The principles of support


The Autodesk support organization defines a set of “support principles” that establish a baseline for quality service. The principles are supposed to be straight-forward, common-sense guidelines such as:

  • Provide credible and timely responses
  • Make and honor commitments
  • Confirm resolution
  • Document cases
  • Communicate professionally
  • Escalate immediately

We do weekly review of randomly-selected cases for principle compliance, and the review scores are a lead measure of our success. They’re a lead measure because management considers principle compliance a predictor of success (where success is measured by the quarterly Net Promoter Score aka customer satisfaction).

In practice, you can go around in endless circles discussing what these principles actually mean. I’ve sat through whole meetings where we never got past the first definition of the first principle.

Based on my experience on the front-line, here’s my take on the principles. I’ve reduced the principles to single-sentence bullet points with [I think] plain English words.

  • Respond quickly with either a solution or some questions that clarify the problem.
  • Let the customer know when you’ll get back to them, and do what you promise.
  • Don’t let cases sit inactive for over 2 days: if you’re waiting for information, follow up with the customer; if you’re researching the problem, let the customer know whatโ€™s happening.
  • Document cases so that others can understand the problem and the solution.
  • Communicate like a pro: clear, concise, easy to understand, and with the same tone you would expect from others.
  • Don’t delay–when you tell a customer you’re going to escalate or transfer the case, do it right away.

Support case word cloud


Here’s a word cloud I generated from the summary and description fields of the Softimage support cases that came in so far this year. The word cloud shows the top 50 words (after I manually removed words like “Softimage”, “using”, “hello”, and others). I did about 30 search-and-replaces to get rid of words that weren’t feature-related, but there’s still some, like “attached”, left in the cloud.

Surveys, the ultimate question, and the net promoter score


The above poll is an example of an ultimate question that can be used to calculate a net promoter score.

If you’re a Subscription customer and you’ve logged some support cases, you may have received a survey invitation. You may have even filled the survey out ๐Ÿ˜‰

These customer satisfaction surveys are based on the Net Promoter methodology, which uses a “would you recommend” question to divide customers into three categories: Promoters, Passives, and Detractors.

The “would you recommend” question is known as the ultimate question, and uses a 0-to-10 point rating scale:

  • Promoters (score 9-10) are loyal enthusiasts who will keep buying and refer others, fueling growth.
  • Passives (score 7-8) are satisfied but unenthusiastic customers who are vulnerable to competitive offerings.
  • Detractors (score 0-6) are unhappy customers who can damage your brand and impede growth through negative word-of-mouth.

Based on the survey results, a Net Promoter Score is calculated:

The general idea is that if you know your NPS, you can focus in on what you need to change to increase promoters and decrease detractors. This requires a dialog with your customers; currently support managers are responsible for following up with detractors to find out what went wrong during the support case.

So far, in practice, what I observe is that we don’t get enough surveys returned, so every single detractor has a huge impact on our NPS. Looking at my own NPS scores, I think I would have to focus not on detractors (unless I really messed up) but rather on the passives. Because if you had only one detractor, but everybody else was a passive, you’d end up with a negative NPS. Hardly something to brag about, and certainly not something that will look good to management.

Time tracking



After a few abortive attempts at time-tracking, we’re rolling out a new time-tracking system here in product support. One of the main goals it to understand how much time specialists spend on “reactive” events (responding to support requests) versus “proactive” events (writing KB articles, developing training, blogging, and manning the forums).

The difficulty for me with time tracking is that these days my time and effort is so interrupt-driven. For example, I’ll be drafting a blog post when I see a new scripting question on the mailing list. So I jump on that to try and answer before someone else does ๐Ÿ˜‰ But then one of my support cases is updated, so I jump on that and reply.

Meanwhile a new case comes in, so I take that and start research. Then maybe I have a meeting, or someone needs me to explain how to do something in our help desk system (I’m the local superuser). Eventually I rewind back through everything, possibly being interrupted again at any time. Or being distracted by my almost compulsive checking of my email inbox and the support incoming queue;-)

I admit this may not sound like the best way to work, but it works for me. I know many others on the support team prefer to handle one thing at a time.

I’ve found I need some utility, like the one in the screenshot, that allows me to have multiple timers going. That way I can accurately keep track of what I’m doing.

The support funnel


A company’s product support model can be visualized as a funnel.

For every N customers who need help, only a subset will require direct one-on-one support

  • At the top of the funnel is the self-service component, where you hope the majority of users can find their answers. Ideally, there’s a support portal that brings together documentation, KBs, FAQs, blogs, forums, videos, and tutorials, and provides an integrated search, and possibly some sort of interactive guided question-and-answer technology. In practice, most self-service sites are usually a loose collection of these different components, which rely on the user’s ability to define good search queries.
  • Next in the funnel are forums and user communities, where you can get help from other users and, in some cases, from staff who monitor the communities.
  • For a company like Autodesk, partners provide the next level of support (this is not so true for Softimage, but is true for most other Autodesk products).
  • Finally, at the narrow end of the funnel, is the one-to-one support provided by people like me, the “technical specialists” (I’m not sure why, but I loathe that title).

Now, the funnel analogy is not “correct” in the literal sense. In a real funnel, everything that comes in eventually flows through the bottom of the funnel. In the support funnel model, the number of customers decreases as you move down the funnel. A more accurate diagram of the support model would be a sankey diagram, which shows the proportionate number of customers served by each level of support: