Tech and Tagging: What are your ESEF options?

As we move into the final year, depending on coronavirus, of preparation for a new digital era for Annual Financial Reports, so much has changed. Wind the clock back a year and the conversations we’re having with clients regarding ESEF were very different.

They were mainly focused around 3 different options that at the time were the most suitable options:

•  Do a full digital transformation, chose an end-to-end solution for six-figures (Spend £££)
•  Lots of time and money on it, you might get close-ish (Spend ££)
•  You’re probably looking at a compliance-style ‘plain text’ document, that’ll sit alongside your glossy PDF (Spend £)

None of these solutions really solved the client’s problem as it would either be extremely expensive or the output was sub-optimal.

This prompted a debate dubbed “one document or two?”, i.e. the approach firms take will determine if they produce one xHTML and a separate PDF, or just one xHTML. Firms tended to fall across a complete spectrum from “it’s just compliance, I don’t care what it looks like” to “it must look identical”. It even prompted some firms (an example being Aviva) to scale-back their Annual Report “look and feel”, to more easily accommodate the requirements.

Since then, the debate has been answered by both a significant improvement in the technology available in the market, as well as a firming up of the stance taken by communications agencies and their clients.

In short, the PDF is going nowhere. For one thing, xHTML doesn’t print well, and glossy accounts must be printable. Secondly, what harm does producing a PDF do? Why wouldn’t you produce a PDF? It’s a standard output from InDesign, and InDesign is the tool of choice for most of the industry, and that’s not going to change in a hurry.

Just as importantly, xHTML can now be made to look very, very PDF-like. Some seriously impressive work by vendors in the tech industry means that you can take a PDF as an input format, tag it and produce excellent quality xHTML / iXBRL that mirrors it incredibly well. There is at least one vendor, probably more, that can do this at a very reasonable price, quickly and to a great standard. We’ve tested around 50 PDFs and the process works well. In short, the technology challenge for ESEF has been solved, and the price-point that goes with it is fairly nominal, if you know where to look.

Technical point – PDF absolutely is the input format of choice when outsourcing. ePubs, which can be taken directly from InDesign too, can also be used, but bring up a number of issues with fonts that we won’t go into, but are painful. PDFs contain all the fonts you need, ePubs don’t, so PDFs tend to win as the input format of choice.

It’s clear from marketing efforts that tech vendors are gearing up for a technology licensing battle on the ESEF front. There’s a relatively short window for firms to buy something (though the expectation of a 12-month delay is entirely reasonable given COVID-19), after all. Testing and pre-tagging needs to happen well in advance of year-end. However, the technology side of things is only half of the problem. And this is where outsourcing vendors come in.

All of the technology out there will enable firms to be able to take their documents, create xHTML with iXBRL embedded, and create their report in accordance with the ESEF mandate. Crucially, however, what that omits is that the iXBRL has to be correct. None of the tech vendors out there can actually do the tagging of the documents, that requires human input.

Here are some fun claims we’ve heard:

Claim 1: “We can auto-tag the documents”

We can give you several reasons why that is very, very wrong. Some think it’s a case of just looking to the row label for a table and fuzzy matching to the ESEF taxonomy – that simply isn’t the case. Here’s one example – “Other financial assets” in a Balance Sheet – would you use the tag IFRS:OtherFinancialAssets? You’d think so, but, often not.

There are also basic trip-ups such as scales, decimals, dates etc that aren’t always entirely obvious from row labels (think the Cashflow Statement), as well as any figures in Primary Statement footnotes, and mandatory tags such as LEI which mean auto-tagging for ESEF isn’t viable.

We’ve tested multiple vendor outputs and the automatic tags simply aren’t good enough. And, even if they were good, let’s say around 90% correct (they aren’t), ESEF requires 100% accurate tagging. So, auto-tagging isn’t going to cut the mustard, not for a good while yet at least.

Claim 2: “Use auto-tagging as a start point and change what needs changing”

In our opinion, this is just as bad as trying to auto-tag and “get away with it”. It’s a classic case of ‘confirmation bias’ – i.e. if you see something, you’re naturally biased to follow its lead, as opposed to questioning it. Same example above – if it’s been auto-tagged as “OtherFinancialAssets”, you’re likely to confirm it as true, than question “actually, that’s in the Non-Current Assets part of the table, so that tag is wrong”.

Whilst this is just one example of a common mis-step, there are many, many more. Arkk’s Belfast team of expert taggers are documenting them as they go through testing.

So properly trained humans need to tag ESEF documents. There is no way around this. You can’t just say “spend a couple of days doing a bit of tagging and you’ll learn it”, the tagging must be absolutely correct, and the nuances of tagging are more involved than that.

If you’re a client looking to solve the ESEF problem, you need to work out if you want to procure technology or outsource it. If you go down the technology route (many will), make sure that you do so for the right business reasons. Sensible reasons would be things such as “control”, “automation” or “doing it all in one place”. Security is somewhat a given with any vendor in this space but check that too.

However, remember that if you do choose technology, you’ll also need to pay externally to either do the tagging or train your team to tag, both of which incur costs. Tagging a document under the ESEF taxonomy is not entirely straightforward, and clients won’t reasonably be able to do it all themselves, without external help, particularly in the first year.

So, that leaves firms with a choice. Outsource the whole process or licence some technology and still outsource part of it. I have no doubt that across the 800-or-so UK firms, and another 4,200-or-so EU firms on top, there will be all sorts of approaches. If we look back to HMRC tagging in 2010, initially many firms opted to licence technology for their iXBRL tagging.

Nowadays, less than 95% of the market outsources it. If you’d like to explore your ESEF tagging options further, take a look at our ESEF page.