On the 27th May HMRC emailed out a pair of `tagged` spreadsheets to software vendors to assist with the `accreditation process`. They separately emailed vendors to not-so-gently remind them that `accreditation` is but a `limited recogition process` passing this test does must not “be portrayed by you as recommendation, endorsement, ‘kitemarking’, or approval of your product(s) or service(s) by HMRC”. I wonder which vendor or vendors have been playing fast and loose on that?!
Back to the sample tagging. This is actually a pair of accounts which have been layed out in Excel to show likely accounts entries, and then have columns to the side that show the likely/expected tag that the data item would be tagged against. It is not a `tagged document` nor an iXBRL file.
The first file, AAAAA Limited, shows just short of 800 rows of data in my spreadsheet editor (Open Office, since you ask). It is colour coded by HMRC to make life simpler for the person reviewing the document to match data to tags, and minimum tagging vs full tagging is identified by italics. I thought I wouldgive you an indication of how complex this appears to be, but the task of counting data points / tagging tasks is not as simple as I may have hoped withouth the actual instance document.
Some of the early information, such as company registration number, registered offices, directors & auditors have a one-to-one `mapping` from the data to a taxonomy tag – e.g. company registration number is `tagged` as uk-bus:UKCompaniesHouseRegisteredNumber. But it is never quite that simple. The registered office address is mapped into `tags` for address line 1, town & postcode but then has to have the further `dimensionality` data of uk-bus:EntityContactTypeDimension=uk-bus:RegisteredOffice added to identify which sort of address it is.
I also noticed that skipping fown to the income statement area that whilst we have a two column layout for the P&L and the B/S, the sample `tagging` only indicates the tags, not other attributes of the data. For instance, Cost of Sales in the P&L is shown next to uk-gaap:CostSales, but the need to select Current Period or Prior Period as an attribute is not highlighted in the document.
All these caveats on one side (and I haven’t even gone into the tuple column yet), I can tell you that the total number of `tags` and `dimensions` referred to in the example doc is 266. How does this sit with the HMRC estimate that a manual tagging approach will take `a day or two`?
Well some parts of manual tagging look straight forward enough, so long as your tool has a text search. Work through the directors report and you will find paragraphs headed up `Principal Activities and Business Review` and `Credit Risk` for example. These have to be `tagged` against the taxonomy items uk-bus:DescriptionPrincipalActivities and uk-gaap:DescriptionCreditRiskExposureManagement. A simple tagging like this cannot surely take more than a minute? Highlight the block of text, open a taxonomy browser window, search for the key words or browse through the taxonomy Directors Report secton.
If it was all that straight forward, then 266 minutes would same fair – less than 5 hours and not far off the lower estimate from HMRC. One could even argue that once you get proficient using a tagging tool, that a whole minute may be over generous for the simple tagging of a block of text. However, it is not all that simple.
Take the P&L and B/S. For each of those items, you have to tag 2 data points not 1 – both the current period and the prior period against the appropriate taxonomy item. In AAAAA Ltd, that’s 28 extra data points to add in; another half an hour. And once you continue into the notes section you’ll find plenty more 2 column sets of information that push the tagging total up – by another 105 extra data points, adding the best part of a further 2 hours.
My total is now up to 399 `tagging acts`, and therefore stands at 6 hours 39 minutes by my rule-of-thumb mechanism. One to two day per set of basic accounts does therefore sound like a reasonable estimate. The question is, at what point does the effort and risk become too great & a software solution to either pre-tag or to generate your stat accounts from scratch, become appealing?