Blood, sweat and tears: a failed project and lessons learned


By Ruby Tomar


In a complex global organization initiatives are taken so that development of products/solutions gets easier, better and optimized. Such initiatives are not main stream projects because they do not contribute to the core of the company. These are typically taken up in the free bandwidth of engineers so that the development of the core is not disturbed. Failure rate of such “side” projects is high as such projects do not get the time and resources required for successful completion. This case study summarizes a real project which failed to deploy and highlights the lessons learned along the way.

Problem and Motivation

User Interface (UI) is the face of any electronic device. To serve a global consumer base, the strings in the UI are localized in different languages. Testing whether the strings display correctly in all the supported languages is a humongous task.

  • Every screen needs to be generated in all supported languages. Our devices support 26 languages; average time spent to generate the screens per language is 140 hours. Therefore, the total time required just to generate strings in all languages is 3640 hours (26×140) per device.
  • All strings in each screen need to be validated for errors and truncations. This is a very taxing and time consuming activity as in most cases the testers do not understand the language being tested. Testing involves manually comparing what is shown in the UI with the “expected strings” document. The comparison is manual and is prone to errors. The problem is aggravated if the testers do not understand the language which they are testing.

Our solution

The UI technology was being developed by another team. We wanted to create a tool to automate the localization testing of the UI which would help in saving time and resources.  We took this up as an initiative or “side” project. The project involved generating snapshots of screens in the target in all the supported languages, transferring them to the host PC, extracting the text from the images using an OCR tool, comparing the extracted (actual) text with the “expected” text, and reporting the errors.


To read entire article (click here)

About the Author

Ruby-Tomarflag-indiaRuby Tomar

Ruby Tomar is an action oriented, decisive and results focused Program and Project Manager with 16 years of experience in the IT systems. With three patents filed and eight disclosures to her credit, Ruby is process and technology savvy with a strong inclination towards innovation and process optimization. She has worked in automotive, consumer, networking, and telecommunications industries and is an avid reader of technical and management research. She has an MS degree in Software Systems from BITS, India and is currently working as a Program Manager at HP. She can be reached at [email protected]il.com.