If you’re trying to reach a global audience today, you need to go beyond translation. Localizing content may go some way to helping your audience understand your message, but it only works if it resonates with the audience. To achieve better quality translations, faster turnaround and efficient use of your localization budget, it’s vital to analyze how your processes are working based on solid data.
If you want to make good decisions on your localization process you could benefit from adopting data-driven localization model.
Gathering all the data available to you about your localization process is only the first step. Data is useless unless we understand it in context. The first thing we need to measure success is to understand how we are doing now and compare that against what is considered “good performance.”
Looking at your existing data will give you “baseline numbers”. This number tells you what is currently happening in each process. What it doesn’t tell you is how you are doing. For example, I recently gave a webinar where 40% of the people who registered attended it. Is that good or bad? A recent customer audit showed a 96% on-time delivery. Is that good or bad? To know how you are doing in any process you need a Key Performance Indicator (or KPI) to measure against.
The KPI is the number that defines success for a process. A KPI can be defined by an industry standard. It can be defined by a contractual obligation. Or it can be a stretch goal for that process. It is the number that, when reached, indicates success. And you are the person that sets it.
Going back to my examples: According to bizibl.com, the KPI for the percentage of people attending a webinar they registered for is a range of 35 to 45%. Based on that, my 40% was pretty good. I bet a lot of you assumed the 40% was bad since, intuitively, it might have seemed low.
Regarding the on-time delivery example: The customer’s contract promised 95% on-time delivery, but the KPI was a stretch goal of 97%. As a result, even though the 95% on-time delivery met the contract, it was below the defined KPI. Therefore the result was bad.
You can see that you can’t really tell anything from your data without KPIs. You typically find your KPIs in one of two ways:
Most industries publish what good results are for different processes or actions. I had multiple choices to pull from for my webinar attendance statistic. Check industry newsletters and publications to find some of your process KPIs.
Not only can you pull data for current work, you can also pull it for past work. Analyzing past data can help you define your KPIs as well.
While it is great to use KPIs as a way of tracking success, you also want to use data to warn you about problems or potential problems. And that is where thresholds come in. When a threshold is crossed, some kind of action must be taken. There are at least two different thresholds you should define.
When the number you are tracking for a process crosses an alarm threshold, it indicates there has been a failure in the process. Something is wrong that has to be fixed immediately.
An alert threshold is a little different. When the number you are tracking for a process crosses the alert threshold, that process must be examined. There may, or may not, be a problem. The alert gives you time to examine the process and fix something before it becomes a problem.
Like KPIs, your alarm and alert thresholds will be set based on industry data or historical data. In the next two blogs we will dig deeper into using alarm and alert thresholds.
Check out my webinar Data Drives Desired Decisions, for a detailed example of using KPI’s and thresholds to analyze your data. Or if you’d like to read my next blog in the series on using alarm thresholds, please click here.