Elise Olding and Carol Rozwell kicked off the afternoon with a session on the hidden costs of unstructured processes: although a lot of focus of BPM efforts (time and money) is on structured processes, as much as 60% of an organization’s processes are unstructured – and probably also unmonitored, unmanaged, unknown and unruly.
Gartner defines unstructured processes as “work activities that are complex, nonroutine processes, predominantly executed by an individual or group highly dependent on the interpretation and judgment of the humans doing the work for their successful completion”, and notes that most business processes are made up of both structured and unstructured processes. Unstructured processes are costing organizations a lot of money in lost productivity, lack of compliance and other factors, and you can’t afford to ignore them. Although most processes aimed to meet regulatory requirements are structured, unstructured processes provide a company’s unique identity and often its competitive differentiation, as well as supporting operational activities.
In order to start managing unstructured processes, you need to get some visibility into them; start by understanding the critical path through the process. This can be a bit tricky, since as you start to map out your unstructured processes, there will be some points at which the process participant just has to wing it and make their own decisions. These are, after all, knowledge workers, and it’s not possible (or desirable) to map every possible process permutation. Instead, map the structured portions of the process, then the points at which it becomes unstructured, but don’t try to overengineer what happens in the unstructured parts. The unstructured parts can be modeled by the notification mechanism (how someone is notified that a piece of work requires attention), the information provided to the participant to allow them to complete the unstructured work, and how the outcome is recorded.
They presented a number of analysis techniques for getting to the heart of unstructured and folklore processes:
- Observe work being done, and challenge tasks that don’t make sense. Keep asking “why”.
- Use storytelling (“tell me what happens when…”) to uncover decision-making logic, methods and best practices: these types of narratives are not well-captured in standard process documentation.
- Analyze the unstructured interactions between people (e.g., customers and CSRs) and extract the themes and patterns. Rozwell wrote a report “Business Narratives Supplement Traditional Data Analysis” that discusses one technique for doing this, although it wasn’t quite clear what it was from the discussion.
- Get clarity around roles and who is the decision-maker in any given process.
There are a variety of different areas of knowledge that you need to consider when analyzing unstructured processes, from identifying what metadata is used for collaboration, to looking at alternative analysis techniques such as mind mapping and social network analysis. Understanding collaborative technologies is also key, since unstructured processes are often collaborative in nature, and make use of the participants’ social graphs.
Their final recommendations are to keep an eye on the technologies that can support unstructured processes, but not to go overboard on monitoring and managing these processes.
@sandy – I’m struggling to buy this. You don’t provide any commentary but from this distance it sounds like they’re urinating in the wind and over complicating something to fit into a box that doesn’t exactly have a reliable shape. Where are the proofs that: “60% of an organization’s processes are unstructured – and probably also unmonitored, unmanaged, unknown and unruly?” And what’s to say that BPM could do anything about it anyway? Isn’t that the nature of problems that people who need to make decisions are up against all the time? @sig talks about these as ‘barely repeatable processes’ – a good way to look at them – where you need a quickly built app that includes the process loops in order to solve the problem. Does that mean they need a formal BPM solution of the kind with which you’re familiar or is there another way? Appreciate your thoughts.
Business narrative is a technique that uses the collection and interpretation of stories told in the workplace. It provides an additional dimension of analysis to raw data such as survey results because it encourages people to tell stories about how they actually get work done. One process for collecting stories is through the use of “anecdote circles.”
The crossfunctional team in the example I described talked about their observations and impressions of the collected data as business narratives. They grouped similar observations, looked for common themes and emerged at the end of the exercise with new insights about the call center process.
from my experiences:
before to structure unstructered/complex things decide if it was useful or: feasable try to structure social, km, case management or cep process. Another reflexion : minimum of chaos is source of imporovment/adaptation/agility. cf frozen process vs hot process
some tips:
-follow an learning by doing/show me aproaches are more easy to build an first version of “main process” models: record each step/branch of the process steps graph
-you can discover and improve by considering exceptions case and simulations : whatif scenarios.
-Use process logs to build concrete facts as support for modelling decision elements
-follow more than one end to end process execution to consolidate bp
-consider an management to operationnal agent roundtrip to refine descriptions (multi scales/multi level approach)
–ask to all stakeholders processes to consolidate terms and rules.
-Try to use paperboard and animated models to validate models. rewind features are very useful.
-ask to Enterprise people to play bp game and comment theirs actions: Run more than one this game, and each time , changing each role performers to consolidate these models.
in final paraphrasing this sentence from Stebdhal, famous french writer : Business Aechitecture is an mirror that walk along the enterprise landscape…
kind regards
djebar
Dennis — the BPMS vendors are looking for ways to incorporate “barely repeatable processes” into their systems, allowing users to create their own ad hoc processes on the fly but still capturing the audit trail so that it’s not just happening over email or the phone in an unaudited fashion. The idea is not to pre-define all of these processes, but to provide tools that allow process participants to have a sufficiently unstructured environment to do what they need to do, and augment that process with their own call-out at that point.
Carol, thanks for the clarification, looks like an interesting technique.
Sandy,
Hi. These are exactly the kind of processes that we address here at ActionBase. One thing to be careful with is that you want to provide enough structure to the process to add value, but not so much as to strangle it. Given that most of these processes are executed today via documents and email, we built our tool as an extension to those standard office tools – allowing the same ad-hoc feel, but adding a layer of management, tracking and reporting.
For many of these processes an initial formal model is overkill (and at odds with the needs of most knowledge workers) – at most you want a guideline or best practice that gets modified as the work gets done. Then these emerging models can later be used to create a more formal model if needed (I’ve blogged on the topic of in-situ process discovery on our blog http://blog.actionbase.com/in-situ-process-discovery).
In general we’ve found that need to provide most BPMS’s with a completely defined formal process model is at odds with the management of unstructured processes. Unstructured processes support needs to be part lightweight process tool, part collaboration tool.
Of course, I am an advocate for our tool – but I think Google Wave could also be a game changer in this space – if Google decides to take it in that direction.
Hi Jacob, I definitely thought of your product when writing about this. I also agree that Google Wave could be a game-changer here: at SAP TechEd last week, I asked the group that had developed Gravity (process modeler inside Google Wave) if they intended to extend the Google Wave ideas to collaboration during process execution, although the answer was somewhat non-committal.