Google is our gateway to the internet. If we want to find something, we google it. Google has a market share for search of 78%, which rises to 90% if you exclude China and Russia. It is the archetypal data aggregator. How, therefore, could banks, an industry infamous for not capitalizing on its data assets, make better aggregators than Google?
The answer lies in the business model. Google serves advertisers, banks serve customers. In the long term, the latter will make for better aggregators.
An explosion of data
As Mark Weiser put it, the most profound innovations disappear, weaving themselves so tightly into the fabric of our everyday life that they become indistinguishable from it. This is certainly the case with the internet. We started by sending a few emails and now we spend 23 hours per week reading news, watching movies, listening to music, shopping and messaging our friends. It has become an integral part of our lives.
In addition, improvements in connectivity –in terms of speed and cost of accessing the internet — have allowed more people to access the internet over time and in more convenient ways. Today, there are 3.8bn internet users, 51% of the world’s population, and the number has been growing by 1m a day since 2012.
We’re also connecting more and more inanimate objects with the internet. The devices create new streams of data about us (e.g. our heart rate) and metadata about how we use these devices.
All of the above has produced an explosion in data. The amount of data being produced is doubling every two years, equating to a 50-fold increase from 2010 to 2020, with machine-generated data the fastest-growing dataset — expanding at roughly 50 times the rate of traditional business data.
The rise of the data aggregators
So much data is now available to us that we need help to manage it.
So data aggregators emerged, platforms that could help to gather and organize the vast amounts of information on the web.
There are essentially two types of aggregators. Those like Google, Facebook and Twitter that are “free”. The consumer doesn’t pay directly for the service; instead, the aggregator monetizes the service through advertising. Then there are those like Netflix, Spotify or Uber that provide us a service either on a subscription basis or taking a fee per unit sold (e.g. per ride organized).
Aggregators become monopolies
Almost all data aggregators benefit from network effects, which make them very different from industrial age companies and which give rise to high market concentration.
A network effect, also known as a demand-side economy of scale, is defined as a situation where more users of a product make that product better. Take Google, for example, the more people that use it for search, the more data there is to train the algorithm, the better the search results become, attracting more uses; a self-reinforcing virtuous cycle.
It is the presence of network effects that drives the tendency towards monopolies. According to economic theory, as firms grow, they can obtain supply economies of scale — where they spread fixed costs over larger levels of output, improving margins — but they are still subject to decreasing returns tp scale over time. However, in a world of network effects, this does not hold true; faster growth leads to faster growth.
Exercising monopoly power
When a company becomes a monopoly, it tends to exercise monopoly power.
According to economic theory, monopoly power is manifested in the monopolist restricting output in order to take up prices and boost its profits. The loss to consumers from having to pay a higher price for fewer goods and services is known as the “deadweight loss to society”. US anti-trust policy, heavily influenced by the Chicago School of Economics, uses price as a proxy for monopoly power and intervenes where growing industrial concentration leads demonstrably to higher prices, as it did by regulating AT&T call charges, for example.
The issue when it comes to platform companies is that they tend to lower, not raise, prices to consumers. This is why the US government does not investigate platforms for monopolistic behavior and why, where the EU does investigate and take action against platforms, it does so on the basis of fair competition rather than a traditional measure of monopoly power.
But the platforms are exercising their power in ways that cause social harm.
Transferring utility to consumers
The economic theory of monopolies pre-dates the digital world. Today, the marginal cost of supply for most digital platforms is zero. Therefore, profits aren’t maximized by cutting supply, but by maximizing revenues. And since most of the digital platforms deal in luxury goods (e.g. holiday lets), demand tends to be relatively elastic and so revenue maximizing tends to expand supply quite a bit or, as Andrew McAfee and Erik Brynjolfsson put it, “The revenue-maximizing price, in other words, is surprisingly low”.
This a boon for consumers. Increasing supply leads to lower prices which leads to more consumer surplus. The issue, however, is that this gain in consumer utility is coming at a cost elsewhere.
Some platforms might give rise to externalities not priced into the transactions, such as AirBnB which has side effects in terms of increasing rental prices and changing the nature of local amenities to favour tourists rather than residents (turning cities into pseudo-theme parks), which has caused protests in places such as Venice and Barcelona.
But in most cases it leads to a straight transfer of utility from either producers or workers to consumers or, in some cases, both.
In the case of Uber, the gain in consumer welfare comes at the expense of drivers who earn less. In an excellent Freaknomics podcast, Stephen J. Dubner shares his analysis that, in a single year in the United States, Uber produced USD7bn of consumer surplus, with consumers paying USD4bn when they would otherwise have been prepared to pay USD11bn. The drivers received USD2.5bn and Uber the rest.
The fact that platforms are driving a wedge between what consumers are prepared to pay and what workers receive is why Nicolas Colin argues that we need a new social pact between consumers and workers. It is also why regulation of the sector is likely to increase and why there might be space for new aggregators.
Hijacking our attention for the highest bidder
In the case of “free” platforms like Facebook, there is a transfer of utility between different stakeholders: for example, the profits of content publishers have plummeted and there are consumer to consumer transfers in that most content on Facebook is user generated.
But the bigger issues from the “free” platforms is that maximizing revenue requires us to spend more and more time on the sites, which produces very significant negative externalities.
In his brilliant Medium post, Tristan Harris sets out all of the tricks that internet platforms use to deliberately hijack our attention so that we spend longer on the sites and so that they, in turn, are able to subject us to more adverts. It is difficult to know or quantify fully the negative impact of us spending so much time on these sites, nor of the opportunity cost (we collectively spend something like 400 billion hours on Facebook alone, which could be spent in other ways). Nonetheless, we can point to studies showing that more time on social sites makes us less happy as well as studies showing our attention span is shrinking (from 12 to 8 seconds since 2000).
Another negative externality is the gradual debasement of the truth. The debasement of truth happens in many ways. Firstly, since there are so many calls on our attention and our attention span is getting shorter, and since so much of the media relies on clicks to generate revenues, there is an inevitable move towards shorter stories and sensationalism, both of which are to the detriment of balanced reporting. Secondly, since internet platforms take no responsibility for the information that is published on the site (the laws in the US give them a clear exemption in this regard, which many people argue should be revisited now that social sites have become the dominant source of information), there is minimal policing of content and almost no attempts to establish its veracity. This has led to countless examples of groups writing deliberately false stories which they calculate would have a higher probability of going viral. Lastly, since the algorithms that Facebook and others use work by serving us up more and more content that they think we’ll like, they create “filter bubbles” whereby we are exposed to fewer and fewer articles that challenge our prejudices or alter our worldview.
In a world where our attention can be easily hijacked and where sensationalized or plainly false content can be widely shared, there is clearly the opportunity to hijack democracy, which is alleged to have happened in the US election and in Britain with the EU referendum. Further, in a world where fact and opinion become blurred, the space for objective discussions becomes more limited, social divisions rise and trust in government and liberal institutions diminishes.
Advertising-based business models will come under pressure
Regulating the platforms to reduce externalities might be an option, but the best long term solution would be a market-based one: the advent of better platforms. Just as Facebook displaced MySpace, so new platforms could emerge to displace the incumbents.
This could happen for any of the charging platforms, which is why they are becoming more vertically integrated and asset-heavy so as to improve customer experience and protect their incumbency. But, it seems that there are more reasons to think that the adverting-based platform models may come under pressure (in the near term).
The first is that many consumers seem to be becoming gradually more aware of the extent of surveillance, which may translate into lower usage or slower growth.
Second, consumers in general are becoming more aware of the value of their personal data and are less likely to share it (without getting additional value for it).
The third is regulation: new laws, such as GDPR (EU legislation but with a global reach), make much more transparent to us what information companies hold on us and give us more control over our own data, including the right to be forgotten.
Fourth is the fact that the return on advertising spend on these channels is likely lower than many people think and so advertising spend will likely fall over time (to be discussed in a later blog)
And lastly, as we face increasing calls on our attention, we will seek services that, as John Hagel puts it, will help us to increase our return on attention, which involves education and helping us to make better choices.
And, as this chart above illustrates — showing the number of Y Combinator applicants citing advertising and SaaS as their main source of revenues — maybe the tide has already started to turn
Banks can forge a different path
If the ad-based platform model goes into retreat, this would create a gap in the market for platforms that take safe custody of our data and help us to improve our return on attention by understanding ourselves better and making better decisions.
Banks are in a strong position to do this. We trust banks. Various studies show not just that banks are more trusted than any other institution to safeguard our data but this trust is a major source of competitive differentiation. In addition, banks don’t have any inherent conflict of interest with their customers: their customers are their customers; their machine-learning algorithms can be put to use solely for the purposes of increasing their customers’ welfare.
And so banks must become data aggregators themselves. They must use their position of trust, their access to rich data sources and their goal congruence with their customers to improve customer financial wellbeing. In the post-GDPR world, it will be necessary to persuade customers to share their personal data, which studies have shown they would be happy to do in exchange for value-added advice. In practice, this means aggregating many different types of data: customer data (financial, contextual, locational), market and risk data as well as information from third parties such as product catalogues. The goal would be to help customers to understand their financial health and financial needs in an impartial way, give them the right advice and help them to access the products and services that best match their requirements, even if those services come from competitors and even if those services are outside of financial services (such as education or insurance).
To become a data aggregator is both an offensive and defensive move. It is offensive in that there is a gap for a trusted advisor to perform this kind of aggregation service. But it is also defensive in that, in the digital age, banks are struggling to find a new value proposition that moves them away from direct competition with the thousands of fintech companies against whom they are not well equipped to compete — in terms of agility, customer experience but especially cost.
And so banks must become data aggregators. As discussed in an earlier blog, the best model is one that combines aggregation with vertical integration to offer consistent and very high levels of customer fulfilment together with the production of some products and services (where strategic, for example).
The model of banks acting as data custodians and trusted aggregators is one that would not only be of value to the many stakeholders (customers, banks, fintech companies, other service providers) but society itself.
An antidote to many of the prevailing platform models, it has the potential to create large positive externalities in two important ways.
Lowering transaction costs: there are massive transaction costs in financial services borne by both sides in each transaction. The individual incurs significant “shopping around” costs which are both monetary and normally non-monetary (time) trying to find the best product or service at the best cost to meet their needs. The financial institution then expends significant costs in onboarding that customer: establishing their identity and performing KYC checks, establishing their financial solvency with credit checks, etc. If the banks become the customer’s trusted gateway into financial (and non-financial) services, these transaction can be cut significantly. KYC checks, for example, only need to be conducted once for all of the services to which the consumer is introduced by the bank, while if the consumer genuinely trusts the bank to find them the right products and services, then shopping around costs disappear (along with significant expenditure in other areas, such as advertising).
Reducing information asymmetries: as well as cost, the other reason why many financial products are underprovided vs. what would be societally or economically optimal is lack of information. For example, if you don’t have a credit record, it will be very difficult to obtain personal credit. If you can’t produce audited financial statements, it will be difficult to get a commercial loan. And so, if banks acting as trusted aggregators can afford greater information transparency, by for example persuading a business to give API access to its online accounting systems, then these asymmetries can be reduced materially.
If transaction costs and information asymmetries fall, then the cost of financial products fall too, which enables more people to access them, helping to address the chronic problems of banking non-provision as well as under-provision (in areas like retirement planning and SME lending)
The pitfalls to avoid
As data aggregators, however, banks take on a different role in our lives which, along with new revenue opportunities, carries new risks and responsibilities. And the risks are higher for a bank compared to an internet platform given the higher levels of trust implicit in the relationship.
Below are some of the pitfalls that banks should avoid if they wish to exercise their role as aggregators successfully, impartially, ethically and in their customers’ best interests.
Advertising. The first one is obvious, banks should not open up advertising revenue streams otherwise they will create a conflict of interest.
Letting algorithms work unchecked. To be able to operate at the scale necessary and to be able to generate ever improving insights, banks will need to employ artificial intelligence. However, they will need to scrutinize the results and be mindful of what we have seen with today’s internet platforms: that algorithms can be gamed. But, perhaps more important is the issue of bias. The vast majority of machine-learning algorithms are supervised, meaning that they are initially “trained” to give certain outputs from certain inputs, which they get better and faster at doing over time. This creates a risk that human biases get transferred. Furthermore, even where learning is less supervised, the learning is normally carried out using training data which may perpetuate biases inherent within it. Already, potentially biased algorithms are being used in a number of critical areas, such as who gets parole. So banks will need to be vigilant, studying which factors are truly relevant to decision-making and deciding when an otherwise important factor should be overridden.
Price discrimination. Price discrimination occurs where a producer is able to get different consumers to pay different prices for what is largely the same product. A classic example is peak and off-peak train tickets. When banks are in possession of much more information about their customers than they have today, they would in a position to set a different price for every consumer based on our own unique demand curves, effectively setting the price at the maximum each of us would be willing to pay. As before, banks should guard against this temptation lest the backlash would be severe. Further, since banks would be sourcing from many third parties, this would create a potential arbitrage opportunity across platforms, undermining the claim that they are acting in customers’ best interests, which is intrinsic to the value proposition.
Not policing their sites. Banks will have to conduct much more ex-ante vetting of the partners they work with (to ensure, for example, that they are financially stable) than platform companies do, which tend to rely on ex-post quality indicators such as reviews and ratings. This might be sufficient if you are buying a book or a jumper, but unlikely to be enough when you’re getting a mortgage.
Using their power to extract excess rent from providers. As discussed above, many platforms have used their monopoly power to transfer utility away from producers and workers to consumers and to themselves. There may be public backlash against them and, as we are seeing, there is likely to be a regulatory response. But, as noted in the introduction to this section, banks will be judged to higher standards than today’s platforms and would be better to act always as good corporate citizen.
Excessive focus on sales. Another pitfall to avoid would be to make too many product and service recommendations, by setting objectives on sales rather than customer satisfaction or engagement. Too aggressive a sales focus would kill customer experience, undermine the idea that banks were trusted advisors and ultimately drive customers away to competing platforms.
Security failures. What has deterred so many banks from investing heavily in data analytics or becoming data aggregators is the notion that they are custodians — of customers’ assets and their data. This is true, but banks must be both data custodians and data aggregators. The case for becoming a data aggregator has already been made at length. And it is not even clear that banks can remain walled gardens in the new world of open banking where customers will share their data with multiple new intermediaries.
So, regardless of whether or not they act as aggregators, banks will have to ensure end to end security for customer data across the value chain; their value proposition hinges on it.
Our world is awash with data. We need data aggregators to help us to navigate this sea of data. But today’s principal data aggregators aren’t doing a good job. Either they are abusing their monopoly power or, worse, they are hijacking our attention in the service of advertisers, causing a plethora of negative externalities. Banks, as data aggregators, are in a position to do a better job. Monetization of their data services doesn’t depend on advertising and their machine-learning algorithms can be put to use creating ever-improving levels of consumer welfare. But they must do so ethically, impartially and always in their customers’ best interests.