Why isn't Banking more Boring? Part Six

In the final instalment of the series we examine the choices open to an existing bank to respond to the industry changes we discussed in Part Five. We assume that the likelihood of an institution purposely setting out on our “Musk-like” transformation - to create The Boring Company of Banking - is low and so we examine the alternative paths available. Whether or not these are truly less difficult - and more importantly whether or not they establish a position for longer-term success - will then be a matter for debate.

Shuffling boxes around

To structure the discussion it is helpful to have a framework to describe the choices an existing bank can make. I like Ben Robinson's excellent four banking business models for the digital age, which in my view gives five distinct possibilities

1. From Universal Bank To Universal Bank but a simpler, more cost efficient and more personalised one
2. From Universal Bank To Universal Bank (as in 1) plus a Partial Open Platform/ Curated Ecosystem
3. From Universal Bank To Thin, Open Platform
4. From Universal Bank To Aggregator
5. From Universal Bank To Infrastructure Provider.

These sit on a continuum of difficulty. The first two require only operating model change, together with the investment of time and money. To choose to pursue these possibilities is largely a matter for the executive team alone. The others require change to both the business model and operating model simultaneously (to greater and lesser extents). As a result they are more difficult choices to pursue and will likely require agreement from a wide set of stakeholders including investors and shareholders, regulators, communities and customers.

That is not to say that 1 and 2 will be easy choices to execute. I just consider them more likely choices, given the fundamental changes required to execute 3, 4 and 5. As a result we will examine the first two possibilities in some depth, with a lighter examination of the final three. However, as we will see as the discussion unfolds, the real point is not the choice that is taken now. It is whether or not that choice generates options for responding to what might come next. 

Continue as a universal bank but a simpler, more cost efficient and more personalised one

We can describe this as the “less people, less buildings” transformation. Banks adopting this approach over the next three years will apply one or more of the following levers

• Further reduction or reconfiguration of branch networks in line with changing customer demand and behaviour, supported by campaigns on digital education and advocacy directed to both customers and employees
• Application of machine learning, cognitive computing and robotic process automation (RPA) to reduce operational headcount. Largely this will target the back-office (customer & product opening and servicing) and enterprise functions (reporting, finance, reconciliations)
• Application of machine learning to improve business performance of decision functions (pricing, next best action, risk, identity, fraud) and use of cognitive computing, chatbots and other engagement technology to reduce front-office headcount (telephony, multi-media centres) and improve customer experience
• Application of cloud and devops practices to reduce (or release) investment spend and change capacity (change-the-bank cost), with likely some modest reduction in run-the-bank costs also
• Continued application of sourcing and benchmarking approaches to reduce spend on major 3rd-party contracts for property, operations and IT
• Structural changes to organisation design to de-layer or simplify management and leadership structures, together with potential business policy and product simplification
• Targeted operating model and IT platform transformation, although limited to specific areas where the business case can be relatively easily made and realised (not core system replacement)
• Introduction of platforms for data and API enablement (if not already in-train) to support the application of machine learning and cognitive technologies. These will also support regulatory initiatives like Open Banking and GDPR, as well as revenue-generating/ retaining initiatives focussed on personalisation and customer engagement.

Virtually all banks will undertake some combination of these in the next three years. The only material variations will be the scale of transformation being sought and what is actually achieved.

Imagination is boundless, but the world of reality has its limits

There is no such thing as a typical bank. So it is very hard to describe "in general" the scale of transformation that can be sought and what might be achieved through this approach, as one bank's material restructure can be another bank's light adjustment. However we can ground our discussion by considering a "model bank" that is

• Primarily focussed on lending to households and businesses
• Largely avoids market-based activities such as investment banking or wealth management
• Based in Germany, France, UK, Canada, Australia or another G20 member country
• Serves that country almost entirely, perhaps with some smaller presence in adjacent regions.

With these qualifications our model bank will have somewhere between 40,000 and 70,000 full-time equivalent staff (FTE), have a branch network of between 1,000 and 2,500 branches, and rely on an IT landscape composed of between 1,000 and 2,500 applications. To be specific we will crystallise our model bank still further and give it 45,000 FTE, 1,500 branches and 1,500 IT applications.

Although our model bank may appear very unlike an all-encompassing universal bank (the base for each transformation), there are two reasons why an institution like this makes a good case to examine

• The size and shape is relatively easy to describe and understand, and whilst big banks can vary significantly in their make-up it is broadly representative of many industry participants
• There is some evidence to suggest that banks who diverge away from this model exhibit diseconomies of scope, i.e. average costs increase as they diversify outside of core banking services. This reasons for this are important and we will return to them later.

Having established the size and shape of our model bank, what is the limit of transformation that can be achieved through this route? Let us consider an extreme view. Imagine that customer behaviour, executive direction and stakeholder agreement is such that the bank can transform to an almost pure online (digital) bank within three years. Assume also that the investment and financial ability to restructure is available to achieve this. What would the result look like?

• The buildings part is easy - zero branches.
• What about the people? Can we imagine this reducing to the 1,000 to 2,000 FTE we might imagine a ground-up engineered digital-native bank would need for the same volume of customer and market activity?
• And the IT estate? Can we imagine it being materially simpler having layered on RPA-based processes, added machine learning applications and introduced new cognitive technologies. Will it resemble the essential platform architecture we described in Part Three and Part Four?

It is, of course, impossible to know the answers to these questions. And as far as I am aware no bank has undertaken a transformation this significant, so there is no prior art. However I think it is reasonable to expect that there will be a "floor" on what can be achieved through this form of transformation

• The application of machine learning algorithms, RPA and cognitive technologies will certainly reduce the number of people and therefore cost associated with managing today's existing accidental complexity. However they cannot, by definition, remove it
• The need to integrate these new technologies with existing systems and data sources will likely introduce new accidental complexity and add to the IT landscape rather than simplify it
• Traditional technology decision-making will continue to put a lot of square pegs in round holes. They might be machine learning-/ cognitive-/ RPA-based square pegs, but they will be square-pegs nonetheless
• Despite what any mathematics undergraduate can tell you - that you have to optimise globally rather than locally to find true optimums - the most likely application of the new 10x technologies will be to optimise local tasks, activities and decisions within existing business areas, departments and operational teams
• The skills to apply these new technologies are themselves scarce. Every use of them to solve an historic accidental problem or to optimise something locally will incur an opportunity cost in not using them to solve an essential customer or business problem.

As a consequence I suspect that the "floor" in the extreme scenario for our model bank would be zero-branches, the same amount of IT, and between 10,000 and 20,000 people. 10x the size and complexity of a ground-up engineered, digital-native alternative and consistent with our previous view that 90% of what exists today is accidental complexity that simply cannot be eliminated.

Of course the extreme scenario is unlikely to be a real consideration for any bank in the next three years, regardless of the rate of customer behaviour change and digital adoption. A more likely outcome is a less-intense version of this. What could this look like? For our model bank I would expect a lower bound of between 30,000 and 35,000 people, 500 to 1000 branches, and a largely unchanged IT landscape. And the upper bound? For all the reasons we have described previously this could look like very little change at all.

However the real issue isn't the scale of restructuring and reduction that is actually achieved. The real issue is whether or not the transformation embeds options for what may come next.

If you are faced with a mountain you have several options, one of which is to live on the mountain

If we accept that there is a "floor" to this transformation, then it would be helpful to know when it will be reached. My guess is somewhere between three to five years. It is hard to imagine our model bank going at transformation longer than this and not reaching the point of diminishing returns. So at this point what options will it have to respond to further change? I would argue that it has no real options at all.

The theory of real options is complicated, but the practical application is less so. As Professor Amrit Tiwana* explains in his excellent book Platform Ecosystems, by analogy with financial options a "real" option is the right to do something without the obligation do it. A "real" strategic option refers to the the future flexibility to use something as a foundation or stepping-stone for as-yet unimagined follow-on purposes.

"Real options tend to be most valuable as well as more plentiful in fast-moving, unpredictable markets and in large, complex projects with greater disparity between the upside and downside outcomes", Professor Amrit Tiwana, Platform Ecosystems

Aren't fast-moving, unpredictable markets with uncertain outcomes precisely the conditions that banks face today? If so, where are the plentiful real options we should expect to see?

More specifically what real options does our model bank have once it has reached its floor? There are some who will argue that the data and API platforms that are being developed today could form the basis of a new strategic architecture and operating model, and hence could be a real option. However I am not hopeful this will be the case.

It is possible that if implemented mindfully and with careful design they could be the genesis of something strategic. However, as we have discussed before, regardless of the technology available to us we still have to understand the right problem to solve and then choose to solve it with the appropriate abstractions. So what is the problem in this case

• The reduction in people and buildings?
• Compliance with regulations?
• Engagement with customers?
• Development of a strategic platform for the future?
• A combination of all of these?
• Or something else entirely?

As Professor Tiwana states

"Real options are a way of thinking that disciplines how a project can be structured against potential losses while preserving potential gains. It is an approach to position for the upside by hedging against possible futures", Professor Amrit Tiwana, Platform Ecosystems

A real strategic option is unlikely to emerge unless this way of thinking is consciously adopted and options embedded in, notwithstanding attempts by well meaning teams to guide projects towards delivering one. Ultimately this transformation has nothing in it to hedge against different possible futures. Once the floor is reached that will be it.

So for our model bank to improve its future chances beyond three years, it will need to set off consciously to embed one or more real options within its transformation approach. The next choice presents an opportunity to do this.

Continue as a Universal Bank (as above) plus a Partial Open Platform/ Curated Ecosystem

We can describe this as the “less people, less buildings, more partners" transformation. The essence of this is the addition of a curated ecosystem of partners for serving existing customer needs in new ways or serving entirely new customer needs. It assumes the core of the business remains the same and the curated ecosystem is an "and" to the core business and not an "instead of".

At a practical level if our model bank makes this choice it will execute many of the same actions that will be undertaken in the first transformation. However the key difference between the two is a matter of philosophy, rather than just a matter of intensity. If 100% of resources (time, money, focus) are directed on "less people, less buildings" in the first transformation, in this one it will be 80-90% on "less people, less buildings" and 10-20% on "more partners". Setting up the enabling platform architecture, data architecture, governance and organisation to support a partially open, curated ecosystem of partner services is a conscious choice supported by time, money and executive attention.

Of course much of the outcome of this transformation will still be the same as before, both positive and negative. Progress will be made towards the floor, albeit with only 80-90% of the intensity. More square pegs will be put in more round holes, albeit 10-20% fewer of them. However the difference - and it is a crucial difference - is the opportunity to generate options. If designed and engineered appropriately the architecture supporting the ecosystem represents a platform where real options can be embedded.

In Robinson's four banking business models the Thin, Open Platform and Aggregator are two-sides of the same coin. Those embarking on establishing a curated ecosystem may not want to become either of these or indeed have the necessary capabilities to win if this is where they choose to play. However, it is in the ecosystem platform that real options can be embedded to hedge against these two possible futures.

Why is it possible in this choice? Because the characteristics of the platform architecture required to support a partially open, curated ecosystem include the base elements required for both an Open Platform and an Aggregator. They don't need to be fully developed or complete in scope of course, but that is where real options come into play. Of course few things in life are straightforward, and there are difficulties and dangers in adopting this approach.

The main difficulty comes from the usual suspects identified by Brooks and Ford. Effectively our model bank ends up with a selection of off-the-shelf components that limit the architecture or it decides to use a 3rd-party solution where the real options are not its to exercise (excluding venturing, which we discuss shortly). The result is the transformation does not embed the optionality our model bank is looking for.

The main danger is a gradual drift away from the ecosystem approach back towards a do-it-yourself model, perhaps because of implementation difficulties or a belief that services can be offered in-house more effectively. As the Reserve Bank of Australia (RBA) showed in their cost income analysis of global banks in their Financial Stability Review 2014, there is strong evidence that those who try to do everything themselves have higher cost-income ratios. This is primarily due to higher levels of staff remuneration for the diversified market-facing activities that they undertake.

In contrast, banks like our model bank have a favourable personnel cost position and - if executed well - the curated ecosystem has the potential to improve it still further. The RBA's analysis shows the advantage can be extended depending on the type of lending and institution undertakes, e.g. relatively homogenous home loans vs. more complex business lending requiring relationship management. A digital ecosystem approach should only amplify such advantages for those that already have them and help close the gap for those that don't.

So if executed well - with the right platform architecture and crucially avoiding the difficulties and dangers described above - this transformation positions our model bank better for the future. It provides for the upside of the ecosystem approach, whilst embedding real options that could further support a move towards becoming a Thin, Open Platform or an Aggregator.

This leads us neatly to the last three possibilities.

Thin Open Platforms, Aggregators and Infrastructure Providers

Of the new entrants into the UK market perhaps the one that has established itself most in the consciousness of customers and the industry is Monzo. They have specifically chosen not to become a traditional bank but will instead operate a marketplace. At the heart of this will still be a customer's relationship with their everyday finances (the current account), but with other needs served by their ecosystem of partners.

Alongside this significant decision on where to play, Monzo has made (at least) one other significant decision on “how to win” - to engineer the platform supporting their business strategy from scratch. Most other new entrants haven’t done this and it is significant. As Professor Tiwana describes in Platform Ecosystems,

"In true platform businesses the choice around the software architecture is not just a technology decision, it is a fundamental business decision. True platform businesses are two-sided, connecting two different cohorts who benefit from the network effects across the platform and within their group. And their architectures are engineered to evolve."

As Robinson's model defines it, Monzo is a (Very) Thin, Open Platform. In contrast our model bank is a Wide & Closed System. So we can describe this transformation as the "get like Monzo" choice.

To execute this transformation our model bank needs to make quite different decisions and undertake quite different actions than in the first two. The exam question is effectively "how do you become like Monzo"? Answering this requires strategic decisions on topics such as

• Which products to manufacture and which customer needs to serve through the ecosystem
• How to treat existing customers on those products that are no longer manufactured
• How to deal with the businesses and markets that will no longer be participated in
• How to establish the technology platform.

For a bank the size and scale of our model bank these sorts of decisions will have significant implications on customers, communities and in some cases local economies. Choosing to no longer participate in certain markets or to no longer manufacture certain products are significant strategic decisions and will require agreement from a wide set of stakeholders, as we noted earlier.

Given this it is unlikely that any bank will plot a direct path to becoming a Thin, Open Platform. However there may be a route that is indirect, with decisions made over three to five years in response to a changing customer and industry landscape. And an indirect route implies that there will be interim choices to be made. The key question then is where are the real options that enable these interim choices? For those executing the first transformation there are none. However for those executing the second transformation there will be options if the dangers and difficulties are avoided.

Of course there are other ways our model bank can generate real options to hedge for this possible future. Two possibilities are

• It invests in or ventures with a new entrant like Monzo (or a platform provider) in a way that gives it the right but not the obligation to exploit its platform in the future
• It establishes its own new entrant and sets if off on the same path, as part of a dual transformation strategy.

Regardless of the approach it is crucial that the executive of our model bank understand that in this choice the software architecture of the platform is not just a technology decision - it is a fundamental business decision. Home-grown, dual transformation, investing or venturing, if the architecture isn't right any real option will be worthless.

What about the Aggregator choice? In this transformation our model bank becomes a virtual advisor and distributor of other institutions products and services. A well executed ecosystem or open platform play will provide options for this possible future too, as it is essentially the limiting case - think "Wide & Closed" becomes "Vanishingly Thin & Infinitely Open". That said I agree with Robinson's analysis that it is hard to argue that others couldn't do this better and it leverages very few of the strategic advantages banks currently have. Consequently I think the route to this possible future will be even more indirect (and less likely) than the previous one.

Finally we come to the Infrastructure Provider choice. To execute this our model bank will need a platform architecture much like the one that underpins our "Musk-like" transformation. And if it had this, why would it make this choice?

Real obstacles don't take you in circles, they can be overcome

Having reached this point it seems we have come full circle. We started our discussion by claiming that over the next three to five years platform architectures will emerge that will enable existing banks to move to a fundamentally simpler operating model (organisation, people, process, technology). We explained the origin of complexity in banks today and illustrated the nature of the new platform architectures, together with how they could be exploited. We then made the case for why they would emerge, recognising it was finely balanced between possible and probable.

We then argued that reaching the tipping point might not just be a matter of time, but would require a strategic move by one or more existing banks. To hedge our bets though - just in case it does turn out to be only a matter of time before a new entrant such as a Thin Open Platform, Aggregator or Infrastructure Provider tips the balance - we looked at the alternative approaches an existing bank could take to respond.

Two things emerged from this

1. The need for real options to deal with an uncertain future
2. The importance of the platform architecture in enabling these.

The need for options should not be surprising. In this series we have purposely simplified our discussion to focus on how a bank can fundamentally change its operating model to respond to the forces of digitisation, i.e. customer behaviour, new technology, new entrants, open banking. As a consequence we have ignored many topics of strategic importance such as macroeconomic conditions, the uncertain political environment, changes in capital requirements, bank risk appetite, the upside of future interest rate increases, to name but a few. So if digitisation alone doesn't demand real options (and it does) then digitisation plus an uncertain world surely must.

However, what is surprising is the apparent lack of real options embedded in the digital transformations being undertaken today. To repeat ourselves, real options tend to be most valuable in "large, complex undertakings with greater disparity between the upside and downside outcomes". The emphasis (which is mine) could be the definition of the transformations ahead. And yet where is the real options way of thinking?

Equally, the importance of software platforms should not be surprising. Ever since Marc Andreessen explained why software is eating the world, platforms have entered the consciousness of most people and businesses. What is not well understood though, is that for platforms "software architecture is not just a technology decision, it is a fundamental business decision".

Given all of this, what should our model bank do over the next three years?

• Inevitably it will need to pursue most or all elements of our first transformation, probably with 85-95% intensity
• In parallel it should consider embarking on establishing a curated ecosystem for direct medium term advantage at 5-15% intensity. It should do this in a way that embeds a real option on a possible future as a thin, open platform or aggregator
• Alternatively it can select a more limited operating model and/ or IT platform simplification opportunity in a simple sub-business or product-line that has an immediate need. Again it should look to embed a real option on the underlying platform
• It should also consider whether or not the time is right to pursue with 1% intensity real options to support our "Musk-like" transformation (or support a possible future as an infrastructure provider)
• For the last three points venturing or investing with a new entrant (new bank or platform provider) should probably be the preferred approach given the skills, talent and environment required to succeed.

This is a lot to do and it will require new understanding and talent to design and deliver. I am sure there are many who think our model bank can't do it. I think it's possible.

What do you think?

* I would like to thank Prof. Tiwana for his kind permission to make reference to and quote from Platform Ecosystems: Aligning Architecture, Governance and Strategy