WebAIM - Web Accessibility In Mind

E-mail List Archives

Re: Average Time to Create an Accessible PDF

for

From: Duff Johnson
Date: Dec 4, 2017 11:12AM

In my old company, Document Solutions, Inc., we started tagging PDF files for accessibility back in 2001. I believe we were the first service bureau in the world to do this work. Within a couple of years we'd evolved a systematic means of assessing PDF files that never led us astray… or I should say, when it did so it was never by much. :-)

Essentially, you have to:

1. Build the estimating process into your overall time-estimate.

2. Auto-tag untagged files.

3. Compare auto-tagged input and output to identify z-order errors, if any.

4. Assess each document (or a representative sample of documents) on a page-by-page basis reviewing the nature of the content and determining the quality of whatever tagging you are starting with.

For documents without form-fields, we used a 1-20 rating scale. "1" more-or-less equalled about 5 seconds / page; enough for an expert to verify that the structure represents the page-content "out of the egg". "20" meant the page was fiendishly difficult and would likely require 1+ hour of work. We found that experienced operators could perform the necessary scoring in around 5-10 seconds per page, depending on the overall nature of the content. Office documents are a lot faster to assess than magazine pages, for obvious reasons.

This may sound laborious, but it actually makes life SO much easier when you really know what you've got, instead of "finding out". That happens anyway, to some extent, of course.

5. Add 1-2 minutes / document for setup and overhead. This can really be a bummer when the documents are very short and/or very simple (i.e., the auto-tagging results are pretty good), but it's very important not to shortchange the overhead!

Documents WITH form-fields have to be assessed in a far more rigorous way, period. There are field combinations (checkboxes) to check, JavaScripts to assess, invisible fields to consider, etc., etc… It ain't just structure-element order and semantics!

Duff.

PS: In my capacity as Executive Director of the non-profit PDF Association...

<one-time shameless plug>

Join myself and many other PDF experts at PDF Day DC on January 29, 2018. Among others, we have are six sessions on managing for Section 508 conformance and accessible PDF. Learn more:

https://www.pdfa.org/event/pdf-day-washington-dc/.

</one-time shameless plug>




> On Dec 4, 2017, at 11:36, Philip Kiff < <EMAIL REMOVED> > wrote:
>
> I agree with Alastair that there is no good answer to the question of an average time to remediate a PDF.
>
> Estimating the time it will take to remediate something is a bit of a micro-skill in itself. I have a set of criteria that I look at to make estimates that includes: file format (pdf, docx, pptx? Do you have the source file), type of document (standard text? large graphical poster? filled with tables and charts?), complexity (many embedded lists? numbered headings? uses complex tables? lots of non-decorative images?), and length. One special case: documents that require OCR I consider individually unique and almost impossible to estimate without starting to actually work on them.
>
> In addition to those criteria, the amount of time required will also depend on what you mean by "accessible". Everyone on this list I am sure is always trying to make perfect documents that meet or exceed all accessibility requirements everywhere. But there are degrees of perfection, in my opinion. Even within my own work, documents that I remediated just a year or two ago, and that passed the PAC 2 checker cleanly, I would now do additional work on were they to come across my desk again. On the other hand, I would also work more quickly than I did a year or two ago, so maybe that all comes out in the wash in terms of time estimates.
>
> On 2017-12-04 10:39 AM, Josh Schroder wrote:
>> For pure-text documents, I can probably do somewhere around 5 minutes per page, sometimes less.
>
> Having stated all those caveats and exceptions above, and for the purposes of useful discussion, I would say that it takes an average of somewhere between 3 and 12 minutes a page for me to remediate 90% of the standard documents I work on.
>
> On 2017-12-04 10:39 AM, Josh Schroder wrote:
>> I'm also a big proponent of CommonLook PDF. It is expensive, but it really speeds things up and reduces the frustration of dealing with the inefficient workflows in Acrobat. If you do this work often, and you consider the cost of your labor per hour, it can potentially be a really good value.
>
> Regarding remediation software, I would encourage folks to look at AxesPDF instead of CommonLook PDF, at least until CommonLook provides some kind of sane pricing structure for third-party licences to folks who want to get paid to remediate PDFs for others. AxesPDF is available at a fraction of the cost and performs almost all the most important time-saving functions as CommonLook's software does. In terms of cost-per-hour, it would literally be cheaper for me to pay someone else to remediate files for me than it would be for me to purchase a license from CommonLook that lets me remediate files for you.
>
> Phil.
>
>
> On 2017-12-04 10:39 AM, Josh Schroder wrote:
>> For pure-text documents, I can probably do somewhere around 5 minutes per page, sometimes less.
>>
>> I'm also a big proponent of CommonLook PDF. It is expensive, but it really speeds things up and reduces the frustration of dealing with the inefficient workflows in Acrobat. If you do this work often, and you consider the cost of your labor per hour, it can potentially be a really good value.
>>
>> Josh Schroder
>> Web Administrator II
>> Office of Strategic Communications
>> Texas Department of Licensing and Regulation
>> (512) 936-8937
>>
>> -----Original Message-----
>> From: WebAIM-Forum [mailto: <EMAIL REMOVED> ] On Behalf Of R.U. Steinberg
>> Sent: Monday, December 4, 2017 7:58 AM
>> To: WebAIM Discussion List < <EMAIL REMOVED> >
>> Subject: Re: [WebAIM] Average Time to Create an Accessible PDF
>>
>> A one page PDF with form fields takes me much longer to remediate than a 10 page straight text PDF. I could spend 8 hours on the form vs. less than an hour on straight text. If the PDF has color that fails contrast, that can be fixed in Acrobat Pro, but that is also a time factor. Complex tables are also a pain. I know that doesn't answer your question, but I suggest you do an inventory of sorts on what types of PDFs you have (forms, text only, with or without color, with or without images, with or without tables, etc.)
>>
>>
>>
>> On Mon, Dec 4, 2017 at 5:51 AM Alastair Campbell < <EMAIL REMOVED> > wrote:
>>
>>>> what everyone considers to be the average time it takes them to make
>>>> a single page document (Word, PDF, HTML) that is inaccessible and
>>>> output it as an accessible PDF.
>>>>
>>> I'm afraid the scale of different is logarithmic different for simple
>>> and complex cases, so there is not realistic answer for this.
>>>
>>> For example, if you have a simple Word document with a couple of
>>> (properly marked up in Word) headings, it is hardly any time.
>>>
>>> An identical looking page that was an image of the same text, that
>>> needs to be OCRed and structured by hand would take a lot of time.
>>>
>>> If there are more complex structures (tables, quotes etc) the time
>>> goes up again. If there are images that need alts, or video content,
>>> the time goes up again.
>>>
>>> For a 100 page document from a good Word source doc, you might spend 5
>>> minutes. An identical looking doc from an un-structured indesign
>>> source might take 5 days. It is that much difference, so any 'average'
>>> would be wildly different depending on what the sources were.
>>>
>>> If you are working with a PDF doc then the difference is mainly that
>>> you can't work on the source, so it had better be the final version as
>>> any changes from the source will undo your work.
>>>
>>> I haven't really tackled HTML to PDF as generally if you have an
>>> accessible HTML version, you don't also need an accessible PDF version.
>>>
>>> I hope that helps in some way, but sorry there isn't a nice answer!
>>>
>>> -Alastair
>>> >>> >>> archives at http://webaim.org/discussion/archives
>>> >>>
>> >> >> >> >> >> >> >
> > > >