The end of the fintech gold rush?

Why did so many new fintech start-ups fail, and how can ‘fintech 2.0’ succeed, asks consultant

Palo Alto, California

Anyone reading the industry and even mainstream press might get the impression that banks and other financial services firms are being threatened, or at least transformed, by a continuing wave of fintech disruptors. The trace of this on the collective imagination is charted by Google Trends (see figure 1).

 

Over the past four or five years, it seems, there has been an inexorable rise in curiosity about fintechs. This is supported by evidence of investors’ ever-growing appetite for the fintech scene. Yet a different chart tells a very different story (see figure 2).

Figure 2 shows the number of fintechs that were founded each year, and comes from the database of The Disruption House (TDH), a firm specialising in the analysis of fintechs in wholesale markets. This chart shows two striking features: first, the fintech boom started at around the time of the financial crisis with a big kick-up commencing after 2009, some years before the term ‘fintech’ blossomed on Google Trends; and, second, the number of firms being launched began falling at the very time – around 2015 – that the term ‘fintech’ started to gain popular momentum.

Why is this? It is natural to question the reliability of the data. Whereas the data behind Google Trends becomes available globally, automatically and instantly as people do searches, the data from TDH (or any similar database) requires human collection and curation. This could potentially lead to a regional bias (TDH is based in London). However, an analysis of the same data by location shows a globally consistent picture, with the drop-off in companies based in the US (the largest region) occurring even sooner than in UK fintechs (see figure 3).

 

It’s also possible there is a ‘discovery’ period before new fintechs appear on the radar. While this may tend to slightly exaggerate the decline, the severity of the drop-off and its continuance into 2019 (not shown in figure 3) mean an ‘appearance lag’ is unlikely to fully explain a phenomenon that is also anecdotally well recognised.

Non-retail fintechs

A more compelling explanation for the disparity between the two charts can be found by unpacking the term ‘fintech’. We can identify three broad groups:

  1. Established companies looking to break into financial services, such as Tesco (with Tesco Bank) and Facebook (with libra).
  2. New fintechs, whether with a financial or technology emphasis, focusing on retail, lending and payments, such as Affirm, Currencycloud, Monzo, Nutmeg and Revolut.
  3. New fintechs focusing on wholesale markets.

The buzz reflected in the Google Trends chart (figure 1) largely stems from category two. However, readers of Risk.net are typically more interested in category three, and it is exclusively this category that is considered in this article. Does the fact that these fintechs in the wholesale markets are described by the same term as the glossily successful retail fintechs mask their effective death? If not, what is the dynamic behind fintechs in wholesale markets? What, if anything, did the creation of thousands of such fintechs in the post-crisis years actually achieve?

Transparency

Before attempting to answer these questions, let me be transparent about my own professional interests. First, I am not a journalist, and this article is effectively an extended opinion piece. Second, I came across the data that motivates this article during work for TDH from which the data is sourced. With the exception of Usha Back, principal research consultant at TDH, no staff from firms for which I consult had prior sight of this article, let alone any part in its writing. However, companies I consult for also engage with some of the firms cited. I should therefore, perhaps, warn readers to watch out for consequential unconscious bias.

 

We first turn to a technology firm serving the capital markets that was founded at the peak of the curve and is alive and thriving. Beacon Platform was co-founded in 2014 by Mark Higgins and Kirat Singh, who had acquired deep experience in risk management technology during careers at Bank of America Merrill Lynch, Goldman Sachs and JP Morgan. Beacon now has around 20 clients, and 70 staff across its offices in London, New York and Tokyo. Last year, it raised $19 million in Series A funding from investors including Barclays and Pimco. The firm won Market Risk Management Product of the Year Award in Risk Markets Technology Awards 2019.

Higgins argues that several factors favouring a start-up in the post-crisis years simply did not exist historically. In the past, banks such as those for which Higgins and Singh worked felt compelled to write their own foundational technology, such as SecDB at Goldman Sachs. More recently, Higgins notes that, at JP Morgan, “we wrote our own database for Athena [the bank’s front-office risk platform]. At Beacon, we use MongoDB and Python, and there are many, many open-source Python packages for all kinds of functionality”. The appearance of excellent and freely available technology has dramatically improved the viability of a new fintech.

And it’s not just the programming tools. Higgins also emphasises the importance of “the commoditisation of everything you need to be a start-up” from cheap office space from the likes of WeWork and Regus to affordable cloud-based software for accounting, human resources and legal infrastructure – and, of course, cloud servers for development and delivery.

The loss of confidence, the hard-nosed reassessment of all-in profitability and the breakdown in morale at banks from 2008 onwards coincided with a dramatically reduced cost of entry for tech start-ups. It is not therefore hard to explain the brain drain from the banking to the fintech sector in those years. But why did it stop?

Conventional wisdom states that nine out of 10 start-ups fail. Based on our analysis, the best estimate for the five-year failure rate ... is approximately 60–65%

Usha Back, The Disruption House

One essential data point that’s surprisingly hard to assess reliably is the success rate of the massive number of fintechs founded since 2009. TDH’s Back summarises what is known: “Conventional wisdom states that nine out of 10 start-ups fail. A number of articles have analysed the start-up failure rate, often with conflicting data and with different definitions of failure, and the reported range varies between 25% and 90%. Based on our analysis, the best estimate for the five-year failure rate – taken to mean a firm is no longer operational – is approximately 60–65%.”

In summary, no more than one-third of fintechs founded towards the start of the post-crisis boom are likely to still be going. Michael Elanjian, head of digital strategy and fintech at JP Morgan, has looked at many of them, and says: “There was a rush from 2010 to 2014. The world was changing rapidly, with pending regulatory requirements having significant market structure implications. What we’re seeing now is a natural maturation and a resultant drop in the number of companies whose product offerings and business models have succeeded.”

To use a term from the life sciences, the post-crisis boom years were a period of punctuated equilibrium in the evolution of software. In a rapidly changing environment, a large number of experimental forms emerged with exceptional speed. Most of these died out quickly as the pace of change in the banking landscape reverted to stasis. The mechanism causing most of the new firms to fail, and eventually checking the creation of further start-ups, could theoretically be the occupation of every new niche by the most successful of the new firms. But it doesn’t feel like that – or, at least, not just that. Rather, the environment in which the plethora of start-ups hoped to thrive proved intrinsically hostile to new life forms.

I wrote about the factors that weigh against fintechs in the capital markets in early 2017, and Risk.net’s journalists have subsequently explored the challenges facing start-ups.

jp-morgan

Beacon’s Higgins confirms one of the themes from those articles: “Selling to banks can be challenging for start-ups. The tickets are ultimately large, but the sales cycle can be long, and banks’ procurement and legal processes have not been traditionally designed to work with small, resource-constrained companies.”

That challenge is attested to by two firms that ultimately managed to get through it, both on JP Morgan’s In-Residence programme.

Access Fintech is a successful start-up founded at around the peak of the fintech wave in 2016. Co-founder Roy Saadon, a second-time entrepreneur who previously had success with Traiana, says that “being a 25-year-old with great technology selling to banks is extremely hard”: “A classic blunder is that after some very positive meetings with a managing director, you’re expecting to get a deal. But, in reality, you’re 18 to 24 months away from seeing dollars.”

Matt Hodgson founded Mosaic Smart Data after a career spanning more than 20 years in rates trading and sales at Citigroup, Deutsche Bank and Salomon Brothers. He says “the sales process has been much longer and much harder than I ever imagined”. Like Saadon, Hodgson has found it can take up to 18 months to close a deal, with up to a year spent in the procurement process. To stay alive long enough to succeed, “you need deep, deep resilience – but also sufficient funding to see you through”.

Influential figures within the large banks are well aware of this, and are trying to do something about it. At Barclays, group chief technology officer and chief innovation officer John Stecher recognises banks “need to simplify procurement”: “In particular, we need to differentiate between domains where there is genuine data sensitivity and domains where there shouldn’t be any issues, such as managing logs.”

For the latter, he has a target of two weeks for signing non-disclosure agreements, sharing whatever is needed with the fintech and getting up and running, but adds: “Currently, we’re at innings three of nine in simplifying our engagement model with vendors.”

Top-five fintech picks

John Stecher, group chief technology officer and chief innovation officer at Barclays, gives his personal pick of five fintechs for banking:

Jumio – know-your-customer onboarding.

Flux – digital receipts and loyalty rewards.

Simudyne – agent-based modelling.

Snowflake – cloud data warehouse.

Flybits – contextual recommendation engine for hyperpersonalised experiences.

JP Morgan aims to ease the difficulty of adoption through steps such as isolating pre-production environments in which fintech software can be tested and accepted. JP Morgan’s Elanjian says there are some intrinsic barriers to success: “To do things successfully in the wholesale banking environment, being enterprise-ready is table stakes. Ensuring young companies are prepared for mass distribution at scale is something we spend a lot of time on.”

There is another hurdle beyond the sales cycle for fintechs that target banks as clients: the difficulty that banks have in integrating anyone else’s software. As Access Fintech’s Saadon observes, “fintechs see the $100,000 of value they can bring to a bank, but not the bank’s $1 million cost of change”.

The exceptional riskiness and high cost of adoption is partly due to the massive accumulation of interconnected software that comprises a bank’s in-house stack. As far as IT goes, banks are enmeshed in their own history.

Many of them are trying to disentangle themselves. At JP Morgan, Elanjian says “the fintech In-Residence programme is part of a wider strategic goal of increasing agility and nimbleness, of accelerating”.

A Barclays bank

The reason that increasing agility is such a large focus for the major banks is that it’s so hard to pull off, given the software and data estates banks find themselves with. Barclays’ Stecher notes: “For a bank like Monzo, they’re going from zero to one, which is easy. Established banks have to go the other way.” He adds that “starting with a lot of intellectual property that’s all proprietary, the challenge is to identify what is actually valuable” and to gain agility while preserving that.

Cloud migration offers a useful case study. The availability of cheap, elastic resources offered by Amazon Web Services (AWS) and, more recently, other cloud providers was a central feature in the fintech boom. While virtually every fintech uses cloud software and most offer cloud-based solutions, Beacon’s Higgins is one of several sources observing that “many of the large banks are still in early days for production use of cloud computing”: “Bank cloud environments are often conservatively configured, with many standard services turned off by default while they get to grips with the new security models.”

For Access Fintech, whose business critically depends upon banks mutualising data on the cloud, this is a battle that has had to be fought and won. Saadon says: “Banks now have a cloud strategy, which is a big help.”

At the heart of the matter, banks demand the assurance from their cloud-based software providers that personal citizens have never had from the tech giants: “Can I push a button and immediately delete all of my data?”

You can’t [implement the data blueprint] with three guys in a garage – disaster recovery is the last thing you’re going to work on – but you absolutely need it to reach the $20 million club

Roy Saadon, Access Fintech

Saadon says “there’s a blueprint around encryption, segregation and recovery of data” that achieves this, and there is a level at which implementing this blueprint becomes essential for a fintech.

Looking at this as a stage in the fintech maturity journey, he says: “The club of firms with $100,000 of investment is massive. The $20 million club, where you reach escape velocity, is tiny. The $100 million club is smaller still. You can’t [implement the data blueprint] with three guys in a garage – disaster recovery is the last thing you’re going to work on – but you absolutely need it to reach the $20 million club.”

JP Morgan’s Elanjian echoes the sentiment: “Companies that understand the bank’s concerns around data ownership, rights, governance and security, and who fit in with the bank’s policies, can have a relatively easy time.”

Conversely, it is evident that fintechs may have fantastic technology, but if they try to change the banks’ policies, they get nowhere.

Neatly summarising the past decade of the fintech gold rush, Access Fintech’s Saadon says: “Fintech 1.0 tried to disrupt things, and that didn’t work. Now, fintech 2.0 sees the banks are still here with their clients, their cash, their inefficiencies and a desire to change. It took banks a long time to reach this point, and they’ve been pushed to it by firms like Amazon, Google and Tencent, rather than [by] the financial crisis. They know they need to look more like those firms, and fintech 2.0 will be companies the banks can partner with to provide new kinds of self-service and collaboration capabilities.”

The buy side

Of course, big banks are not the only possible customer. If history is a guide, we’ll find these new start-ups will, as with established vendors before them, settle into making a steadier living from smaller regional banks and the buy side. The glory and the reference value may be less than from a marquee sale to a top-10 investment bank, but the sales cycle is much shorter and implementations are much easier.

As Simon Lumsdon, head of technology at Hermes Investment Management says, buy-side firms are more comfortable with and proficient at licensing third-party software because, unlike the large banks, “we can’t afford to develop everything ourselves”.

Is investment banking disruption-proof?

This article breaks fintechs into three categories: established companies looking to break into financial services, such as Tesco (with Tesco Bank) and Facebook (with libra); new fintechs, whether with a financial or technology emphasis, focusing on retail, lending and payments, such as Affirm, Currencycloud, Monzo, Nutmeg and Revolut; and new fintechs focusing on wholesale markets.

The first two of these are indisputably game-changing. When a High Street supermarket first gained a banking licence and offered current accounts to its customers, something new had happened. If a social network with over 2 billion customers succeeds in launching a global currency, then it promises – or threatens – to be even more disruptive (which helps explain the headwinds currently facing Facebook’s project). Furthermore, new challenger banks and lending companies have been launched in the digital age, and are providing highly automated app-based products run on a low-cost base that are stealing business away from incumbents. The economics of these new entrants are largely lubricated by the monetisation of client data.

Investors and clients are now looking for signs of a read-across from the genuine disruption in retail financial services to investment banking, and Michael Elanjian, JP Morgan’s head of digital strategy and fintech, believes the bank is well placed to provide it.

JP Morgan and its peers ostensibly have data riches, too – not just pertaining to clients, but also to the markets in which they operate. And market data is probably easier for a tightly regulated entity to leverage commercially. This isn’t new. JP Morgan itself used market data and in-house expertise in risk management to launch RiskMetrics 27 years ago. If the bank or its peers pull off something akin to this again, it will be notable – but probably not a disruption on the scale of what has happened in retail. Significant disruption, almost by definition, is more likely to come from without than within.

If the concept of a “challenger investment bank” isn’t on the horizon, perhaps the external impetus will come from fintechs serving the sector? Or perhaps not. Surveying the fintechs that have tried to make waves in investment banking, Mosaic Smart Data founder and chief executive officer Matt Hodgson identifies two different sources: “On the one hand, there are pure tech firms, often from Silicon Valley. And on the other, there are firms founded by business leaders with specific domain expertise who can identify precise pain points. The pure tech firms have struggled to gain traction, but some of the firms with domain knowledge have come through with cloud-based solutions that offer immediate value.”

Of several essential qualities, Hodgson believes the most important critical success factor shared by winning fintechs is timing: “When is the market right for these companies?” Fintechs with a deep understanding of their clients who bring the right resources and domain expertise to bear at the right time can do well. As Barclays group chief technology officer and chief innovation officer John Stecher says, to succeed with clients such as Barclays, fintechs – unlike firms that work in retail – “need lots of capital to scale globally across different regulatory regimes, and lots of industry-specific knowledge”.

In this widely shared view, it is dauntingly difficult for an outside entrant to make an impact because, unlike in retail, the arcana of wholesale markets require such a long induction. If a decade or three in investment banking is the sine qua non for success in capital markets technology, perhaps investment banking really is disruption-proof.

For firms such as Hermes, the explosion of new fintechs has been an unalloyed blessing. While established vendors have a natural incentive to run their old products as cash cows, rather than to invest in new ones, “fintechs often do one thing very well using current, cloud-based technology”.

This has enormously enriched the software supply in areas such as trade surveillance, where companies such as Red Deer, founded in 2013, and Eventus Systems, founded in 2015 and winner of Market Surveillance Product of the Year in Risk Technology Awards 2019, now compete against older firms such as Apama, Digital Reasoning and Nasdaq (with Smarts).

The fact that these new fintechs are able to leverage the burgeoning range of open-source libraries is also risk-reducing.

“Using freely available capabilities like Amazon voice recognition is far less likely to result in undetected bugs than developing a proprietary solution, as vendors would historically have needed to,” says Lumsdon of Hermes. “And utilising Azure or AWS also significantly reduces the risks associated with running on an outmoded platform.”

Furthermore, the relatively small and well-connected network of buy-side chief technology officers confers a sense of informal group wisdom about which new companies appear the most credible.

This is good for the fintechs, too. The buy side engages in less hand-wringing before getting a contract approved. For example, on cloud security, Lumsdon will be looking for a fintech’s ISO 27001 conformance certificate, rather than engaging in a multi-year programme of in-house standard-setting. The upshot of this more rehearsed approach to software licensing is, he claims, that deals can get closed in three months.

Fintechs, but not as we know them

TDH’s database lists well over 2,000 fintechs serving wholesale markets that have been founded since 2009. It is hard to know whether the five-year survival rate is the widely quoted 10% or principal research consultant Back’s more conservative estimate of one in three. Direct experience tells us a thousand flowers did not bloom. On top of the risk of starting up any new business, the peculiar difficulty of selling and delivering to investment banks has winnowed out the majority. The impetus to create new fintechs for this segment is, for the moment, largely exhausted.

While it is cheaper than ever to bring high-quality, richly functional software to market, the fintechs that are flourishing are those with the deepest reserves of resilience, expertise, cash and luck. Whether it is even helpful to call companies serving this sector ‘fintechs’ is moot. Their relationship to retail fintechs, some of which have profoundly reshaped their market, is perhaps redolent of the banks’ colourful application of the term ‘robotics’ to simple process automation tasks such as password resets. This is not to disparage the fintechs in the capital markets: over the years preceding the fintech boom, the vendor marketplace was becoming ever more concentrated and ever more tired, and the firms founded in the post-crisis years have given it a much-needed jolt of new life.

But, unlike their peers in retail, the current generation of capital market fintechs has not disrupted its clients. Instead, the small minority that are succeeding are helping incumbent banks find their way along a hard, slow and necessary path of structural cost reduction.

Buzzword appeal

Software companies working in financial services have always had to deliver solutions that meet the specific functional needs of their clients. However, they have often led on – and got to the table with – claims that they offered breakthrough technology. The mystique of the buzzword du jour has long been used to pique the interest of sophisticated clients that otherwise may tend to doubt the value that an outside firm can offer.

In the fintech era, the one technology that is ubiquitous among start-ups is the public cloud. Beacon Platform co-founder and chief operating officer Mark Higgins identifies cloud leverage as a factor that helps firms such as his displace established vendors that struggle to migrate from their pre-cloud platforms. Higgins is one of those who’s sceptical of the current extent of cloud use at the banks. But even while banks may be slow to realise the cost savings attainable from the cloud they are committed to the journey. Events such as the recent Amazon Web Services data breach at Capital One are unlikely to derail that. Access Fintech co-founder Roy Saadon says of the breach: “The weaknesses, such as leaky S3 buckets, were known before it happened.”

Barclays’ Stecher goes further. He says the bank “conducted a deep dive, and determined it was pretty nefarious, but Barclays’ control regime would have protected us against that attack. Capital One’s response was mega-impressive in terms of the speed with which they identified, announced and addressed the issues” compared with similar historical breaches at on-premises data centres. Paradoxically, “it gives you comfort”, he concludes.

Alongside the cloud, the technologies that have perhaps been most associated with fintechs are artificial intelligence (AI) – taken to encompass machine learning (ML) and natural language processing (NLP) – and blockchain, or distributed ledger technology (DLT). Returning to TDH’s database, we can chart the ubiquity of these technologies as a function of the year in which fintechs were founded.

Figure 4 shows a marked rise in both technologies over the course of the fintech ‘gold rush’ years, with AI being cited as a capability by over half of the firms founded in 2018 and DLT by over 20%.

In research for this article, I found software firms that have traction with investment banking clients are not necessarily choosing to use blockchain where it might technically be applicable. Also, situations where there is a real need for ML expertise still tend to be infrequent, and, when needed, solutions are often achieved through expert partnerships. For Beacon Platform, this centres on research into deep hedging with JP Morgan – the idea that offsets for options positions and other derivatives can be calculated more accurately by using large, real-world datasets than through traditional modelling.

 

Access Fintech’s Saadon sees AI as having the promise of strategic value: “For us, AI addresses a real problem of creating scale for repeat processes through pattern detection, anomaly detection and correlation-based alerting.”

In contrast, he regards blockchain as “a trend whose brightest impact is forcing a conversation on topics that were taboo, but that hasn’t yet driven any real problem-solving”.

At Barclays, Stecher is also reluctant to place too high a value on the intrinsic magic of any particular new technology: “AI is oversold. Most developers haven’t done it, but that doesn’t matter. Our most useful asset remains the transportable ‘full stack’ developer who gets the overall architecture and can be plugged into a wide range of projects. Niche skills such as ML or Cuda [Compute Unified Device Architecture] are much easier to hire for as and when we need them.”

Beyond new technologies, another paradigm shift often associated with tech start-ups of the past five or 10 years has been the pattern of generating revenue from services, rather than chunky software licences, and often making the underlying software available as open source on GitHub. Beacon’s Higgins says this “must be tough”, and prefers the difficulty of the long sales cycle that ultimately promises the reward of big-ticket sales.

Given the lack of any categorical difference between new firms such as Mosaic Smart Data and Beacon and well-established software houses, in terms of the business model, the technology used or the kinds of solution offered, one might question whether, in wholesale markets, the word ‘fintech’ really has any meaning.

For Higgins, a fintech is literally a firm offering financial technology, and so encompasses firms such as Murex or FIS. Access Fintech’s Saadon prefers to use fintech to refer to cloud-based companies offering solutions for cost mutualisation and network plays. More pragmatically, Matt Hodgson, Mosaic Smart Data founder and chief executive officer, says, perhaps without a frenzy of enthusiasm, “‘fintech’ is such as a well-established term that it would be difficult to get rid of”.

Ian Green consults on software sourcing, data analysis and algorithmic methods as applied in electronic markets and also endurance athletics. Earlier in his career, he spent 15 years at Credit Suisse, latterly as the bank’s global head of fixed income e-commerce.

  • LinkedIn  
  • Save this article
  • Print this page  

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact [email protected] or view our subscription options here: http://subscriptions.risk.net/subscribe

You are currently unable to copy this content. Please contact [email protected] to find out more.

You need to sign in to use this feature. If you don’t have a Risk.net account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here: