A Case Study and Methodology For Designing Open Collaboration | Management Innovation eXchange
Tuesday, May 31, 2011 at 04:31PM
Ralph Kerle in Collaboration, Innovation, Innovation, KM, knowledge_management, opensource, swarming
Related Posts Plugin for WordPress, Blogger...


This is the story of how we used open source principles at Red Hat to develop a new workflow for our technical support staff based on collaboration. There are two key parts to the story: discussing the process we used to develop our new collaborative work flow, and an overview of that new collaborative work flow itself.

Red Hat's Global Support Services (GSS) organization is accountable for customer loyalty. In that capacity, we're in the business of solving complex technical problems experienced by our customers. Our technical support engineers, technical account managers, and software maintenance engineers work from 16 countries throughout the world, providing 24 by 7 service in nine languages. We support a wide variety of enterprise software solutions, from our flagship Red Hat Enterprise Linux, to our JBoss Enterprise Middleware Stack, Red Hat Enterprise Virtualization and newer Cloud Solutions. Many of the questions we receive from customers are consultative in nature. That is, beyond helping customers with straightforward questions about our products, we advise customers on intricate technical architectures, and help them solve complex problems. To that end, it is critical that we effectively leverage the knowledge of our talented staff of Red Hat certified technical associates in a scalable fashion.

Until we implemented the changes described in this story, Red Hat used a traditional tiered support model (see diagram below). A level 1 associate would work on a case until they could no longer make progress, and then escalate to level 2. Level 2 would do the same, and escalate to level 3 as needed. Each escalation had some overhead of an escalation process, and each involved assigning the issue to a specific engineer who was more senior than the previous one (this was a manual process). While this model served most of the industry well for many years, it has some drawbacks:

  • Individuals are motivated personal metrics of success – how many issues can they close in a week? This can lead to behaviors that might not be productive. For example, associates closing cases prematurely to improve their numbers, or associates not helping each other for fear of their own numbers slipping as a result.

  • The culture that forms in this model is one of knowledge hording. People see that the more senior engineers have very specialized knowledge, and that gives them personal respect as well as positional power. The more power their knowledge gives them, the less incentive they have to share. This contributes to cultural barriers forming between the different levels.

  • Support cases may take more time than necessary to reach the right resource. We've seen examples where the escalation process for a case might have taken two weeks. As soon as the level 3 person saw the case, they solved it instantly. In other words, we had the ability to solve the case instantly, but that ability was locked until a case went through the process.

The old model:

Old Escalation Model


With Red Hat's rapid growth year after year, we knew we couldn't continue to simply add technical support staff linear to revenue increase. In order to scale effectively, we had to innovate around how our associates worked. So, in February of 2009 the management team of Red Hat GSS began a journey to determine if we could serve our customers better, and ultimately increase customer loyalty, by improving the ability of our highly trained, globally diverse technical support associates to collaborate with each other. The goal for our collaboration project: get the right people looking at the right issues faster, so that customer problems are resolved more quickly.

The fundamental idea was that if we could take away certain structural, cultural and procedural barriers that separated different groups of associates, we could increase the flow of knowledge, reduce the duplication of efforts, and ultimately provide customers with more accurate, more consistent, and faster results.

In addition, we wanted to design a system where the goals (both individual and organizational) were based on outcomes rather than activities. More on this in the “Benefits and Metrics” section below.

Key Innovations & Timeline

During the 18-month period of developing this new program, there were a few key innovations both in the process design as well as the design of the new workflow of our associates. These are described in detail under “Challenges and Solutions”.

  • Process design (how we developed the new workflow)

    • Use open source development principles in the process design (do it transparently, allow anyone to contribute ideas, try things out quickly and fix them as you go).

    • Doers as designers: make managers advisers and recruit a team of individual contributors to design the actual program

    • Global participation: ensure design team has the perspectives of every team in every region

    • Use elements of design thinking. Specifically: define the problem from the perspective of the doers, think about root causes of the problem, brainstorm possible solutions, and rapidly prototype different solutions until the right one is found (gathering and discussing feedback as we go).

  • The new workflow for our support associates (what we actually designed)

    • Base the workflow on principles of community and transparency true to our open source principles

    • Understand that we're not merely talking about a change in workflow, we're talking about a culture shift away from individuals striving for personal indicators of success to everyone working toward team indicators of success.

    • It is important to define collaboration and socialize internally what it means in our context and why it will make us both as an organization and as individuals more successful 

Challenges & Solutions

Deciding to increase collaboration is one thing, but actually enabling increased collaboration and then fostering its use by associates is no easy task. While our associates have always freely collaborated within small groups (people who sit near each other, for example, or people who work on a similar technology), our vision was to take advantage of our organization's collective knowledge by fostering collaboration on a global scale. That meant connecting people from several different functional roles, across 16 countries, through nine languages, and dozens of areas of domain knowledge, who had traditionally used separate, disconnected knowledge and case management systems.

Finding the right design process

Figuring out how to design a workflow that could accommodate this ambitious goal proved quite difficult. We assembled a project team and went through several iterations without feeling like we had something solid enough to fully implement. We started with a few pilot groups collaborating on a small scale, but found logistical as well as fundamental attitudinal problems stymied the open collaboration we envisioned. While we had successfully articulated the goals of our efforts, we were not making progress on a solution.

Then, after several months of false starts, we finally had a breakthrough. What if we got rid of the project team, which consisted mostly of managers from different areas of the organization, and instead asked the associates themselves to get together and figure out the best way to achieve the goals? This would, after all, be consistent with open source principles – enable people to work together to develop a solution that makes their lives easier.

The existing project team became what we called the “core team”. This team consisted only of managers, and their role would change from actually designing the solution to being more of an advisory board for the project. We then solicited volunteers for a new team, which we called the “design team”. This team was composed of about 15 technical associates; a representative from every region, every team, and every skill level. These associates were all enthusiastic and energized by the project – but that didn't mean they all thought it was a good idea or agreed on how to implement it.

My role was to facilitate the design team meetings (using design thinking techniques), make sure the conversation remained productive and on track, and to make sure all voices were heard.

The design process wasn’t easy. We alternated meetings between early morning and late evening to accommodate various time zones. Some associates were joining our calls at 5:00 AM, 2:00 AM, 11:00 PM or other odd hours in their local timezone. Emotions ran strong; some of the discussions questioned the validity of certain roles and opened the door of possibility to major change. People feared of losing their relevance, of sacrificing their exclusive status as experts of a particular topic, of having to take on more work than they previously had, of having to spend more time discussing problems with team members instead of just doing the work themselves. The implicit understanding that many people in technical fields have is that knowledge is power. Doesn't increasing collaboration and knowledge sharing mean that you give up that power?

The Solution: A new collaborative workflow

The design process taught us that breaking down silos and improving collaboration would be a major culture change and require a comprehensive change management effort that could potentially take years to be fully realized. Far beyond the scope of the original vision, we spent the next twelve months designing and revising what would become the new workflow for all our technical associates worldwide, along with new unified knowledge and case management systems, based on what I now refer to as “intelligent swarming”. Along the way the design team worked through dozens of questions, tough conversations, heated debates and disagreements.

For example, one key realization that we came to was that not everybody had the same understanding of what it meant to “collaborate”. Does it mean that you work on a problem until you reach the end of your ability, and then ask somebody else to work on the problem? Or, does it mean you work on the problem and ask for help as you need it, but maintain ownership of the case until it closes? Or, should we strive for people to proactively contribute to cases where they can help, based on monitoring cases appropriate to their skills, regardless of of whether they are asked to help? The answer would entail very different workflow guidelines.

During one design session early on, I asked everyone on the team to each draw a picture of how they thought the workflow should look in an effort to move the conversation forward. Many different diagrams were shared, several of which kept the principles from the existing escalation model, but renamed the terms. That is, people still envisioned putting a case through some kind of process by which a more senior person was tapped to help, but instead of calling it “escalation” they called it “collaboration”. This was revealing, and became a theme. Understanding the concept of collaboration at an intellectual level is one thing, but actually internalizing it quite another.

Then one diagram came in from a frontline support associate in Beijing that looked truly original:

New Swarming Workflow

Rather than envisioning a traditional phased workflow, this associate wanted to visualize the different states a solution could be in during its life cycle based on the concepts we had been discussing. This workflow diagram has been updated and tweaked several times since it was first created, but it largely resembles the original and has served as a simple but effective guidepost for the project. Some elements of the workflow are similar to the old one – for example a frontline (junior) associate is the “first touch” on a new case and the first thing they do is search the knowledge base to see if they are dealing with a known problem or a truly “new” problem. If the case owner can solve the case on their own, they do so. There is no reason to collaborate for the sake of collaboration. Any issue that has been solved previously should be found in the knowledgebase, leaving only previously unsolved problems to the collaborative workflow.

If, however, the case owner needs help, they do not hand the issue off to someone else. Instead, they signify they need help by changing the status of the case to “waiting on collaboration”. This is where the new model really kicks in. All associates belong to different technical “domains” that focus on different topic areas. Associates of all skill levels view all the cases in the system through a filter for their domains. They review cases relevant to their skills, and can help guide each other toward resolution. More junior associates take more cases up front, while more senior associates spend more time contributing to cases owned by other people through the new collaboration model. Thus senior associates see more issues than they did before, which means their expertise is accessed more quickly, junior associates are sticking with cases longer, learning directly from the senior associates rather than simply handing issues off. Customers, ultimately, are getting the answers to their questions faster.

Benefits & Metrics

The workflow effectively solves its primary goal: get the “right” people working on the “right” cases faster, so that customer problems are resolved more quickly. The workflow has several additional benefits:

  • Knowledge sharing: associates who have less knowledge in any given area learn directly from the senior associates. At the same time senior associates are recognized for sharing their knowledge rather than hording it.

  • Removing barriers: the divisions that separated the tiers of support blur as people collaborate more, resulting in a broader sense of “team” and aligning everyone around the goal of solving customer problems effectively and efficiently

  • Cross-geo collaboration. We're also seeing a lot more direct collaboration between associates in different parts of the world, further removing barriers between teams that tend to exist from office to office. One global workflow forces discussion and resolution of snags related to how things work in different regions


Early data indicate that the combination of Knowledge Centered Support (a methodology for incorporating knowledge creation and reuse into the workflow of solving customer problems) and our new intelligent swarming model allows us to handle greater call volumes without increasing staff. We also see shorter time to resolution of support cases, which leads to increased customer satisfaction. These efficiency gains are early indicators of a positive ROI and are a direct result of empowering our associates to take the lead in designing their own processes and procedures, the open-source way.

In the coming months our goal is to move from adoption of this new model to proficiency, working out the kinks and refining the metrics that we use to measure success. One thing we we no longer do is make goals around specific activities. An important realization we came to as we changed our workflow was that setting any kind of goal on an activity invariably results in undesired behavior. It's important that we strive for goals that reflect outcomes and also take into account the team success over the individuals success. For example, its tempting to set a goal such as “Each associate will contribute to X number of cases per week”, thinking that this would spur people to collaborate on cases. But, what you might find is that instead of truly collaborating associates simply leave “low value” comments (an activity) on large numbers of cases to meet the goal, without really helping to solve any specific cases. A better goal would be something like “The technical domain the associate is a part of will reduce time to closure on new cases by X percentage points”. The outcome here is that the team actually solves cases, rather than perform a specific activity.

The Big Picture

Below is a diagram that shows how swarming (called SBR in this diagram), combined with Knowledge Centered Support (KCS) enables us to leverage collaboration for new cases, while re-using knowledge from previously solved problems to provide answers to customers more quickly. We're also looking at ways to improve our knowledge of the whole customer by better engaging the entire team that interacts with a customer (called the virtual account team in this diagram).

Solve Once for Many


Embarking on this journey ultimately taught us lessons that we didn't expect – not just in how to manage collaboration for our associates' day to day jobs, but in how to use open collaboration as a management team to innovate on process design in our organization. It turns out the answer was right in front of us all along – use the open source development concepts our company was built on.

Our new collaborative workflow is far from perfect. There are still tough questions, problems we encounter, and improvements we recognize we need to make. Over the coming months we will focus on measuring the effectiveness of our new intelligent swarming model and increasing its efficiency. To do this, we'll keep building on the lessons we've learned in developing the program: resist the temptation to apply changes or improvements from the top-down, and let the people who do the work design and improve the process, transparently.

This story was originally published on opensource.com in a shorter version: http://opensource.com/business/11/5/designing-open-collaboration-red-hat-global-support-services
More information about Knowledge Centered Support can be found via the Consortium for Service Innovation: http://www.serviceinnovation.org/knowledge_centered_support/
For further information on the Management Innovation Exchange.
Article originally appeared on The Creative Leadership Forum - Collaborate - Create - Commercialise & Transformational Change (http://thecreativeleadershipforum.com/).
See website for complete article licensing information.