Understand the Changes to the Customer Journey Complexity Score
Previous year, we released a feature that allows customers to score the complexity of their journey models. Now (with the release of January 17, 2023), we have updated the scoring approach, based on customer conversations (user research) as well as data-driven insights (comprehensive process model data analysis). This blog post provides a walk-through of the revised scoring approach. Key novelties are:
- Separation of journey model dimensions and the operational complexity of the underlying processes
- Nuanced consideration of nested (sub)-processes by recursion;
- Revised thresholds and weights based on process model data analysis;
- ‘T-shirt sizing’ of complexity for an intuitive assessment as ‘low’, ‘medium’, and ‘high’ complexity.
Organizational Complexity in Business Process Management
In the context of business process management, complexity can be defined as the “non routineness, difficulty, uncertainty, and interdependence […] associated with [organizational] activities” . Removing “difficulty” – a hard-to-define property – and merging “non routineness” and “uncertainty” simplifies the definition and gives us journey complexity and process complexity:
- Journey complexity provides a measure of levels of uncertainty, and interdependence of a journey through an organization.
- Process complexity provides a measure of levels of uncertainty, and interdependence of a business process.
Because a journey typically involves several processes, journey complexity takes into consideration (i.e., aggregates) the complexity scores of these processes. For journey and process complexity, we have developed model-based scoring algorithms, whose behavior we describe below.
Journey Complexity versus Journey Model Dimensions
In the new version of customer journey complexity score, we distinguish between journey complexity and journey model dimensions:
- Journey complexity considers the journey’s operational complexity, i.e., the complexity of the underlying processes that are directly or indirectly linked (via value chain diagrams, processes, or other journeys).
Journey complexity in SAP Signavio Journey Modeler.
- Journey model dimensions provide an at-a-glance overview of the size of the journey table (grid size), as well as of the number of non-empty fields. Detailed counts, e.g., of the number of linked personas and images are provided as well.
Journey model dimensions in SAP Signavio Journey Modeler.
Elements of Process Complexity
The complexity score of a process model (i.e., of a BPMN diagram) is determined based on the following elements.
- Flow: How many decision and parallelism splits are in a business process and how deeply are they nested?
While there are two perspectives on flow complexity, i.e., decision and parallelism, these perspectives are intertwined, for example when a “decision gateway” (any gateway that is not a parallel/AND gateway) is nested into a parallel gateway. A decision implies uncertainty, parallelism implies interdependence. A parallel gateway has a base complexity of 1, a decision gateway has a base complexity of 1.5. For each gateway, base complexity is multiplied by its nesting level. Then, all scores are summed up to a pre-normalized score. This score is then normalized, on a scale from 0 to 1, where a pre-normalized score of 0 indicates minimal complexity, and a pre-normalized score of 20 or higher indicates maximal complexity (1). For instance, the process depicted by the figure below contains two nested XOR splits, and hence has a pre-normalized score of 4.5 (1.5 + 2 * 1.5) and a final flow complexity score of 0.225.
- Handovers: How many handovers between roles are in the process?
More roles/actors in a process imply more interdependence and complexity. For each handover to a “new” role we add 1.5 to the handover base score. For each handover back to a role that has already been involved in the process, we add 1 to the handover base score. The score is then normalized, on a scale from 0 to 1, where a base score of 1.5 or lower indicates minimal complexity (0), and a base score of 10 or higher indicates maximal complexity (1). For instance, our example process has three handovers, which amounts to a handover base score of 4.5 (3 * 1.5), which is then normalized to 0.353.
- IT systems: How many IT Systems are there in a process, and to what extent are they accessed by multiple roles?
Each IT system gets an initial base score of one, when it is first utilized via a task, and each additional utilization adds 1.5 to the base score. For all IT systems, the base scores are summed up to a total IT system complexity base score. The score is then normalized, on a scale from 0 to 10, where a total base score of 1.5 or lower indicates minimal complexity (0), and a total base score of 10 or higher indicates maximal complexity (1). For instance, our example process has two IT systems that are used by two tasks and three tasks, respectively. This amounts to an IT system complexity base score of 6 (2 * 1.5 * 2), which is then normalized to 0.6.
- Documents and data objects: How many data objects (incl. documents, i.e., we use ‘data object’ as an umbrella term for BPMN data objects and documents) are in a process, and to what extent are these data objects accessed by multiple roles?
Data object complexity is computed in the same way as IT system complexity. Each data object gets an initial base score of one, when it is first utilized via a task, and each additional utilization adds 1.5 to the base score. For all data objects, the base scores are summed up to a total data object complexity base score. The score is then normalized, on a scale from 0 to 1, where a total base score of 1.5 or lower indicates minimal complexity (0), and a total base score of 10 or higher indicates maximal complexity (1). For instance, our example process has two data objects, each of which are used by two roles. This amounts to a data object complexity base score of 6 (2 * 1.5 * 2), which is then normalized to 0.6.
- Linked processes: How many other processes are linked via link events to the process model?
For the sake of simplicity, each linked sub-process adds 3 to the linked process complexity base score, and each process that is linked via an event adds 1. The score is then normalized, on a scale from 0 to 1, where a base score of 2 or lower indicates minimal complexity (0), and a base score of 10 or higher indicates maximal complexity (1). Because this process does not have any process that is linked, its linked process complexity score is 0. Note: sub-processes are managed via recursion, i.e., instead of a fixed complexity addition, the entire sub-process is scored in detail, until up to four levels of recursion. If the nesting-level of the underlying process landscape is too deep, a warning will be displayed, and an approximate score (lower bound) is provided.
Process-Level and Journey-Level Aggregation
To determine the final complexity score of a process, all sub-scores are summed up, weighted, and then averaged. Here, the weights are as follows:
- Flow and handover complexity: 35% each;
- IT system, data object, and linked process complexity: 10% each.
On journey level, the complexity of each linked process is multiplied by 0.2 and finally, the entire score is computed by multiplying by 100 – based on our assessment, a complexity score of more than 100 will only be achieved in very rare cases.
To provide an intuitive assessment of journey complexity, the complexity score is mapped to a ‘T-shirt size’, ranging from very low to very high, according to the following mapping, which has been informed by systematic estimates based on the complexity scores of thousands of process models, (x is score value):
- Low: 0 ≤ 𝑥 ≤ 20
- Medium: 20 < 𝑥 < 60
- High: 60 ≤ 𝑥
In summary, the update to complexity score brings the following changes and benefits:
- Distinguishing between the underlying operational complexity that comes from linked processes and the dimensions of the journey model makes the scores easier to interpret and more actionable.
- Considering the scores of entailed (sub)-processes by recursion and refining scores and thresholds based on real-world data provides a more accurate complexity assessment.
- Providing ‘T-shirt sizes from low to high complexity facilitates an intuitive assessment.
 Jahangir Karim, Toni M. Somers & Anol Bhattacherjee (2007) The Impact of ERP Implementation on Business Process Outcomes: A Factor-Based Study, Journal of Management Information Systems, 24:1, 101-134, DOI: 10.2753/MIS0742-1222240103