Assisting unassisted support - why document management matters

A few days back I upgraded to version 4 of the Mozilla Firefox browser because I wanted to test the new user interface and general stability with CogniDox in particular (it's fine) and see if it really is faster (it seems to be).

Like the average user, my pre-reading on the new features was minimal. Ok, zero.

All was well until I looked to the bottom right corner for an installed extension I wanted to use. But nothing there. No status bar. Went to the View menu to enable status bar but nothing there. Surely Mozilla can't have cut off all those 3rd party add-ons so brutally by removing the screen real estate where the icons reside? So I searched the Web for an answer. Within minutes I knew that the status bar had been replaced by the new Add-on manager and I knew exactly how to enable it. My icons were back.

It didn't matter to me during those minutes that I was an example of "unassisted support" (aka web self-service), but I was. Because I turned to the general Web rather than to a specific company support site, I guess you could also say this was an "unmediated" support search.

I've just been reading a Technology Services Industry Association (TSIA) research paper called "The Current State of Unassisted Support" which came out in February 2011. They're an umbrella body for many well-known enterprise and consumer technology companies (membership ranges from Accenture to Yahoo!, passing e.g. Juniper Networks, Red Hat and VMware along the way). The point being that when they do a member survey, it includes a sample of the biggest names in the industry.

Left to my own intuition I would have predicted that unassisted support is on the increase. And maybe it is, as companies cut back on technical support staff numbers and rely instead on other channels. We all know that answering phone calls and e-mails for assisted support is an expensive option. The more you can reserve these channels for complex and unusual queries, the better.

But what the report shows is that unassisted support success has been declining - the success rate in 2003 was 48% and is now down to 39%.

It seems that users have a low opinion of the chances of finding their answer in the FAQ or knowledge base. The bottom line seems to be that the more money a company spends on handling a support call, the happier the user will turn out to be. This might seem like a statement of the blindingly obvious, but it is not good news for any industry that needs the speed, accuracy and low cost of a successful unassisted support scenario.

It looks like the behemoth CRM platforms (or ERP systems with a CRM module) that were installed in the '90s or early '00s have not delivered on their promise. They are too cumbersome for ensuring up-to-date content, and frequently offer a bad user experience.

If we just switch the emphasis so that online communities, collaborative Wikis and social networking sites deliver user-to-user support, are we likely to see a major improvement? There might be some, but I can't see any logical reason why we should expect a quantum leap. Knowledge needs to be there in the first place, before someone in the crowd can impart that wisdom. Another statement of the blindingly obvious maybe, but the use of technology alone won't make the difference we crave.

As product development times get shorter and time-to-market priority increases, any shortcuts taken on product knowledge capture and management are stumble stones on a future path. We may be agile, but undocumented development agility is not as great a thing as is often claimed.

So what do we do? It's worth trying again until we get this right. As the TSIA report says, if a company receiving 10,000 customer incidents a month were able to increase self-service success by just 5%, over the course of a year that would save nearly $1 million.

Part of the solution lies in a new generation of knowledge management products. These may share some of the same buzz-words as 1st generation efforts, but the huge difference lies in their usability. Another helpful technology is federated, intelligent search that can find answers across different content repository silos. Take an example from Mentor Graphics - by incorporating search into every 'create a help ticket' operation, they're able to achieve self-service fix rates of 40%. Being able to measure what users are using (analytics) and giving them the chance to score the quality of information is useful too. Many of those help tickets are not bug reports or user errors, they're enhancement requests that should be diverted back into your product roadmap.

But we ignore the human designer element at our peril. When we as product developers build something new, then for a while knowledge about how it works lies in our heads. Making time to transfer that knowledge may seem sweetly old-fashioned, but it's going to make a difference if you later want to see the benefits of unassisted support. Until you can "engineer the experience" to the extent that nobody ever needs to ask how, then this is needed.

That puts a lot of the responsibility on to software products such as CogniDox - how can we help to capture knowledge quickly and efficently in a world that will never again see the long product cycles of the Waterfall era?

We're up for that.