Localisation workflows: the impact of process well-handledness on automation

In the context of an ever growing need for LSP to optimize their offerings to their clients, TMS stands out as a seemingly obvious choice to deliver higher quality content faster and cheaper. As the trend is to deploy TMS workflows full speed ahead, do those systems cater well to different use-cases as they can be found in traditional localisation processes, involving frequent updates, change of scope at language or file level as localisation is tied in more closely than ever to content authoring? We started with the assumption that TMS only supported well-handled use-cases with fixed sets of well-defined transitions and limited support to non-well-handled cases with flexible unpredictable transitions. As a corollary, we deemed that perhaps certain use cases are not suitable for automation in TMS. Thus, the objective set forth with this paper was to evaluate the strengths and limitations of TMS using the concept of workflow well-handledness. To this avail, we began by looking at the current state of affairs in traditional localisation workflows and found that, although the traditional translation, editing and proofreading cycles and stakeholders follow a defined model, there is a need for added flexibility at each step to optimise turnaround times. We then positioned TMS in the supply chain, looked closely at localisation business processes and in what ways solutions based on TMS could improve current localisation workflows. It became apparent that TMS is not a monolithic system and workflow engineers need to closely work with project stakeholders and be aware of all process applications to design, implement and ensure relevant reporting. The practical steps to deploy a workflow in TMS around key process areas and specific goals were detailed, revealing that there should be just enough human tasks to cater for change in requirements and the system must be robust enough to allow time-critical maintenance on running production workflows. Workflow patterns were introduced allowing us to scrutinise this flow around complementary perspectives, that of control-flow, data, resource and exception handling. Finally, we put the concept of well-handledless to the test in a case-study of industrial Lionbridge TMS using workflow patterns. The case study confirmed that in control-flow patterns support is rather basic and limited to well-handled use-cases, with a fixed set of well-defined transitions and predictable resources. This is mainly because such systems focus on the end-user to meet SLAs rather than offering powerful underlying workflow.