18 January 2010

XBRL - so how much effort is REALLY involved?


For years now (well, the past two to three years) the assumed wisdom has been that the creation of a set of XBRL documents for filling with the SEC would require between 40 to 80 person hours. Lately the figure of 30 to 50 has been used, with "10% taking more than 80 hours".

A recent survey raises serious questions about these estimates. While second and subsequent filings have taken significantly less time (as expected) then initial filing time commitment is much higher than has been touted. The lesson is that XBRL, and especially Phase II detailed tagging of footnotes, will cost businesses significantly more than they have been told to expect.

The facts are coming out, and it is time to stop pretending, and stop misleading. When 57% of respondents say that the effort for XBRL filing with the SEC exceeded 120 hours, there is much to do to restore the credibility of those providing resource estimates. Some real numbers would help.

The promise

Supporting the "its quick and easy" argument are quotes In a Journal of Accountancy article in November 2009,

Typically, companies take between 30 and 50 hours to complete their first-year submissions in XBRL, with about 10% of companies taking less than 10 hours and 10% taking more than 80 hours, according to Paul Penler, executive director of assurance services for Ernst & Young.  In year two, when companies will be required to detail tag footnotes, they might expect the process to take between 200 and 250 hours if the company does it itself or with a filing agent, he said.

I believe these estimates are bogus at best. But lets also be clear, Paul is not the only one who has made these estimates. The XBRL community has consistently underestimated the actual efforts involved in building taxonomies, tagging documents, time to adoption, rate of adoption, etc. And all these translate, simply, into underestimates of the true costs of XBRL. I have been as guilty as anyone in the XBRL community, of underestimaing effort and overestimating businesses willingness to invest their limited resources into something that I can see the benefits of, but the the investing business person does not yet understand. I sit on numerous calls listening to the high priests of XBRL making statements like "They just don't get it" and "If only they would..." and "We really need ot get the message out" and the best - "They just aren't listening". 

No plan survives contact with the enemy

Well the message is out. It is not as easy, as quick or as cheap as promised.

Last week the results of a joint AICPA/XBRL US Inc survey on company readiness of XBRL was released. What a difference real results make. The "plan" - convince those that will have to actually spend the money that it is cheap and easy, has just met the "enemy" - the reality of cost and effort.

I love the wording in the results. Always start with the "good news" first. Then the rest seems muted.

Preparation time for creating XBRL-formatted financial statement decreases dramatically after the first filing, according to respondents who have prepared multiple XBRL financials:

  • 57% said it took them over 120 hours to complete their first submission; 64% of those that filed a second time around said it took less than 40 hours
  • 45% of those that filed a second time said it was significantly easier the second time

Isn't that fantastic news: preparation time for creating XBRL-formatted financial statements decreases dramatically after the first filing. Ummm, we knew that and have been saying it for years.

 But - not 10%, but 57% said that it took them over 120 hours. Not 10% taking more than 80 hours. In fact, looking at the detail of the results and you get only 16% saying that it took them less than 80 hours. Not 90%.

And repeat effort? Well, I think we can guess that the results would not match the promise. I particularly like the line in the summary report: "45% of those that filed a second time said it was significantly easier the second time". Significantly easier for only 45% of filers. I certainly hope that the 3rd and 4th filings are significantly easier. Being fair, 14% of respondents to that question had not filed for a second time, so the total percentage of significantly easier is greater than 50% of second time filers.

This has to raise questions about the estimates being provided for the effort involved in creation of detailed tagged footnotes. I've written before that this is a major concern to me. I remain very concerned that the overhead involved, at a time when companies are tightly stretched for resources, could tarnish the image of XBRL as a standard for increased efficiency and effectiveness.

Certainly the "real" effort will play nicely into the hands of the consultants and major accounting firms who will happily sell resources to assist companies in meeting that resource gap. But XBRL is not about selling consultant services, except perhaps for some partners in those firms that have touted XBRL for years, invested heavily, and have felt the heat internally to deliver some real revenue for all this investment.

I'm reminded of a great quote, usually unattributed, that we read before every war - "Those that are talking don't know, and those that know aren't talking". The simple fact is that many of the mouthpieces have never actually spend the time required to create an SEC-ready XBRL instance document. They have never been able to take themselves away from their very important meetings to actually spend the 80 - 120 - ? hours involved in creating the XBRL.

Some Recommendations

It is painfully clear that the effort estimates that have provided a backbone for the arguments in favor of XBRL have been wrong. Regardless, I remain convinced that XBRL will deliver benefits across the information supply chain, especially to regulators and investors. I remain convinced that the range of opportunities for XBRL to increase business reporting efficiency and effectiveness are vast and that we have not even begun to think of all the ways XBRL will improve reporting, or the range of applications.

However, there are some lessons and recommendations that flow directly from this latest information:

  • Footnote tagging will not be 200 - 250 hours of effort. That is a bogus number founded on nothing more than the best dreams of the advocates for XBRL. XBRL US and the SEC should run real exercises creating real instance documents, and publish the time and effort required.
  • Companies should significantly increase their estimates for resource requirements.
  • Companies in the "first wave" should start their footnote tagging exercises NOW.
  • Companies coming up on their first XBRL filing should start planning as soon as possible, and should be having "trial runs" well before summer.
  • Mouthpieces of the large firms should reconsider their estimates, and begin to give some honest assessments of the effort.


  1. Lies damn lies and statistics.

    Would not pay too much attention to this.

    It attempts to put some purely subjective numbers on a process, with the pro and anti zealots pushing the bell curve in their favored direction.

    Some musings while I sip my cappuccino while sitting in the conservatory being warmed by the winter sun..

    215 responses out of how many filings?

    How many of the respondents are well versed in keeping accurate times of work done on specific projects and honest enough to strip out the facebook/beebo/private email/twitter/blogging hours?

    How secure do the respondents feel re their job?

    How long did the previous filing system take?

    For the second filing, are we comparing like with like?

    What kind of order, rag or otherwise are the underlying systems in?

    As reference:

    The X-Files is a American cult science fiction television series.... It first aired in September 1993 and ended in May 2002. The show was a hit for the Fox network, and its characters and slogans (e.g., "The Truth Is Out There", "Trust No One", "I Want to Believe") became pop culture touchstones in the 1990s. Seen as a defining series of its era, The X-Files tapped into public mistrust of governments and large institutions, and embraced conspiracy theories and spirituality as it centered on efforts to uncover the existence of extraterrestrial life.


    XBRL is a American cult science fiction reporting system.... It first aired in 199X..... The show was a hit for the big consulting firms , and its characters and slogans (e.g., "The Truth Is Out There", "Trust No One", "I Want to Believe") became filing/reporting culture touchstones ...... Seen as a defining product of its era, XBRL tapped into public mistrust of governments and large institutions, and embraced conspiracy theories....as it centered on efforts to uncover the existence of painless, costless, effortless error free reporting.

  2. Perhaps we need an XBRL taxonomy for reporting XBRL work. At least three concepts could be better defined: 1) "required," 2) "necessary," 3) "actual." Required could mean "what it took to comply with an external mandate" or "what it took to get the job done given one's choices about how to comply" -- it would need to be disambiguated. "Necessary," perhaps should mean, "the minimal amount that could have been done." "Actual," could include "necessary" + voluntary work beyond necessary.
    There might be other elements in an XBRL Use Taxonomy, and they might sum to a value for "actual" that significantly exceeds the values for "necessary" and "required."

    We tried to finish the rule as quickly as possible so that there would not be time for the many interested parties in the financial reporting supply chain -- from accounting and auditing companies to lawyers to software companies to financial printers, to consultants and others -- to develop more elaborate XBRL revenue models and the presentations necessary to convince CFO's to pay more people more money and have people spend more hours on the trivial task of applying data tags to facts in financial statements. Those who might have profited from more time to ramp up revenue models may be disappointed that the rule was finished as quickly as it was. But investors should be happy they're finally getting the benefits of data instead of documents in this one little corner of the capital allocation market. The ROI will increase as the transformation from documents to data spreads across the capital markets as it's done in other markets (music, travel, retailing, etc.).

    A second part of an XBRL Use Taxonomy could include savings -- concepts like:

    External Consultant Displacement: Savings from purchasing competitively priced software and support to do the same things you might pay a third party much more to accomplish.

    The cost/benefit analysis in the SEC rule provides some examples. For full transparency, it would be nice to see not just the debits on an XBRL Data Tagging Chart of Accounts, but the credits as well. Alas, since many of the credits accrue to third parties, that would be a challenging chart to populate with accurate data.

    Just like opacity and uncertainty and fear and not wanting to be left behind as others developed more "profitable" revenue models contributed in various ways to a dot com bubble and a housing bubble and a post-9/11 "security" bubble (one in which we're still living), so too do these factors contribute to an XBRL services bubble. The thing that could make an XBRL services bubble unusual, however, is that it holds a key to its own resolution: less expensive business processes facilitated by the use of open standards, including XBRL itself.

    [End of Part 1 of Comment]

  3. [Part 2 of Comment]

    Surveys might be nice (they might also be self-defeating, deceiving, or downright deceptive – especially non-scientific self-selecting surveys), but show me the facts according to well-defined concepts!

    My bet is EDGR can help companies beat the survey results handily. Regardless, there's an interesting discussion of the competitive landscape, including industry connections that might delay maximum competition, in a Miles Jenning Seeking Alpha post entitled “EDGAR Online's Unique XBRL Database and Analytics Platform,” here:


    Disclosure: I, like the author of that post, am long EDGR. The reason is that EDGR appears to be in the best position to profit from the development of an iTunes ecosystem of corporate disclosure. They probably won't do to accounting firm executives what Apple did to record label executives (cut into the budget for extravagant parties for one thing), but they could be at the heart of a new capital markets information ecosystem.

    One question is whether regulators will listen to the fears of long-established market incumbents or comprehend the transformative effects of moving from document models to data models. But even if regulators are too slow to move toward the future, the trend is powerful enough already to take us there. Document-based disclosure is something the Facebook generation up with will not put. That's also why I'm long BR, which is nice because it will continue to profit from legacy systems, but also understands how to create systems to exploit new technology. They're like an old-fashioned record label, except they're practical enough to understand that moving from CD's to mp3's requires new business models, and that the move to a semantic Web only further increases the pressure to develop new and more efficient business models.