Suggestions about an agency fuzzy grid
Thread poster: Armando Calderon
Armando Calderon
Armando Calderon  Identity Verified
Mexico
Local time: 23:51
Member (2016)
English to Spanish
+ ...
Mar 8, 2017

Hello,

I am about to sign an agreement with a translation/localization agency. They specialize in IT, UI strings, marketing, etc. They have shared with me their fuzzy grid. I know that they are plenty of posts about fuzzy matches, but I do not really know if this particular grid will be functional for me in the long term if we talk about software strings and TMs provided by the agency.

Their grid is as follows:

Trados Grid


What do you think? Would I need to re-negotiate my rates? The rates they could offer to me were $0.08 USD per word and $21.00 USD per hour (EN>ES).

Any help and suggestions would be much appreciated.

Thank you.



[Edited at 2017-03-08 22:25 GMT]

[Edited at 2017-03-08 22:26 GMT]


 
Mario Chavez (X)
Mario Chavez (X)  Identity Verified
Local time: 01:51
English to Spanish
+ ...
Walk away slowly Mar 8, 2017

I find that grid a bit insulting, especially the so-called contextual match. Machines and software programs don't have brains to think and can't understand context, just frequency algorithms (if x word appears next to y words a sufficient number of times, then it's a context-valid match).

Whenever I sign a contract with an agency, I bring them my own rate card (or fuzzy grid).


 
Mirko Mainardi
Mirko Mainardi  Identity Verified
Italy
Local time: 06:51
Member
English to Italian
Context match Mar 8, 2017

Mario Chavez wrote:

I find that grid a bit insulting, especially the so-called contextual match. Machines and software programs don't have brains to think and can't understand context, just frequency algorithms (if x word appears next to y words a sufficient number of times, then it's a context-valid match).

Whenever I sign a contract with an agency, I bring them my own rate card (or fuzzy grid).


Mmmh, are you sure about that? AFAIK, a "context match" (or ICE, or 101% match) is a 100% match preceded and followed by two segments with a 100% match (those 2 TUs are the "context"). If they pay you 0% for them, it means you don't have to touch those segments at all.


To the OP: I believe the per hour rate should be higher compared to the per word rate (e.g. 0.08 psw > 25/30 per hour). Also, if you can't negotiate rates upwards, then perhaps you could try to have them remove the 75-84% fuzzy band, or at least get a lower "discount" for it. Why should 85-94% matches be paid the same as (the "less useful") 75-84% ones?

As for the reliability of legacy TMs... well, that's a risk, but if they're giving you those TMs, I guess they'll be expecting you to use them and adapt to their style/terminology anyway...

[Edited at 2017-03-08 23:14 GMT]


 
Mario Chavez (X)
Mario Chavez (X)  Identity Verified
Local time: 01:51
English to Spanish
+ ...
Context Mar 9, 2017

Mirko, after working with Trados, Idiom WorldServer and DVX3 for close to 15 years, I'm pretty confident in my comments above. Sure, 100% matches and context matches may not be touched (per client's instructions) and therefore I don't have to touch. But I still have to read them. So I insist on getting paid for them, unless the client compensates by offering a reasonable (i.e. high) word rate for the base rate.

Apparently, reading surrounding context isn't getting much attent
... See more
Mirko, after working with Trados, Idiom WorldServer and DVX3 for close to 15 years, I'm pretty confident in my comments above. Sure, 100% matches and context matches may not be touched (per client's instructions) and therefore I don't have to touch. But I still have to read them. So I insist on getting paid for them, unless the client compensates by offering a reasonable (i.e. high) word rate for the base rate.

Apparently, reading surrounding context isn't getting much attention these days?

Since all things are negotiable, to Mirko and the OP: feel free to negotiate the grid to your tastes (i.e. your professional experience). There's usually a middle ground that can be reached.

Another way of approaching the fuzzy grid/rate card dilemma: offer to charge a lower per-word rate for all of the text (no grid). I have done that to the mutual satisfaction of the parties because it simplifies things for me and for them. The trick is to find that happy medium rate that you can live with. It simplifies invoicing as well.
Collapse


 
Philippe Etienne
Philippe Etienne  Identity Verified
Spain
Local time: 06:51
Member
English to French
A grid is meant to reflect the time you save processing fuzzies compared to translating from scratch Mar 9, 2017

Armando Calderon wrote:
...
Trados Grid


What do you think?...

Well, this grid means that:
- your editing word rate is a tenth of your translation rate
- you save half your time on 75-95% fuzzies compared to retranslating them from scratch
- 95-99% fuzzies extracted from the TM save 2/3 of your time

Any line with a zero means zero money, therefore zero work.
Find a way to skip/hide/lock/mark the related segments and don't reread 100%/reps/CM if they're not paid.
It's vital that the agency realise and acknowledge that your time is worth money and that you don't take responsibility for dodgy segments if the correlated discount band is 0€. Give them a few examples of rotten matches so that they change their mind and pay you everything they want reread. Or let you know that those issues are taken care of downstream in the translation worflow.

It may sound like hair-splitting, but who exactly started splitting hairs with match bands, discounts and stuff?

From my experience (starting from, hum, Trados 3), in my language combination, I would find your grid weak. I know it's a type of grid commonly offered, and I turn down such terms. Or offer to raise my rate by 30%.
I am comfortable with the following, which I think fairly shares the time-saving benefits of a CAT tool.
100%/reps: 20-25%
85-99% match: 60%
0-84% match: 100%
But I also take on anything more beneficial to me, such as no grid.
I find the grid above reflects the time I save using a CAT tool. In other words, my hourly rate working on a translation full of fuzzies is a bit higher (the CAT tool needs to be paid for) than with a similar translation without fuzzies.
For me, anything remote from this means that using a CAT tool leads to a lower hourly rate than not using it.
Which is idiotic since tools are supposed to make you work more quickly AND earn more per day, and not to be ripped off on an industrial basis and with a smile.
I wrote about grids and consequences in more detail there:
http://www.proz.com/forum/money_matters/268387-rates_for_weighted_words.html#2291899

You could also ask whether they tick the internal fuzzy match/homogeneity option (another potential 10-40% hourly revenue loss), but that's another story.
With such a discount grid, I'm sure they do.

Philippe

PS: I hadn't paid attention to the "editing discount grid". Does anybody read every second word in a text to make sure there are no errors left? Have agencies invented discounted editing? It's funny how some agencies have no clue whatsoever about what they're doing. Well here at least you're paid to reread each word, which is good.

[Edited at 2017-03-09 07:45 GMT]

[Edited at 2017-03-09 07:54 GMT]


 
Jan Truper
Jan Truper  Identity Verified
Germany
Local time: 06:51
Member (2016)
English to German
What Philippe said Mar 9, 2017

Armando Calderon wrote:

Their grid is as follows:

Trados Grid

What do you think?


I think they are trying to make undue profit at your expense in the 75-85% range.
I don't know about English>Spanish, but for English>German I usually have to rewrite the whole sentence in this range, so I charge the full 100%.

And as Philippe said,
"ask whether they tick the internal fuzzy match/homogeneity option (another potential 10-40% hourly revenue loss)".
That's a biggie.


 
Thayenga
Thayenga  Identity Verified
Germany
Local time: 06:51
Member (2009)
English to German
+ ...
CAT tools aren't free Mar 9, 2017

Mario Chavez wrote:

I find that grid a bit insulting, especially the so-called contextual match. Machines and software programs don't have brains to think and can't understand context, just frequency algorithms (if x word appears next to y words a sufficient number of times, then it's a context-valid match).

Whenever I sign a contract with an agency, I bring them my own rate card (or fuzzy grid).


I agree with Mario. Plus, you need to renew your CAT tool licence ever so often. So you should actually add a position to your invoice, what percentage of your costs is covered by the time you've spent using your CAT tool (in some cases apparently only) for the agency's benefit.

And it needs to be made perfectly clear that 100% matches are not even being glanced at. Consequently the translator is not responsible for any mistakes of any nature within the 100% matches. If I would be into "hair-splitting) I'd go as far as saying that even the act of skipping these segments takes time.


 
Armando Calderon
Armando Calderon  Identity Verified
Mexico
Local time: 23:51
Member (2016)
English to Spanish
+ ...
TOPIC STARTER
And "homogeneity" enters into the game Mar 9, 2017

Hello everybody,

First of all, I would like to thank you for your interesting responses and points of view. They are really helpful to me.

Phillippe was right. They use homogeneity (internal fuzzy matches) when calculating project wordcounts. I really do not know what is this. Is it another Trados function used by agencies to hair-split rates?

Maybe one day, CAT tools will integrate a "chronometer" just for statistical purposes which probably will b
... See more
Hello everybody,

First of all, I would like to thank you for your interesting responses and points of view. They are really helpful to me.

Phillippe was right. They use homogeneity (internal fuzzy matches) when calculating project wordcounts. I really do not know what is this. Is it another Trados function used by agencies to hair-split rates?

Maybe one day, CAT tools will integrate a "chronometer" just for statistical purposes which probably will be used by agencies for "another obscure purposes". Beware, editors!


Edit: typo

[Edited at 2017-03-09 20:07 GMT]
Collapse


 
Philippe Etienne
Philippe Etienne  Identity Verified
Spain
Local time: 06:51
Member
English to French
Homogeneity Mar 9, 2017

I tried to explain the idea behind it there:
http://www.proz.com/forum/memoq_support/307876-pre_translation_a_weird_problem.html#2598863

Armando Calderon wrote:
I was thinking: Maybe one day, CAT tools will integrate a "chronometer" just for statistical purposes which probably will be used by agencies for "another obscure purposes". Beware, editors!

I.T. is no longer a tool, it's a lifestyle!
Without sounding too paranoid, who knows what server/client setups can already record. Expect soon Big Data-crunching systems to unearth obscure trends, dissect translator behaviours and more generally squeeze as much juice as poss from IT to slash costs, part of which is us.

Anyway, let's write again these wise words: "there's a lot of money to be made in the translation business... provided you're not a translator"!


 
Armando Calderon
Armando Calderon  Identity Verified
Mexico
Local time: 23:51
Member (2016)
English to Spanish
+ ...
TOPIC STARTER
Grid viability with homogeneity Mar 9, 2017

Thanks for the link Philippe, very useful information about this homogeneity function.

So, with homogeneity there would be more fuzzies, therefore less pay.

I was thinking about proposing a similar grid as yours and see what they say:

100%/reps - 20%
85-99% - 60%
0-84% - 100%

Is this option viable even with homogeneity?

Thanks!

Armando


 
Philippe Etienne
Philippe Etienne  Identity Verified
Spain
Local time: 06:51
Member
English to French
Better late than never Apr 4, 2017

Armando Calderon wrote:
...Is this option viable even with homogeneity?

Let's see the extent of the damage with a random example with enough words to make it relevant.
I took a real case of dismantling instructions, with a TM populated with past translations.
CAVEAT: each text will yield different outcomes, so it's really an academic exercise to illustrate the impact of homogeneity.

Homogeneity OFF
Match types: Words: Percent: Equivalent words:
-------------------------------------------------------------------------
[Saving space]*
-------------------------------------------------------------------------
Total words = 7380
CATCount = 5086,40

Homogeneity ON
Match types: Words: Percent: Equivalent words:
-------------------------------------------------------------------------
[Saving space]*
-------------------------------------------------------------------------
Total words = 7380
CATCount = 4601,90

==> FEE LOWERED BY 9.5%

Now let's see with the agency discount grid:

Homogeneity OFF
Match types: Words: Percent: Equivalent words:
-------------------------------------------------------------------------
[Saving space]*
-------------------------------------------------------------------------
Total words = 7380
CATCount = 4799,70

Homogeneity ON

Match types: Words: Percent: Equivalent words:
-------------------------------------------------------------------------
[Saving space]*
-------------------------------------------------------------------------
Total words = 7380
CATCount = 3814,48

==> FEE LOWERED BY 20.5%

WWC discrepancy (income delta) between the two grids:
- 7.5% with homogeneity OFF
- 17% with homogeneity ON

Stronger grid = less sensitivity to the Homogeneity status (obviously)
Big TM, lots of matches = more sensitivity to the Homogeneity status (obviously)

impact on Homogeneity on WWC depends on content: catalogues, product descriptions and generally repetitive text can decrease the overall fee by a massive amount (-70%), even more so if your grid is ridiculous.

When those next-gen CAT tools became widespread 5-10 years ago, all with this option available, it meant that agencies ticking that box would get their translations AT LEAST 10% cheaper compared to before with the ubiquitus old-Trados (pre-v2009). For us translators who lived the shift, this option meant an hourly income lowered by 10% or more. The alternative was to increase our base rate accordingly only to maintain our income.
In this other discussion about this rip-off feature there:
http://www.proz.com/forum/memoq_support/289898-make_sure_agencies_leave_homogeneity_switched_off_when_running_statistics_in_memoq.html , a WWC went from 2000 to 1100 just ticking that box (45% less).

Nowadays, one just has to assume that weighted wordcounts are based on internal fuzzies/homogeneity turned on. And remember that WWCs from 2017 are at least 10% lower compared to 2000 with the same discount grid.

Good luck,
Philippe

* .csv old-Trados compatible MemoQ analyses available


 


To report site rules violations or get help, contact a site moderator:


You can also contact site staff by submitting a support request »

Suggestions about an agency fuzzy grid







Trados Business Manager Lite
Create customer quotes and invoices from within Trados Studio

Trados Business Manager Lite helps to simplify and speed up some of the daily tasks, such as invoicing and reporting, associated with running your freelance translation business.

More info »
TM-Town
Manage your TMs and Terms ... and boost your translation business

Are you ready for something fresh in the industry? TM-Town is a unique new site for you -- the freelance translator -- to store, manage and share translation memories (TMs) and glossaries...and potentially meet new clients on the basis of your prior work.

More info »