The Complementary Middleman

Part one in a four-part series on the future of service

“The Middleman Strikes Back”, Ezra Galston, Techcrunch:

One of the great promises of the Internet — a democratic, transparent, open network that would disintermediate entrenched industries, remove fee-taking middlemen and thereby lower the costs of goods — has seemingly been realized.
Priceline and Expedia mostly killed travel agents; Prosper and Avant are fast disrupting bank loan officers; and the car salesman has been diminished in favor of eBay, Autotrader and Craigslist.
And yet, just as surely as we thought they were gone for good, middlemen have come surging back to life — rebranded as our best friends: the personal concierge.

Ecommerce is eliminating the everyday middleman because buying online is a better experience in every dimension.

It’s faster than going to a store or picking up a phone. It has clearer information about price, whether something was in stock, and when you might get it. It expands the number of things we can buy, regardless of geography, and is open 24/7. On top of all that, the prices online are better because there’s no brick-and-mortar overhead and customers can easily comparison shop and share information about the experience. Now, the competition is a click away, not fifteen minutes up the road.

But we’ve lost something in this shift online. We can use ecommerce as a stand-in for the cultural forces at play: when we take it apart, we can see the aggressive move to self-service that’s happening in both online and physical retailers; the new and incomprehensible scale of the available goods, services, and information available to the average person; and the spread of an always-on, speed-obsessed culture that is arguably hostile to healthy human relationships, where the default expectation of our environment is instant and 24/7.

We’re entering an age of superabundance with humans at the periphery.

In the process, we’re losing both the human touch of service and our own free time as we do more and more work ourselves that used to be done by others.

In response to this shift, we’re now seeing an explosion of concierge-like products that promise to take care of household chores, help you shop, do background research, make calls and schedule appointments, book travel, get advice, or, most generally, do anything as long as it’s not illegal, all with a personal touch.

However, these services aren’t a return of the traditional middleman because scarcities that allowed traditional middlemen to function are vanishing.

Instead, I like to think of these products collectively as “Middleman as a Service” (if MaaS doesn’t sound too ridiculous to you) [1], and as I see it, there are several critical aspects to this type of product.

First, MaaS will be deployed by customers to complement self-service purchase options, not replace them.

Second, MaaS are the next step in the evolution of something more akin to the search engine than a traditional platform or marketplace product: a “do engine”.

Third, these products will be built on messaging interfaces and will accelerate a fundamental shift in human-computer interaction. We’re used to conforming our brains to pre-defined UI when working with computers. However, we’re about to see a wave of UIs that conform to our own brains, so to speak — to how we structure the problem at hand and think and feel about it — which will enable us solve problems more rapidly and naturally than ever before.

Fourth, messaging platforms are native app killers, and MaaS are killer apps for messaging platforms. The default consumer service product strategy will shift from “build an app” to “build an API” with the messaging platform providing the UI, and this will, in turn, reinforce messaging platforms’ position as a new method of product distribution, weakening the position of Apple and Google as app store gatekeepers.

There’s a ton to unpack, so in this first essay in the series, I’ll be looking at the first point: what a MaaS is, how it’s not a traditional middleman, and how it complements self-service, rather than replaces it.

So why isn’t MaaS a traditional middleman?

The business model has to be different

Taking a step back briefly, why won’t MaaS be a return of what we traditionally think of as a “middleman”?

I think the answer boils down to two things: commodity manufacturing and the internet.

Traditional middlemen could rely on scarcities to extract economic rent, which basically describes the “I’ll charge more because I can” effect where someone brokering a transaction can charge more without actually doing anything valuable in the process.

The scarcities middlemen could rely on to extract rents from customers were access to the means of exchange (like a broker to a stock exchange), reliable information (like knowing how much something actually costs and what competitors are charging), local inventory (actually having it in stock), logistics (being the only ones who could get it to you), and access to the person who’s producing the thing they’re selling (access to the wholesaler or manufacturer).

The capabilities and supply chains that support commodity manufacturing are eliminating former scarcities around logistics and access to production. We can make more, better things more cheaply and at greater volumes than we ever could have before. In the last thirty years, better operational processes improve quality and reduce waste, industrial robotics removes humans from the manufacturing line, and globalization and containerization allows manufacturers to reduce costs by moving the parts of production where people are needed to cheaper locations have led to manufacturing becoming commoditized. This trend is likely to continue, or even accelerate, over the next decade as advances in 3D printing technology drives manufacturing costs to zero.

At the same time, the internet has removed scarcities around access to means to exchange, information, and inventory. You can now exchange stocks without a broker and book your own flights and hotels without a travel agent (intermediary automation). You can compare prices in a couple of seconds, regardless of whether you’re in a brick-and-mortar store or in front of computer, and read hundreds of reviews from (ostensibly) other people that allow you to evaluate whether something is as good as the salesperson or manufacturer says it is. Social media allows people to name and shame businesses that have poor service or customer-hostile policies. And being the only place in town that carries something matters less when you can order something online and have it delivered economically to your door in anywhere from a couple days to an hour.

Collectively, these changes remove formerly reliable scarcities for traditional middlemen, exerting downward pressure on margins.

There’s a second factor at play, too: software is radically changing the cost of running a business.

Ecommerce is slowly but surely eating traditional commerce [2] for the same reason that software is eating the world more broadly: software offers low-to-no transaction costs and is cheaply and infinitely scalable, unlike traditional operations.

A business starting with a brick-and-mortar presence has to worry about all kinds of operational issues — utilities, rent, staff, stock-outs, and carrying the right inventory — in order for someone be able to come in and buy something. [3] These costs make it into the price that traditional players charge, and they either don’t exist or exist in much smaller magnitudes for an ecommerce player. With ecommerce, it’s both cheaper and significantly faster to set up shop.

It’s also far cheaper and easier for an e-commerce player to scale operations.

If you’re a traditional player and you want to operate 24/7, you’ll need to staff up. If you’re a software company, you make sure your servers don’t go down.

If you want to triple the transaction volume in your store at the same throughput (eg. Black Friday in the US or Single’s Day in China), you’ll need to increase your staffing for that day far ahead of time. If you’re a software company, you spin up another server whenever the traffic hits.

If you’re a traditional player and want to offer your product in an additional 130 countries, you need to invest months of preparation and spend lots of money. If you’re a software company, you announce that your service is now available in 130 countries and flip a switch, like Netflix just did.

Ecommerce — or really any business that processes transactions and scales through software — will win on price because it’s so much cheaper to run, and that goes for both upfront and recurring costs.

When we take these trends together — formerly reliable scarcities becoming less scarce, collapsing margin, and ecommerce changing the fundamental operating costs of running a business, collapsing prices — a negative feedback loop starts to emerge.

Ecommerce competitors with lower operational costs can drop prices while maintaining similar margins and have the option to decrease their margins even further in a bid to the customer relationship and capture even more demand and available spend at scale.

This makes the business model of traditional middlemen unviable. MaaS and what comes next, by necessity, will have to take a different approach to business.

The experience has to be different

Ecommerce is deadly because it changes available business strategies while providing a better, more convenient experience for the customer.

There’s no phone call or trip to the store, more choices than you’d find locally, and 24/7 availability. There’s accurate information about inventory, shipping, and price. The store can securely remember your preferences, purchase history, and payment options.

Buying online is faster, easier, more trustworthy, more personal, and cheap and getting cheaper. Ultimately, though, the interaction experience is driven by the software capabilities and the dynamics of competition in ecommerce.

When multiple e-commerce retailers selling the same products compete, their margins will approach zero in the process of competing with each other. Margins will approach zero because it’s easy for customers to buy somewhere else, because you can drive margins down if you’ve got more scalable software that drives down operating costs, and because the absolute magnitudes of money involved in high volume, low margin scenarios are still attractive — as is owning the customer relationship, which is itself driven by winner-take-all dynamics.

In markets with commoditizable products, low-to-zero margins are going to be the new norm for businesses if they’re lucky to survive. [4] To survive with low-to-zero margins, they’ll need to scale the entire experience — finding the thing to buy, buying it, asking questions about it. The experience will be self-service because that’s the model that scales with and like software. For customers, self-service is — and will be — the new default. [5]

How it complements self-service

Our world is increasingly characterized by superabudance. We produce more goods, services, and information more cheaply than ever before in human history. At the same time in the developed world, we work longer hours when we have work, deal with the paradoxes of choice that come with that abundance, and spend more time doing things ourselves that used to be done by others as businesses shift to self-service by necessity. The new scarcities are free time and meaningful human interaction.

Low-to-zero margins for businesses, the self-service default for customers, and the quality of the superabundance/scarcities in the environment defines an opportunity.

For ecommerce, there’s opportunities to differentiate on service to attract customers, or to create a higher-margin product centered on the service experience. In either case, the service product will rely on the operational and technical capabilities built for the self-service product as a kind of API and will use those capabilities to execute quickly at every step of the transaction. (I think this dynamic will play out for luxury and commodity products alike.)

MaaS will complement self-service because it’s more expensive to offer and hard to do well. Making MaaS-like service the default could raise costs across the board (likely not the right call if you’re selling commodity stuff), and you might lose some customers expecting self-service.

For customers, there’s an opportunity to save significant amounts of time by having others solve daily problems in a way that’s as easy as buying something online. But it’s hard to beat the pure utility of self-service in situations where you know what you want and think it’d be faster to take care of it yourself because it feels likely that the service will cost more and will be slightly slower.

Instead, MaaS will be something you choose in certain situations when you don’t want to self-serve [6] — and the harder and more foreign the problem, the more valuable the service.

While MaaS products are emerging in response to the opportunity defined by low-to-zero margins, the self-service default, and superabundance, they’re not going to replace self-service as the default because they’ll be more expensive for businesses to offer, will rely on a robust self-service purchase experience to create a great service experience, and might not be the first choice for a customer solving a problem because it could be slower and more expensive of solving a problem — at least for now.

The New Model

MaaS products aren’t a return of the traditional middlemen because traditional middlemen aren’t coming back. Scarcity-driven business models with brick-and-mortar overhead are breaking as the internet and commodity manufacturing put downward pressure on margins, prices, and operating costs while ecommerce provides a faster, more convenient user experience.

A MaaS launched today will have to make money another way. As scalable, efficient self-service ecommerce becomes both the default experience for users and table stakes for businesses, there’s an opening for a complementary product that uses self-service infrastructure to solve problems.

In the next three parts of the series, we’ll take a deeper look at what a MaaS product looks and feels like, including why it shares more DNA with a search engine than a traditional platform (Part 2), why it will be built on messaging and shift how we think about interacting with computers (Part 3), and how it will shift product strategy from “build an app” to “build an API”.


[1] Why not just use “concierge”? Concierge services are usually included for free as part of an existing purchase. MaaS is employed by the customer directly, and it’s the customer who pays for their services. I’d argue that “personal assistant” is a more useful metaphor.

[2] If you look at e-commerce as a percentage of US retail revenue excluding gasoline, food and beverage, and building supplies as Benedict Evans did in his analysis of Amazon in 2014, e-commerce’s share of overall retail revenue could be a couple percentage points higher than that.

[3] The operational cost divide is even more stark when the cost of carrying and selling additional inventory is close to zero for the e-commerce player and real for the traditional player. Media is a great example of this: it costs Apple and Amazon nothing to rent out another copy of a movie compared to a brick-and-mortar video store, which has to purchase the right number of discs ahead of time to meet demand and deal with returns.

[4] The market for luxury goods is obviously different, but it might not matter. A MaaS product might eventually be built or bought by conglomerates (like LVMH) to differentiate the shopping experience, aid discovery, and drive conversion rates across its portfolio.

[5] It seems like this is happening in traditional brick-and-mortars as well — think of what’s been happening in big box stores, supermarkets, and pharmacies over the last 20 years.

[6] If you’re particularly wealthy or time constrained, maybe you’d employ a MaaS for everything. A MaaS has the potential to capture the best customers and have attractive margins because it’s moving up the stack in the customer relationship and further away from the low-to-zero margin self-service purchase experience. It’s hard to put a price on the ability to do things that you couldn’t normally do yourself, faster than you thought was possible. It makes you feel superhuman.

Thanks to Dan Teran, Phil Sarin, JT White, Harrison Johnson, Jacqueline Kurdziel, Trista Kempa, and Courtney Caccavo for reading drafts of this essay.

If you’re interested in the future of service, shoot me an email at We’re building it at Q.