The Global Support Services (GSS) organisation at Sophos has a strategic knowledge management system whose primary goal is to deliver case deflection at the lowest possible operating costs. We are wary of adding new features that do not help us realise that goal. This approach gives a clear pointer to the kind of metrics we should track, filtering out many measurements that can in theory be made but that add little to our understanding of where to apply scarce resources.
As explained more fully in this blog post on Knowledge Management at Sophos, a multi-lingual knowledgebase system has been built, and integrated with online resources like product-focused websites and customer notification channels such as Advisory articles, targeted emails and Twitter. Working closely with all of these are two separate online communities (discussion forums).
SophosTalk (http://community.sophos.com) was launched in November 2009 and complements traditional phone and email Support channels for customers. It is structured around the Sophos product lines such as Endpoint, Gateway, Data Protection and Mobile Security. In addition, there are discussion boards for specialist topics such as the small business versions of our products, and issues relating to data loss protection. There are boards for notifications from us, and for comments and feedback from community members. Members who are registered with our Beta product launch programmes are granted access to restricted boards where they can ask questions about other users’ experience, suggest changes, and learn more about new features. Though not visible to standard community members, we’ve launched ten of these to date, and they’ve been very successful in driving Beta products to full FCS-standard.
SophosFreeTalk (http://openforum.sophos.com) was launched in November 2010, and is the sole Support channel for consumers of the Sophos free tools. It, too, provides boilerplate boards for notifications and feedback, as well as boards for the free tools themselves. Each tool is further supported by a separate board for help, tips ‘n’ tricks, and FAQs, these taking the place of the standard phone and email support, a policy that helps Sophos drive down the overall cost of providing free protection. A measure of SophosFreeTalk’s success is that there is virtually no overspill from the community to the standard support channels from users prepared to try phoning or emailing despite the embargo on doing so.
Both communities are readable by anyone, and it is therefore impossible to say how many anonymous readers are online at any one time. However, to post or answer a question, rate or tag content, or respond to a poll, users must be registered and signed in. It is therefore possible, in both communities, to gauge growth by tracking registrations and levels of content submission. I’ll be discussing some of our community metrics below, including a compelling indication of return on investment (ROI).
Community management is a tightly focused and economical affair. The two systems share a community administrator who builds new boards and generally looks after strategy and metrics. There is a tiny team (one, with some back-up) of community moderators (deleting the occasional spam, moving posts to the correct location, etc). She is supported by product-focused technical moderators who are domain experts on the look-out for serious issues such as dangerous (usually well-intentioned) advice, and by a hugely productive but small group of external VIP users, about which, more below. Beta program and special interest moderators step in as required.
So, what about metrics? Because it was added to the functions of the pre-existing GSS Knowledgebase Team, communities management at Sophos naturally inherited a focus on case deflection at low cost. The challenge was therefore to select only those metrics that would help us identify and encourage case deflection, while ensuring that the communities remain healthy. We derive benefits other than Support cost savings, such as the gathering of customer feature suggestions, the opportunity to push news to our customer base, and generally build online credibility. A community that is stagnating will eventually stop delivering all those returns on investment as well as case deflection.
An obvious measurement of community health is ongoing registrations. New membership means more questions asked, more answers posted, more quality feedback and more general comment on how we’re performing as a business. Here’s a headline metric, using data gathered from the day of launch (I’ll use examples taken from SophosTalk throughout this paper, as it’s the more mature community):
It’s important to understand that not all registered users deliver the same levels of contribution. Online community membership shows the classic Zipf distribution of contributions, where a tiny number of activists punch way above their weight (the “VIP” or “super-users”); a larger number of members individually stand out against the background, but not at the level of the VIP users; and a very large number of users individually contribute a small amount. Finally, many users are content to simply read what is there, may never register, and even if they do, post neither questions nor answers. It is essential to nurture all four of these membership types:
1. The VIPs can be relied on to act as moderators as well as contributors. They initiate interesting threads, they answer questions, they alert the rest of the community to abusive or wrongly located content, and in general radiate a sense of ownership. We have VIP users with many hundreds of posts to their credit.
2. The middle ground users are the workhorses of the community, collectively providing a richness of technical expertise that the VIP users (because of their numbers) cannot match. These users contribute tens or low-hundreds of posts.
3. The hundreds or thousands of occasional contributors provide the “long tail” of community content, posing the questions that collectively cover most of the problems likely to be encountered by other users. Many of these users register, post a single question, and disappear from view, though moderators can see they’re often silently reading other threads weeks and months after receiving a resolution to their own issue.
4. The unregistered or un-signed-in onlookers are a major source of Support cost savings, invisibly finding answers to their questions, and as a result, not phoning or emailing the Support teams.
A healthy growth in registrations is what will deliver that mix of contribution levels. Without it, a community will increasingly rely on its VIP users who will run out of things to say if not continually prompted by questions and comment from other users. Here’s how SophosTalk has performed, year on year:
It’s important to understand where a community’s content is coming from. Many Sophos employees signed up to SophosTalk, but we didn’t want all of the answers to come from them even though they’re experts. We went so far as to set in-house rules preventing immediate responses. The community is there to nourish a peer-to-peer Support network, and this is where the real ROI comes from. For this reason, we run monthly reports on the split between Sophos and non-Sophos contributors and contributions:
From this, it is clear that we’ve been getting an increasing number of posts from non-Sophos contributors for each one from within Sophos, with an exceptional spike of 15 : 1. Given that we need an average of only 3.8 posts to close a thread, this is real peer-to-peer Support at next to no cost of ourselves (there are minor overheads accruing from community management and on-going operating costs). Many customer questions or cries for help are resolved with no involvement from Sophos. This happened for the first time as early as 36 hours after SophosTalk launched, where a customer question (education, USA) received three partial answers (network security, Peru; computer security, Germany; partner/reseller, Belgium) before closure from a source in academia, Austria.
There are numerous things a community management team can do to ensure the community works smoothly. For example, are users able to find answers to their questions? For this reason, we’ve closely tracked the performance of searching within the community. To assist with this, we’ve encouraged tagging (keywords), and we’ve added content to plug gaps. Searching, in general, will succeed if the community is disciplined and irrelevant content is discouraged or weeded out. Here’s how SophosTalk has performed:
In fact, we report on about a dozen headline metrics monthly. Some are simply to do with billing and not much concerned with the success of SophosTalk itself, or the ROI we realise. Others track contributions by product and by community role, others again focus on indicators of content quality. I don’t have space to cover them all, but I hope this paper has given a feel for how we strategically positioned our communities and how we identified the metrics to track.