Hi guys, several weeks ago, as some of you already know, me, Vasco and Wim started this project: www.agile-localization.com. During this time we have been taking notes and thinking about several topics that we want to talk tackle in the book. After collecting all topics, we realised that we could write a pocket book in parallel with several “How to´s”. This small pocket book could attract readers that are not interested in a full story of the original book, but instead, they want practical exercises to implement within their teams/products. This is the result. Below I want to share the first “How to” and I would like to get your feedback. Your feedback would be important for me to improve and write a final text.

How to use crowd sourcing for validation/testing

Several years ago one of the authors worked in a big mobile manufacture company as a Localization Manager. He was responsible for localizing several products within the organization. These products were delivered to millions of users all over the world, so the localization quality was extremely important for the company in order to keep a good brand image. Imagine yourself buying a phone and when you start using it you barely can understand the translations. Most probably, you will not get the best image of that company. This was happening not only with customers but also with employees. They were not happy with translation quality delivered by vendors.

In the beginning, the author thought this was normal situation. The problem is that people in the team were too involved with the product and they were familiar with product features, therefore it was easy for them to be unhappy with translations provided by vendors. However, the situation started to be too annoying due to repeating complaints of translations. Everyone started to think that the organization couldn´t afford to ship products with such a poor quality level. Adding more testing would not help much, because based on experience, the author thinks that increasing testing will not improve the linguistic quality that much. So, what would be a possible solution for this?

How about if the company would use crowd sourcing to help the validation of the product? This is exactly what they did, so let´s see how it was done:

1) The author started to take screen shoots of the most important products screens. The screen shots were taken in English in order to compare the English version with the one they would have on their own devices.

2) A survey was created with a question for each different screen shot asking: “Do you agree with the current translation?” a)Yes b) No. If No, please can you provide the correct translation.

3) The survey was spread inside of Beta Community and within the company

4) A script was created to analyse all answers

5) All translations with an acceptance below 95% were written on “Improvement List”

6) An avg translation score was calculated to measure how accurate the translations were for each language

7) For each word on the “Improvement List”, the top three suggestions were collected

8) All words within the “Improved list” and their respective suggestions for improvement were sent to the marketing department in order to select a final translation

9) The changes were implemented into the product

10) The survey ran again

11) All metrics were calculated repeatedly

12) The quality was improved drastically

This is an example how crowd sourcing can help companies to increase a product quality.

What do you think? Do you think its useful or not really? Please leave me your comments :)

Did you like this post? Do you want to get more in the future? Subscribe my newsletter below and follow me on twitter: @lgoncalves1979.

I am trying to improve these blog posts with a help from professional designers and editors to give you even more valuable content. If you want to support me on this effort, feel free to contribute with any amount of money that you think it´s fair.

Thanks guys,
Luis