A RESPONSE TO THE REGISTER ON TECH NATION 2016

Over at The Register, Marcus Gibson has written an article about Tech Nation 2016 entitled, Oh TechNation. Britain’s got tech talent. Just not like this – Giving shiny-happy UK ‘digital’ sector survey the digit.

Before we get started, Marcus publishes The Gibson Index, which he describes as ‘the world’s first national database of early stage technology companies’.

Our small but expert inhouse team of BBC-trained researchers are backed up by more than 60 specialist business and technology journalists, investment experts, academics and sector experts, who keep us posted on new SMEs that cross their radar.

Startup Intelligence magazine
Startup Intelligence magazine

Not only does this sound eerily similar to our first incarnation as Startup Intelligence back in 2011 (we received commissions from blue-chip companies to go out and find (for example) the top ten machine-learning companies in Mexico and used journalists on the ground to do this), but it’s clear that Marcus is an expert on the tech SME space and committed to solving the problem of classifying, mapping and understanding it in order to build an accurate and truthful picture of its health.

His ire at the apparent inadequacy of Tech Nation 2016 is well-founded inasmuch as he thinks the methodology is based on SIC codes. Had that been the case then it would likely be wide of the mark in many of its findings. The purpose of this post is, in the spirit of joint enterprise and progress, to set him straight.

The article begins:

It was ironic that this month’s crowing Tech Nation report, which wrongly claimed the UK digital sector “was growing 32 per cent faster than the rest of the economy,” almost coincided with the total collapse of one of the sector’s biggest – and loudest – stars, Powa Technologies.

Compiled by Tech City and NESTA, an innovation “charity” famed for producing endless streams of reports no one reads, the Tech Nation report makes some bold claims for itself. But when those claims are scrutinised they are shown to be riddled with error and exaggeration.

And then on Linkedin:

Last week the mainstream media dutifully lapped up the content of the dreadful ‘Tech Nation’ report, produced by the ‘Likely Lads’ of SME research – Duedil and ‘charity’ NESTA.

Not quite, Marcus – it was us, not Duedil. We were the core data provider to the report, and there were many other contributors.

There are a number of points we consider erroneous but we’ll start with the one that goes to the heart of our methodology.

The report’s findings are based on Standard Industrial Classification codes.

SIC codes played no part in our software’s data-gathering methodology, and our system is what underpins the findings about the size of the digital economy and the permeation of digital tech into traditional industries. The only part they played in the compiling of the report was as one of a number of an administrative datasets used by Nesta to calculate Gross Value Added (this is explained on page 59).

First, Tech Nation claims to be the “most comprehensive analysis of the UK’s digital industries to date,” with “1,800 responses logged from a potential 1.56 million businesses.” From my experience based on talking to 60,000-SME community this is an obvious fallacy – or an incredibly low email return rate.

It is the most comprehensive analysis of the UK’s digital industries to date. Its comprehensiveness doesn’t come from, nor should it be conflated with, an email survey. As the report says, ‘Led by Nesta, this year’s report analyses data from the Office of National Statistics, Burning Glass online job ad database, GitHub, MeetUp users and GrowthIntel’s pioneering website analysis.’

Later Gibson says,

‘A new company name may be recorded at Companies House and gleefully listed by Tech Nation, but [sic] certainly not a new business, contrary to the latter’s misleading claims.’

Companies House was not the decider here. In order to be listed by Tech Nation as a digital business, a legal entity had to meet a minimum threshold based on the presence of related words on its website (the full methodology is on p58 of the pdf report, or pages 114-5 if you’re lucky enough to have a hard copy).

The report itself explains, ‘In this report, we have worked with GrowthIntel to build a more timely and detailed picture of digital tech than official data sources allow. We have used a “big data” method to create a “core business dataset” based on a sector classification scheme developed by GrowthIntel which is independent from SIC codes. This dataset is built using machine learning techniques on data scraped from company websites.’

The article makes further reference to sectors when querying the report’s definition of ‘digital’.

Tech Nation’s definition of “digital” is also awry – but, nevertheless, is all-encompassing. Anything involving a PC is digital. It includes “marketing” and “design”, sectors normally included in business services, as well as games and social research, though its researchers omit to highlight the UK’s often world-leading expertise in building, construction and integration software, such as BIW Technologies, based in Woking, Surrey.

We did not simply subsume marketing and design sectors under an imprecise, all-encompassing definition of digital, triggered by an employee switching on a computer. Tech Nation’s definition of digital is clear and printed on page 4: ‘What is a digital tech business? Business that provides a digital technical service/product/platform/hardware, or heavily relies on it, as its primary revenue source.’ It is not the case, therefore, that ‘anything involving a PC is digital’.

We classified businesses as digital when they met the above criteria using a combination of supervised and unsupervised machine learning. This process created tags that corresponded to businesses’ areas of activity based on the text they product and that others write about them. Of the 500 topics identified, 228 described industrial activity and of those, 49 were digital. Any business assigned at least one digital tag was defined as a digital tech business. That’s where the figure of 58,000 digital tech businesses (out of 320,000 businesses) comes from (again see page 58 for more detail). “Design” may frequently come under business services in the SIC codes but that doesn’t tell us how many digital companies that archaic grouping hides.

Lastly on the above quote, BIW Technologies didn’t feature, possibly because their website is no longer operational, which may have something to do with the fact that they were acquired in 2010 by German company CONJECT.

Where the report completely abandons reality is when it claims that digital tech businesses now make up significant proportions of “traditional industries” – 36 per cent of aerospace and defence, and 32 per cent of electronics industries.

Everyone else may well classify telecoms and networking as part of electronics or engineering but Tech Nation’s definition says a digital firm is a, ‘Business that provides a digital technical service/product/platform/hardware, or heavily relies on it, as its primary revenue source’. If it is the case that companies in nearly all sectors have, over time, come to increasingly rely upon digital services/products/platforms and/or hardware for their primary source of revenue – and the data suggests it is – then we would argue that this is something worth knowing.

Finally on the sector surveys/SIC codes point:

‘…key individuals behind the report display a deep, yet sincere lack of knowledge about the UK’s core IT geography. Not surprisingly, as Tech City leaders all seem to hail from overseas. One might remind them of the forensically comprehensive sector surveys carried out by all Regional Development Agencies, not to mention the DTI’s surveys, in the Noughties – long before the Tech Nation authors had left school.’

We’re not sure what Tech City’s leaders could be expected to do with a pile of surveys carried out years ago by agencies no longer in existence that used traditional, static, and out of date sector analyses. Given the pace of digital economic change it’s unlikely that either the methodologies or findings of any such reports published in the 2000s are of relevance in 2016. Plus, what have the Tech City leader’s nationalities got to do with it? It’s a data-driven report.

A number of separate points are also made. We’d like to challenge those too.

Digital media is not growing.

Far from providing proof that digital media is growing, there is overwhelming evidence that some digital sub-sectors are in decline – notably, the large number utterly dependent on public sector cash, such as councils, schools and arts organisations, some of whose budgets have been halved in the past five years.

The fact that some digital subsectors are in decline, for which no evidence is provided (but is bound to be the case), is not a counterweight to the argument that the digital tech economy, as a whole, is growing – for which we have provided a wealth of data.

Digital tech companies overpay their senior staff and don’t train junior staff.

Tech Nation says the average digital salary is “£50,000, or 36 per cent higher than the national average”. What this actually shows is how little training of junior staff it undertakes, or employs, and how excessively overpaid are many senior staff.

It doesn’t say that £50k is the average digital salary; it says it’s the average advertised digital salary (see the Burning Glass methodology on page 59). Separately it’s hard to see how this is evidence of how little training of junior staff it undertakes or employs, or that senior staff’s wages are inflated.

Armchair extrapolation is a fun game we can all play. For example, we could just as easily argue that an average digital salary of £50k is evidence of the fact that there are probably fewer slave-driving minimum wage positions available in digital tech vs the whole economy, thus the average salary is higher. Who can say?

Tech Nation is reporting on sole traders, not real businesses.

In addition, hundreds of thousands of digital “companies” are not solid trading companies in the traditional sense but mere accounting entities for one man, home-based software developers.

We’re confident this isn’t the case because our software doesn’t track sole traders. Therefore all 58,000 ‘digital’ business noted in Tech Nation are trading companies.

Digital growth should only be applauded when it’s profitable.

Of course, while there is much to applaud among the UK’s innovative firms involved in the digital world, there is also much to fear. The collapse of Powa Technologies, once valued at $2.7bn, is an omen. London payments firm Transferwise, another “unicorn” worth more than $1bn, is profiled in the report. According to the latest reports filed with Companies House, it continues to make losses, as does company information firm Duedil.

Profit is not necessarily the only aim of a company. Obviously big losses are usually bad unless you’re sure you’re setting yourself up for much more revenue later, but many fast-growth companies plough back all the money they make into expansion. Nothing wrong with that. It’s tough to draw big conclusions from anecdotal failures.

We genuinely value the author’s input. His dismissal of traditional business and government sector classification means we’re on the same side. We exist as company to present businesses with a different and more accurate method of understanding the economy (so you can sell into it).

This report has its flaws. Understanding an economy is an incredibly hard problem to solve. GrowthIntel is the first non-state body to build a system to understand the structure of an entire economy. but we didn’t hit publish and dust our hands. This is a long-term mission.

But we are all working towards the same end. Tech Nation is the most comprehensive analysis of the UK’s digital industries to date. The methodologies underpinning the analysis (ours and those of the other contributors) are some of the most advanced in the world. As such, the report’s insights and findings are of enormous value to us all – entrepreneurs, policy-makers and journalists alike.