Development of customer-specific and cross-channel KPIs

    Development of customer-specific and cross-channel KPIs

    Last year, CURE collaborated in numerous projects focusing on the design of performance dashboards for campaign analyses as well as on the development of specific key performance indicators (KPIs) and their inclusion in the reporting. The required data basis was gained from onboard analytics tools of the various social networks (e.g. Twitter Analytics, Facebook Insights and YouTube Analytics) as well as from metrics or predefined KPIs of the employed monitoring and engagement tools. For our team, this involved the special challenge of consolidating various systems and processes in order to convert big data into smart data.


    Social networks offer increasingly effective integrated solutions for target-oriented analysis, comprehensive reports, and complex KPIs for measuring the performance of your channels. However, the horizon of these benefits is always limited to a single network. But how can you catch a glimpse beyond the horizon, jointly analysing and comparing your own channels in diverse social networks? Apart from highly specialised metrics and KPIs, every social network offers metrics and KPIs that also exist in other networks, but whose designations might be different. For example, I might say “share” and mean “retweet”. To avoid divergences when generating reports, all metrics must be checked for identical content value when manually compiling the data. Another difficulty is that the calculation of seemingly identical KPIs might actually be based on different algorithms and metrics whose logic usually remains hidden at the backend and is rarely disclosed. A third variable is the enormous amount of time required for the manual analysis of separate social network analytics and for consolidating them.

    Here an example of the different ways how a KPI might be calculated

    The interaction rate:
    The interaction rate is a good example of how differently KPIs can be calculated and interpreted. This figure can be calculated by means of various formulas and under consideration of various parameters. For example, the metrics recorded as interaction or their weighting may differ. Is a share worth more than a like? Do follower numbers and time periods play a role, or do you merely need information on the interaction per post? Should your own comments be excluded from the post, and do you only evaluate those with subject-related created content? Thus, the detail level of the computed KPI will ultimately depend on the extent to which the underlying formula and the selected parameters fill your need for information.


    The use of monitoring or engagement tools is one way of catching a glimpse beyond the horizon and using consolidated analysis and reporting possibilities across various social media channels. However, these tools have some minor weaknesses and differences in terms of the data quantity, topicality, and quality.

    The data consistency greatly depends on numerous factors. For example, consider the following:

    • Does a social network generally offer an open interface?
    • Which data can be retrieved via the API, and in what quality? (Service levels)
    • How smoothly does the tool connect to the respective API? (Technical connection)
    • What is the computing performance of the tool provider? (Number of servers)
    • How detailed is the crawler configuration? (Scope, interval, frequency)
    • How good are the algorithms of the respective tool? (Which KPIs are offered?)
    • How comprehensible and useful are their APIs or GUIs? (Interfaces for processing the data or web-based user frontends)

    Practical example:
    When the interaction (clicks, likes, retweets, replies) of a tweet surges within a short period, you can track this “live” in your own timeline or by refreshing the tweet. Current data on this can also be gained from Twitter Analytics. If, however, a monitoring tool is used, the displayed interaction may vary due to the factors described above. Every crawling process costs computing performance or may be limited by the respective social network due to a payable service level. This, in turn, reduces the frequency of the data synchronisation. Therefore, the interaction presented in the monitoring tool may diverge significantly from the actual figure.


    There are various KPIs that are used across different channels and tools. But are these KPIs truly relevant to the solution or to the measurement of your targets? To be able to select suitable KPIs or perhaps even to develop new ones, it is vital to fully understand the subject matter and the question that is being investigated. Preconfigured KPIs should therefore be carefully examined, broken down, and scrutinised for their metrics. This is necessary to ensure that assumptions and decisions are based on precise KPIs. The same applies to the development of custom KPIs. Here too it is of great importance to unify the right metrics in a meaningful formula. As a general rule, if a figure does not deliver any target-oriented insight, it is not a KPI.

    Leave a Reply

    20 + 18 =

    Close Menu