How to reasonably predict the time and cost to complete a translation project

A lot of our clients depend on our services regularly.  Something we do to streamline our collaboration is share statistical heuristics that can help predict the value of their regular work and when they can expect us to deliver.  In practice, computers are used to analyze every word of content and produce a valuation down to the penny. Since only our team have access to all the tools and detailed training, heuristics were developed to improve transparency and collaboration.  In this article, we will dissect these heuristics and help you predict the value and time required to deliver translations of standard Word documents. The sections of this article are as follows: 

• Value
• Time
• Exceptions

The benefit of these statistical heuristics is their simplicity, but the compromise is their accuracy.  While the heuristics won’t get you a perfectly accurate valuation, they will get you in the right direction.  

Value

Many variables go into a project’s valuation, including characteristics related to the file type, document type, source language, target language, quality threshold, timeline and many others.  To acknowledge the uncertainty in the estimate of the heuristics, ranges will be provided for each one. The following heuristic captures most of the variance with only knowing word count:

Dollars = ((English Words rounded up to the nearest thousand)/1000)*(8/10) ± 20%

“English Words rounded up to the nearest thousand” refers to the English word count, whether it’s the source language or target language.  When translating into English, the word count tends to expand or contract. For those who are curious, below is an abbreviated conversion table made of additional statistical heuristics to convert to and from English:

To English Multiple	Word count	From English Multiple
1.3	Slavic languages: Russian, Hungarian, Czech, Slovak, Polish, Ukrainian, Serbian, Croatian, Bosnian, etc.	0.8
1.2	German, Norwegian, Swedish	0.8
2.0	Finnish	0.5
0.5	Japanese	2.0
0.7	Chinese	1.4
1.5	Hebrew	0.7
1.0	All others	1.0

Of course, the above formula is useful if you know the word count although a good guess is equally helpful since this is all a guess anyway.  

Word count can be estimated from page count, which is far easier and more accurate to determine.  Since clients typically depend on mental math for these calculations, we propose converting from words to pages by assuming that pages have between 200 – 300 words on them, but the number can be double that in particularly dense documents, like contracts.  Below is a heuristic if you want to depend on page count instead of word count:

Lower Limit: Dollars = pages*40

Upper Limit: Dollars = pages *100

The lower and upper limit capture most of the variance in the valuation, but not all of it.  Also, note that there is a minimum charge currently set at $120, so the cost of 1 page (or 250 words) and 3 pages (900 words) might be the same.   

Time

Time is a generally simpler calculation, but admittedly, even when trying to be accurate, it is an educated guess.  How fast can a translator type? How fast can an editor read? Can they keep that up for hours on end? In the industry the heuristic used is that a translator can type about 2000 words per day, while an editor can review about 4000 words per day.  Of course, this varies by language, person, and many other variables, but it’s a starting point. For larger projects, people can work faster, on average, but at the same time, the need to guess is eliminated as it’s usually safer to ask in the vendors in those instances.  In general, language service providers (LSPs) guess, adjusting for experience:

Days = 1 + ((Word count)/2000) + ((Word count)/4000)

Notice the “1” at the front.  This extra day is used to account for time zone differences, scheduling and any number of things that could happen during the project.  Here is the heuristic again in terms of pages:

Lower Bound Days = 1 + (pages/10) + (pages/20)

Upper Bound Deliver = 1 + (pages/3) + (pages/10)

As for the time of day, the standard is end of business unless the client makes a specific request.  

Exclusions

In the heuristics shared above, no pre-processing, post-processing or quality assurance is taken into account – this is purely for translation work of typical Word documents, which is usually most of the work. Here are a few things to think about if you want to extend the above heuristics to include additional things:

Pre-processing: In general, regardless of the task, adding 20% to both value and time might be a good start.  5-15% could be accurate, but not conservative enough.

Post-processing: For each pre-processing task, there is typically a post-processing task.  This is because pre- and post- processing tasks are like unpacking and repacking a file for the translation process, so you can use the same 20% heuristic for value and time.

Client Review: Client Review (when the client reviews interim deliverables) is never accounted for in our timelines.  Thankfully, this is something that you have a better chance of guessing than we do. Client review takes place after translation, but before post-processing and QA.  

Quality Assurance: This activity usually takes one day or less, but can increase dramatically for more complex projects that include audio, video or testing. In cases were it takes less than a day, the quality assurance is typically free.   

Multiple languages: Adding languages multiples the value (so two languages is 2x), but has no effect on the time.  The translation process, regardless of which language, take place simultaneously so there is no impact on the timeline.  One exception is if a rare language is used and extra time needs to be added just for scheduling purposes.  

Scheduling and time zone issues: Scheduling and time zone depend entirely on the language as language and geography are strongly correlated.  Also, since timelines are measured in days (and not hours) simple rounding can usually accommodate for this. Up to two days would be added depending on the time of day the project was approved by the client and the number of languages involved in the process.  

If you’d like to learn more about how BURG Translations helps you ensure high-quality translations, contact us today.