresponding need to guide users to the right content.
What is personalisation?
Put simply, there is a fundamental difference between non-personalised and personalised sites: the former present the same content to all users regardless of profile, preferences and clicking behaviour. This results in a “static” experience, relying on navigation and search as the main points of interaction. In contrast, a personalised site is one where the experience is changed to be different to each user based on a number of pre-defined factors.
Over the following few paragraphs we will see how personalisation carries many benefits for both user and administrator; that measurement of personalised sites is getting more complex, more sophisticated and also more insightful; and how the convergence of several factors means that not only is personalisation here to stay, but also that the benefits of personalisation from a business perspective will only increase correspondingly.
Explicit vs implicit personalisation
In order to ascertain the benefits of personalisation, it might be useful to provide a broader overview of the different types. Very broadly speaking (a deeper analysis is out of the scope of this article), personalisation can be “explicit” or “implicit”.
Explicit personalisation is based on the user’s profile, and can be passive (allowing users to make choices as to what they’d like to see) or segmented (allocating content choices based on a logic layer applied to a particular profile/profile type).
Implicit personalisation is, to me, a far more interesting proposition. Here the clicking activity of a user on that website is tracked and monitored, and every incremental click then used to determine what content the user then receives. There are also hybrid systems that claim to capture the best of both worlds.
The future of personalisation probably lies in advances made in the field of adaptive personalisation. Here, the system creates the logic that determines the content that each user receives. Over time, each website users’ behaviour is analysed and the system then processes the incremental behaviour of users to create a model around that user or user type, and consequently creates the content based upon that model’s predictions.
From a cursory glance of the above different personalisation types, it’s probably clear what the benefits of personalisation are. From an business perspective increasing personalisation is a win-win scenario, as deeper insights into your site users necessarily follow through to create personalised browsing experiences that reward users in terms of not only getting the right content in front of their eyes, but also accordingly creating a more pleasing user experience. And, particularly with adaptive and implicit personalisation there is an innate systematic urge to continuously improve in order to keep systems accurate, with the resultant benefit of ever-better analytics and insights.
Your website is a business tool. Personalisation brings market and business intelligence to your organisation, while at the same time rewarding your users. Adoption is therefore a no-brainer.
In attempting to gauge how best to measure the effectiveness of personalisation, we should start with some higher-level principles from the world of analytics. We need to bear in mind both quantitative measures of success - the “what” - along with the qualitative factors - the “why”. Each on its own is a necessary, but not a sufficient, condition. They must always be taken together.
The world of analytics has progressed greatly in recent years, from the early and painstaking days of log files and tagging to the near-ubiquity of Google Analytics and the remarkable levels of sophistication and granularity that such systems offer to everyday users. It is no longer the realm of specialists and experts. Ultimately measurement is about improving levels of service; this is best achieved by creating a better experience for users.
Context is everything. No metrics in themselves are independently meaningful, particularly when anything at all can be a source of data. This is a double-edged sword as obviously there is a need to collate disparate data sources, but the upside is that dedicated web managers have the freedom to be creative not only at the back-end but also correspondingly in terms of the relevant touch points that impact upon the user.
From a broader perspective, this scenario has been bolstered by several key industry trends. The rise of big data means that there are better tools to accurately filter large datasets, and obviously there are also more data sources. And with more data sources, there are more platforms - notably the rise of “mobile” as a phenomenon. Finally, the ubiquity of social media is an important part of this mix, with key metrics, sentiment analysis and broader “buzz” analysis considered crucially important, particularly in the B2C sector.
Necessity engenders invention, and greater complexity of the kind described above has resulted in a set of analytical tools that allow us to understand what is happening on our sites in a much more sophisticated and granular fashion. The advent of “universal analytics” allows users to tailor their dashboards as needed, integrating relevant and pertinent datasets and obtaining a much better overview of (for example) an entire marketing funnel or even a wider ecosystem of funnels. On a whole number of different points we can now access greater insight: previously fiddly KPIs such as importing cost data; customer lifecycle/lifetime value; recency, frequency and monetary value; custom widened dimensions for bespoke measurement and (perhaps most importantly) a better view of mobile data.
The meaningful measurement of personalisation is a fluid and dynamic state of affairs. Despite this it is possible to formulate some higher-level conclusions. Firstly, it’s clear that there are increasingly sophisticated approaches to web personalisation and web measurement. This arises from the fact that we are now faced by the myriad challenges of providing better user experiences across more devices, different screens and with vastly more content and data. The upside of this is that we can, if we are thorough and sufficiently methodological in our approach, derive greater understanding and insights from our uses and their preferences. Not only can we therefore maximise return on investment, but a happy ancillary effect of this is that we should, in most cases, correspondingly deliver a far better proposition.
It could be argued that we are in a “perfect storm” for personalisation. Our devices are becoming increasingly smarter - they are getting better at monitoring us. The rise of the “quantified self” movement means that the sharing of personal data will only become more ubiquitous - look at the ‘Health” app on Apple’s recently released iOS8 operating system, and the rise of wearable tech everywhere. At the same time, data storage is becoming cheaper along with the related data processing and analysis services providers and tools. And, across the board, there is more expertise. The future will entail better systems, greater and more fulfilling user experiences and even more insights which are easier to access. Everyone wins.