Service Desk agents are often Swarming already, even if their leadership doesn’t know

Jon Stevens-Hall
4 min readJul 28, 2020

The UK’s Times Newspaper recently broke a story (paywall) alleging personal data breaches by members of the country’s new and hurriedly-assembled COVID-19 contact tracing workforce. This is obviously a significant story, but one particular line caught my eye:

“Thousands of (contact tracing) workers have joined social media groups that act as informal support networks attracting dozens of posts each day requesting help with IT problems and advice on how to handle cases”

It is entirely understandable that the focus of the article is on privacy, but this sentence reveals another interesting angle. This is an example of a common phenomenon: support professionals often feel the need to turn to non-official tools and channels, to get their job done.

In my role as a Product Manager, I’m involved in the creation of software products for Service Desks and the people working with them. In doing this, I have conducted many customer observation visits to service organisations and their frontline contact teams. I love those days: it is fascinating and educational to sit with support agents as they work. One key thing I have learned is that the use of side-channels for mutual support is a norm, rather than an exception.

Typically, such side-channels are based on conversational tools such as Slack, Microsoft Teams, or Skype (although I did once encounter an entirely unofficial knowledge management database, built using shared email folders, which was nicknamed the “Joeledge Base” after Joe, its creator and curator).

A typical example: at one service provider in Scandinavia, I discovered that support desk agents had devised a practice of broadcasting “customer plus problem statement” messages to a group Skype conversation, simply to identify if anyone else happened to be seeing the same issue that their customer had just reported to them. More frequently, though, what I’ve seen is less structured than this. Conversations simply spring up spontaneously, informally, with customers placed on-hold while agents seek out each others’ knowledge and insight.

The frequency with which I have encountered these side channels has long left me concluding that there is something of a systemic problem with the management orthodoxy of frontline service.

Organisations following such orthodoxy typically attempt to achieve an acceptable level of service through processes of top-down managerial control . A support centre may employ a diverse range of management inputs, from hiring policy, through team structures such as tiered support levels, to assistive tooling, and reactive data analytics. Agents themselves are not commonly granted significant autonomy, nor are their individual talents allowed to emerge. Success is assumed to result from managers doing a good job of turning the dials on those inputs (this is a topic covered extensively and superbly in Peter Johnson’s excellent “Making Light Work”).

However, these problem with these processes is that they are limited to the foreseeable and the scalable. In reality, it is not easy to implement them successfully in complicated situations. There is too much breadth to cover, and too many novel scenarios. In a pandemic response, training time is limited, but human factors are highly significant, perhaps even more so than in a typical customer service scenario. These agents have to overcome both technical challenges and the difficulty of communicating vague and shifting policies to confused and worried people. Both on a technical and a personal level, agents need to get their job done, and where their training and enablement falls short, it is not remotely surprising to learn that many have sought to find their own ways to do it.

In any given service environment these side channels differentiate themselves from the core processes. They are more collaborative, emergent, self-organised, dynamic, and cross-functional, and are driven by the participants’ confidence that this is the best route to an answer. In this respect, I would argue that they could be defined as a kind of Swarming.

Many support organisations would operate more effectively, and their customers would be better served, if they more formally embraced and utilised these behaviours. As with any type of collaborative “Swarming”, the enablement of dynamic, self-organising interactions between skilled people can be a catalyst for solving issues faster, without forcing them through structure of tiers, queues and silos before the right skills are reached.

I previously wrote about one interesting example of conversational swarming at a service centre, which had set up contextual channels into which frontline agents could dip at will while dealing with incoming issues. More generally, we have been investing in ChatOps capabilities in our own ITSM product line, as we see increasing demand for conversational interactions with tooling (something already well established in the DevOps community).

The uncontrolled distribution of personal data is of course a serious issue, and I do not seek to dismiss the significance of that part of the story. But what is easily overlooked is that the COVID-19 contact tracing agents, placed abruptly into a difficult new role of national significance, have been left to create their own conversations to get things done.



Jon Stevens-Hall

The intersection of digital transformation, DevOps, and ITSM. Articles by a senior Product Manager in the enterprise service management space. Personal views.