Beth Pollock & Andrea Rutherfoord, Citrix Systems, Inc.

In 2008 when our department started using a CMS, we had already made a business case for the implementation based on the impressive cost-savings realized in Localization. Intuitively, though, we knew that we would also save on authoring costs as a result of content reuse. Shortly after rolling out the CMS, we had to deliver product documentation for several large releases (some unplanned) in quick succession. The largest of these releases, which we worked on through 2009 and the first quarter of 2010, was delivered in half the staff time we had estimated during pre-CMS project planning. The question became, “How can we demonstrate what we know intuitively about content reuse ROI?”

Our first task was to come up with a formula. Multiplying the number of hours required to create a topic by the number of times the topic was used, minus the sunk cost of developing the topic should get us the amount of time saved.

time saved=(number of hours to create topic)x(number of times used-1 sunk cost)

The time saved calculation is based on the premise that a topic would be researched and written for each instance in which it was needed. Before we introduced structured authoring, author assignments were made per-deliverable within a product or project. Consequently, multiple authors would research, write, and maintain the same task or feature independently, which we knew was a costly duplication of effort and reduced accuracy and consistency.

To start measuring, we needed first to calculate the average number of words per topic and the average amount of time required to author a topic. Once we had an idea of how long it takes us to author one topic, we could figure out how much time was saved by reusing that topic. Defining reuse was another tricky variable. Since there are many ways to define the concept of reuse, we agreed to establish a standard and to count only the instances of reuse that met that definition. For the purposes of our study, we counted only

  • Content used in multiple topics as conrefs
  • Topics that appear in multiple deliverables (excluding those included in reused maps to avoid double counting)
  • Topic maps that appear in multiple deliverables
  • Content published to multiple formats

Using historical data, we

  • Counted the number of non-duplicated files within a given release cycle and the aggregate word count for that file set
  • Used our file and word counts to calculate the average number of words per topic
  • Reviewed our resource tracking data for the release cycle we were studying to get the number of staff hours dedicated to the release cycle in question
  • Estimated how long it takes to develop N words of content, with N being the average number of words per topic

So if a topic took 8 hours to develop and we used that topic 6 times in one release cycle, we would count the time saved as 40 hours:

40 hours saved=(8 hours to create topic)x(6 times used-1)

Since the data gathering would be time-consuming, we wanted to have some evidence that the study was worth the effort. Therefore, we did a pilot study of one author’s content. That author looked at her content for the release cycle in question and counted the number of files she had reused. The results were better than we’d hoped for: in this case, reuse had saved us 268 days based on our established formula.

Encouraged by the numbers of our pilot, we expanded our study to all authors. Since the individual authors are most familiar with the details of the content and the maps they produced, we asked them to provide numbers based on the parameters outlined above. The results were astonishing. They also explained why we were able to get a large release (and multiple smaller releases) delivered on time and with fewer resources than we’d ever had for projects of similar size and scope. According to the figures, we had saved 1800 days—approximately the work of 7 full-time equivalents (FTEs)—through content reuse over a period of 15 months (January 2009—March 2010).