The Power of Digital Assets and Intangibles
Digital Assets And Their Combination With Other Intangibles Will Decide Which Banks Win And Which Lose In The Intangible Economy
In the first of three articles on intangible assets, I highlighted how intangible assets have come to dominate the creation of wealth in our economy, increasing productivity and returns on capital, and thereby deciding who wins and who loses. In a second article, taking the banking sector as an example, I explored how incumbent firms, FinTechs and the global platforms each have very different strengths when it comes to intangible assets. Asymmetric warfare, based on intangible assets, is in full swing to determine who rules the banking sector. Here, again using banking, I examine the role of digital assets in generating revenue and profit, especially in the context of other intangible assets.
In the intangible economy, digital assets are pivotal
In the intangible economy, digital assets play a pivotal role. Information technology (IT) is both a significant intangible asset in its own right and the key connecter of other intangible assets. Inevitably, banks must build their strategies around digital assets, just as in the industrial age firms planned their business around machinery, factories and connections to transport networks. As the banking analysts Cornerstone put it, “Technology costs will move from a tax on the organization to more of an investment in competitiveness” 1. In developing strategies, banks will need to take into account the distinctive characteristics of intangible assets, such as scalability and synergies, and identify how to combine digital assets with other intangibles. The major digital assets that will drive bank revenue and profits, explored in turn in this paper, are:
- Scalable digital operations
- Digital platforms
- Digital assets that can be monetized in their own right
These digital assets combine to support customers and partners, as illustrated below.
Scalable digital operations
Intangible assets (such as data and algorithms) have the potential to scale but banks require a practical means to scale through digital operations in order to maximize this notional value. For FinTechs and challenger banks this is no problem: they are built digitally from the bottom up, so from the start they have scalable software-based operations, with tiny variable costs per transaction. If incumbent banks are to compete in the long term, they need operations and IT systems that can scale to match these ‘digital-native’ businesses.
At present, most incumbent banks have digital transformation plans that move them forward one step at a time from where they are now. This is sensible and pragmatic, but is not sufficient. Banks will not be able to compete without a parallel strategy that works backwards from what their cost structure needs to be. This will be nothing like where they are now, nor even where they expect to be after their current digital transformation plans. Transaction costs may need to be orders of magnitude lower. Another vital consideration is that, owing to concentration effects, the number of banks that reach scale in any given market may be small.
There are four options for achieving scalable digital operations:
- Greenfield. An increasing number of incumbent banks have decided to adopt a greenfield approach by building digital-only or mobile-only banks. For instance, RBS has Bó for retail banking and Mettle for small businesses; BNP Paribas has Hello bank!; CaixaBank imaginBank. In a variant of this strategy, BBVA has invested in challengers such as Atom, a UK retail bank, and Holvi, a Finnish bank for small and medium-sized businesses. The critical manoeuvre in this strategy is making the cut-over from old to new, i.e. bringing its legacy customers and data over, and leveraging existing brands and partner relationships. Otherwise, this strategy amounts to following two to three years after the challengers, without leveraging the bank’s strengths in intangible assets.
- Brownfield. Banks may decide that in some parts of their business they can come close enough to the goal of scalability through a brownfield approach based on simplifying and transforming their current IT. Here, automation across every function will be indispensable as, put simply, people don’t scale. Legacy banking systems will be kept but functionality from these systems will be exposed as Application Programming Interfaces (APIs) and micro-services that individually can scale. In addition, though right now computing is typically less than 5 percent of admin costs and scalability of computing power seems like a ‘nice-to-have’, it will become a ‘must-have’ as banks digitize operations and computing becomes a much higher percentage of costs. After all, this is why Amazon and Google developed their scalable cloud platforms.
- Outsource/insource. Where the bank is not at scale, outsourcing to other service providers will be a compelling option, especially if a function offers little potential to differentiate in the eyes of the customer. Of course, the mirror image of this strategy is to insource additional volumes from outsourcing banks.
- Per-click operations. Many digital businesses have achieved global scale quickly by accessing external services on a per-click model. Uber’s rapid growth was possible because its operations are essentially just a bundling of services sourced from partners on a per-click basis – partner APIs lie behind Uber’s geo-positioning, route calculation, maps, push notifications, payments and receipts. Banks can identify where banking functions and commodity services are available on a per-click basis and incorporate them within their digital operations.
No doubt different strategies will apply to different bank functions and products. One way or another, banks need a five-year plan to ensure that in each part of their business they have scalable digital operations. There is no point in a longer-term planning horizon because no one can see further ahead – and anyway, by then the game will be up.
The platform is rapidly becoming the dominant business model for the 21st century, and so fundamental to any intangible strategy. In addition to driving revenue in their own right, platforms draw in more customers, spin off more data and create new data interfaces – all intangible assets that can be leveraged in other areas.
A platform is essentially a multi-sided marketplace that connects parties on each side, with network effects creating a virtuous circle that attracts ever-more producers and consumers to the platform. Banks can build platforms around areas of banking, such as trade finance, asset management and wealth management. Alternatively, banks can target platforms at particular customer segments, such as small businesses, millennials or high net-worth individuals. A further option is building a platform whose sole role is connectivity, with Bloomberg the posterchild. In any case, the starting point must be total clarity around customer jobs-to-be-done – say, exporting goods or saving for retirement. Ask why a customer would come to the platform? What jobs do they want done? Equally, which partners are required and what is in it for them? A number of banks have set off down the road of building platforms, but too many have viewed the platform as a vehicle simply for distributing existing bank products, as opposed to working back from customer jobs-to-be-done and the partners needed for those jobs. 2
Partnering to tap into external platforms and network effects
Not everyone can be the dominant platform – by definition. Banks should, therefore, consider where it makes sense to adopt the alternative ‘cheap and cheerful’ strategy of accessing the network effects of other platforms, on the principle of ‘If you can’t beat them, join them’. For example, 58 banks across Europe have decided to use Raisin as a distribution platform for savings products in order to access a larger network of customers than is possible via their own channels; and JPMorgan Chase (JPMC) has linked its auto-loans into TrueCar, a market-leading platform for car buyers to identify auto-dealers in their area and compare prices.
Strategies for the global platforms (GAFA and the Chinese platforms)
A further strategic dimension should be assessing opportunities and threats from the global platforms that have become such dominant features in our business landscape. The global platforms might be banks’ competitors, distribution channels or customers – or all three.
- Competitors. In China, Alipay and WeChat have come to dominate certain financial services segments. In Europe, Amazon, Facebook and Google have registered as third parties to aggregate payment data and initiate payments under Payment Services Directive 2. In the USA, Amazon Pay ranks second only to PayPal in online payment solutions. Moreover, Amazon offers businesses fixed and revolving credit lines, invoice generation and pay by invoicing. In response, banks need to identify where their business is vulnerable to the major platforms and develop defensive strategies.
- Distribution channels and interfaces. In an earlier blog, ‘Digital Banking in a Post-App Era’, I explored how the global platforms have become the High Streets of today’s digital world and why it is essential for banks to have a strategy for distribution via these digital High Streets. This strategy will need to include integration with platforms’ intelligent agents such as Siri and Alexa, which are likely to become the standard interface for accessing frequently used digital services. Banks are already experimenting here. For example, UBS has piloted a new service with Echo that lets customers ask financial questions to the device's AI helper Alexa.
- Customers. Banks should look out for revenue opportunities from providing financial services to the platforms themselves and to their customers. For example, Zopa, the UK peer-to-peer lender, has struck a deal with Uber to offer car loans to its drivers.
In summary, as a result of the power of platforms and the network effects that sit behind them, platforms should form a distinct strand within the strategy of each bank business unit, whether the chosen option is to build or to partner. In either case, GAFA, and in some markets the Chinese platforms, will have to feature too because they are dominant in today’s digital landscape.
An algorithm is a set of rules for solving a problem in a finite number of steps. In some ways, the written operational procedures that banks depend on today can be regarded as algorithms because they similarly define a series of steps. Computerization, however, has transformed the ability of banks to deploy algorithms. Computerized algorithms bring greater consistency in decisions, allow much larger volumes of data to be employed and increase the speed of decision-making.
Machine Learning (ML) and Artificial Intelligence (AI) bring a further step-change in the potential to apply algorithms: where hitherto people programmed an algorithm’s rule set, ML and AI models allow computers to derive their own rules and progressively improve decision-making. In future, all the decisions that are fundamental to banking – credit, risk, fraud and investment – will be made (or at least supported) by ML and AI models. To take just one function, lending, algorithms can be used to assess creditworthiness and price credit risk, streamline loan operations, identify potential borrowers, predict loan default, determine loan write-offs and decide treatment of delinquent debtors. In the intangible economy, algorithms will be one of the most valuable assets that banks hold.
Don’t let a thousand flowers bloom: decide which flowers your garden needs
Most large banks are already actively exploring ML and AI, with individual business units generally taking the initiative independently. However, from the viewpoint of the bank as a whole, the ‘let a thousand flowers bloom’ approach makes no sense because there is no resource in shorter supply in the global economy than data scientists. Banks should identify where ML and AI have most potential to create value and focus their data scientists in these areas. This requires decision-making: banks must decide where the ability to make better decisions will have most impact, and recognize domains in which partnering may provide a better option – likely to be areas such as anti-money laundering where aggregation of data and knowledge is vital.
Maximizing productivity of data scientists – the world’s scarcest resource
We are beginning to see interesting initiatives to democratize AI. For instance, Google has released Cloud AutoML to automate building and tweaking neural networks for image recognition so that non-experts can develop their own models. Even so, for the foreseeable future, ML and AI (perhaps like all emerging technologies) remain artisanal in nature. Solutions are hand-crafted. It is all about the workman and his tools. The priority, therefore, must be increasing the productivity of data scientists, providing them with the right tools and – above all – ensuring that time is spent on algorithms, as opposed to sorting out the underlying data or waiting for computing to be provisioned. This requires investment in self-service tools and assets such as data catalogues that explain what the data means and where it comes from. Mundane data management tasks may be outsourced so that data scientists can devote their precious time to the algorithms that will differentiate the bank in the eyes of its customers.
Curation of algorithms
Like any critical asset, algorithms need to be looked after. Leaving it all to the data scientists is not enough once algorithms become a critical basis for revenue generation. Another vital consideration is that algorithms can introduce regulatory and reputational risk. Banks need to curate algorithms in several respects:
- Performance. Managers and auditors, not just model-builders, need to track and understand the performance of algorithms. For instance, banks must know how algorithms will perform in different market conditions. Will they stand up to a crisis?
- Bias. Algorithms are not altogether bias-free. For instance, bias can be introduced in selecting the data used to train a model. Preventing bias is not as straightforward as it might appear. Take credit-risk assessment as an example: the credit score of a borrower appears to be a reasonable variable to use, but in fact credit scores are influenced by past credit decisions, so in practice people cannot get credit because historically they could not get credit.
- Fair use of data. Traditionally, lenders have employed data that relates directly to the likelihood of repayment, such as credit history and debt-to-income ratios. As more and more data becomes available, the issue of fair use of data arises. What would customers think if they found out that their bank was using social media data to assess credit risk?
- Interpretability and explanation. Since the rules of ML and AI systems are not programmed in a traditional sense, they can resemble a ‘black box’ where it is impossible see how a decision has been reached. This raises regulatory issues – for instance, GDPR includes a ‘right of explanation’ for individual citizens to know how decisions about them have been reached. In any case, bank executives will surely require oversight themselves. All this calls for banks to become expert in the emerging field of model explanation, understanding both how a model works in general and how individual decisions were made. Managers will have to decide where to make trade-offs between models’ performance and interpretability.
- Audit. Once bank performance is increasingly determined by its algorithms, audit and risk assessment of algorithms will become fundamental, as will maintaining a register of algorithms.
Monetization of algorithms
Banks will want to capitalize on opportunities to monetize algorithms outside their own operations. After all, if you have a scalable asset, why wouldn’t you want to market its use, not only generating added revenue but also harnessing more data to improve it? A case in point is Metro, the UK challenger bank, which has partnered with Zopa – Metro brings the customer deposits; Zopa brings the algorithms. Similarly, the AI-based lender OakNorth is commercializing its algorithms in countries outside its home UK market.
In order to develop and manage algorithms as a coherent set of corporate assets, a centre-of-excellence model stands out as an obvious approach, especially when it comes to ML and AI. Reasons for this include that the state of the art is not yet mature; expertise is scarce; ML and AI are general purpose technologies (GPTs) with application across the whole bank; and multiple ML and AI models will draw on similar data. Critical too will be measures of model performance and their impact on cost, revenue and profit. These metrics will be prominent in the dials that executives monitor most closely in tracking and predicting bank performance.
Data is first and foremost a revenue-generating asset, not a compliance issue
Most banks have long-running data quality programmes, but compliance is typically their principal driver. Of course, compliance matters, but the importance of algorithms means that banks should think about data first and foremost as a vital asset for revenue generation. In many respects, the data is more valuable than the algorithms. With data you can build algorithms, but algorithms without data are worthless. Google is happy publish its algorithms because it is the only firm that has the data on customer search queries. Few banks can expect to succeed in the intangible economy if they are not masters of their data.
Data value = data x ability to exploit data
As with any asset, you first need to know what you have. Banks should build a map showing what data they hold and its business value – or potential value. I say ‘potential value’ because as with many intangible assets, the value of data is not intrinsic but depends on how or if it is brought into play. In other words, data value = data x ability to exploit data. For most incumbent banks there is a huge ‘data value gap’: the difference between what their data is worth at present and what it would be worth if it were classified, associated with other data and made accessible to those who need it in a timely manner. Data maps, a sort of ‘meta-meta data’, will become a valuable asset in their own right, because without these it is very hard to make strategic decisions about data at an enterprise level.
Closing the data value gap
In large part, closing the data value gap is a matter of improving data quality through traditional disciplines, such as data cleansing and applying meta-data. However, new factors are coming into play as banks extend their use of algorithms and harness new types of data such as unstructured data and ‘big data’ from outside the bank. As an example, for many ML and AI models, where data is stored and how is critical. In addition, as banks hold more and more data, the cost of data storage and management will become a significant concern. Likewise, because the value of much data ages fast – a breaking news story is worth infinitely more than yesterday’s news – access to data in real-time may be important, for example, via data streaming.
This is an area of rapid technology innovation where the received wisdom around how best to do things has yet to evolve. For most banks, unlike (say) manufacturing companies, the challenge is not the volume of data but its inter-relatedness and its timeliness. In the meantime, the challenge is keeping abreast with a flood of new technologies, understanding how and where they fit.
Complex and conflicting regulatory requirements
Data management, however, is not merely a technology issue. Banks face multiple and often conflicting rules around data. For example, a customer’s right to be forgotten under GDPR points towards data deletion, but this may conflict with a bank’s obligations to retain data to detect insider-trading and money-laundering. The key is understanding how to address multiple regulatory requirements, while still being in a position to exploit data to the maximum for revenue generation. Ultimately, this comes down to deciding what data is retained, and if it is retained, putting in place policies and controls to determine who has access (and do they have access to data in its raw form, or data that is aggregated, anonymized data, pseudonymized, etc.).
Unpicking these complex issues calls for close collaboration across multiple technical domains – compliance, risk, legal, business units and IT – especially as this is a multifactorial problem, requiring cost, revenue, risk and technical considerations to be weighed. This collaboration will need to be underpinned by clear governance mechanisms to reach the right decisions (and record why decisions were made). Without this collaboration and supporting governance, bank data remains more a liability than an asset owing to the regulatory and security risks that it brings. It is ironic that years of transaction data across huge numbers of customers and counterparties can be a weakness, while FinTechs and challenger banks have much less data, but because they have been built digitally from the ground up, they are able to harness it. 10 x 0 = 0; 1 x 1 = 1.
Data access and monetization
Once data is classified and made accessible (i.e. turned into an asset) it can be monetized. Whereas traditional management information and business intelligence models involve ‘pushing’ data to consumers of information, maximizing the value of data entails reversing the flow through a ‘pull’ model. ‘Self-service’ becomes the goal, where consumers of data are provided with data, metadata and a set of tools.
There will also be opportunities to monetize data outside the bank’s own operations, for example:
- Data services. Banks can seek to provide data services, both in order to generate revenue and to increase stickiness. In personal banking, customers’ choice of a bank will increasingly be shaped by the tools offered to analyze and advise on spending. For merchants, Wirecard, the German payments provider, has built a service on top of its ePOS solution that takes merchants’ payment data and provides back to them a machine learning solution for analyzing customer value and migration rates3. Data integration is another strategy: Barclays’ DataServices transfers data on payments and cash balances directly into customer accounting systems.
- Revenue from data sales. GDPR and other data regulations notwithstanding, banks will derive revenue from data sales. For example, companies such as Cardlytics provide targeted offers from retailers to bank customers who have opted to receive offers.
- Data creation. The success of the global platforms has derived from their ability to create a new asset from digital exhaust – the data and information left behind when people browse the web. This new asset is monetized through targeted advertisements. Banks will have their own digital exhaust from web, contact centre and location data which, once combined with other data, can serve up immensely valuable sales referrals to merchants. Clearly, as banks seek to monetize this digital exhaust, transparency and maintenance of customer trust will be prerequisites.
- Data aggregation. As the banks’ ability to derive value from data increases, they will be active in aggregating and acquiring additional data, through offering customers added-value services in return for permission to use it and partnering with data vendors who hold complementary data sets. Here 1 + 1 = 3.
The sooner banks start down the road of thinking about their data as a vital corporate asset the better because it is hard to make up lost ground. First, resolving data management issues around years of complex inter-related data takes time. Second, developing algorithms – which is why you want the data – is a learning process that depends on iterations, so it too just takes time.
Most banks require much more impetus here. A few banks have appointed chief data officers (CDOs) to bring more focus around data management. In general, however, their role is more about policies and governance than genuine ownership of corporate data. As firms realize that leaving data management to business units is a recipe for good intentions and not much else, the role of CDO is bound to strengthen. Perhaps the scope of the CDO will be extended to become a Chief Algorithm Officer as the value of data and algorithms are so closely inter-linked.
Digital assets that can be monetized in their own right
Having built digital assets to support their own business, banks may find opportunities to monetize digital assets in their own right, as shown below.
In investment banking, JPMC and Goldman Sachs have decided to make their trading systems, once closely guarded ‘crown jewels’, available to customers. Deutsche Bank has gone a step further and open sourced its trading software, Autobahn, so that customers and partners can not only use the software but also add to the code. When it comes to retail banking, every bank already possesses an intangible asset that is extremely valuable in the digital economy: a secure digital identity coupled with a payment mechanism. This identity and payment asset can be exploited in countless market segments: who needs a way to establish digital identity and take payment? Everyone.
In many cases, the opportunity to monetize digital assets will come through APIs. Capabilities that were developed as part of an overall bank process (such as providing account data or initiating a payment) may be commercialized as stand-alone services via an API. Partners will consume bank APIs on a per-click basis as part of their own distinct customer proposition. As more and more elements of the economy are digitized, there will be an increasing range of opportunities to embed payments and other banking functions within the operations of other sectors. Banks should consider monetization of any functions that they have digitized to support their business – not just banking functions. For example, Know Your Customer (KYC) checks are needed in a range of sectors (accountancy, legal, real estate … ) as a precursor to doing business.
In addition, there will opportunities to create new digital assets that are adjacent to the bank’s business. For instance, Commonwealth Bank of Australia (CBA) has developed an Android-powered console, Albert. This is an EFTPOS tablet which is pre-loaded with a payment app and other apps created by the bank. CBA has fostered an ecosystem of suppliers and has lined up 800 registered developers to create apps specifically for the terminal. Conceptually, this is a merchant equivalent of Amazon’s Alexa in that it is an asset that, while admittedly physical in itself, extends the reach of the bank, giving the bank and partners an extra channel to generate revenue and strengthen ‘stickiness’.
To succeed in the intangible economy, banks will have to put digital assets at the centre of their strategies to drive revenue, profit and customer loyalty. Thinking about scalable digital operations, platforms, data and algorithms as distinct assets will in itself mark a step-change – right now they barely feature, if at all, on bank balance sheets. Human capital and organization capital – people, skills, roles, processes and governance – will all need to evolve in support. Moreover, as with chess pieces, banks will have to learn the moves that are possible with each digital asset and decide how to bring them into play alongside other intangible assets within an overall game strategy.
2. A framework for brownfield firms to map out platform strategies and to anticipate the moves of digital-native competitors is detailed in Bill Murray, ‘Liberating Platform Organizations’, LEF, 2018
3. ‘Digitise Now’, Wirecard Annual Report, 2017