The Rise of Tech-Enabled Middlemen

Marketplaces have been a hot category of startups over the last few years, from ‘sharing economy’ platforms to all the on-demand delivery apps to B2B portals linking businesses with vetted service providers. The core value of the marketplace model is to create market efficiency by cutting out middlemen and connecting the buyers and service providers directly. They make transactions better, faster, and cheaper.

So it’s interesting then to consider the (still early) trend of another crop of startups focused on putting middlemen back in. Many processes are too complex, or many types of searches too personalized, for straightforward two-sided marketplaces to serve as the simplest solution. Plus consumers are reaching “app fatigue” after a decade of downloading and figuring out how to navigate a new app for every little thing.

This leaves an opportunity for a tech-enabled middleman (TEM)…a tool that figures things out and coordinates them for you. TEM startups target ways of helping people that could traditionally only be done on an agency model, but they leverage new technology to create a scalable approach that is a hybrid of algorithms and human customer service. Most can be described as “your personal _____ agent”. While someday (albeit probably not in the decade ahead) artificial intelligence could handle the full scope of operations and human interactions, TEM operations are partly automated through AI but also employ human concierges.

(There are a number of startups similar to this using 100% natural language processing and no humans, however everyone I’ve seen remains too clunky or single-purpose for consumers to trust it in handling anything important. Most consumers are not technology purists…they like having a human at the other end.)

For example:

Service has technology and customer service specialists that quickly solve your problems with companies so you don’t have to haggle with 1–800 customer service lines yourself (think canceled flights, Verizon overcharges, poor service experiences, etc).

Operator and Magic handle your shopping list for you, searching for goods, reading reviews, ordering products, and coordinating immediate delivery.

Fin searches for information like a personal assistant who instantly Googles whatever you want to know (like the AI in the film Her).

Journy applies this to travel with an enormous database of activities and events curated by local guides who give every traveler unique recommendations, plus answers travel questions and books reservations.

Super coordinates all the to-do’s and paperwork of home ownership and home maintenance.

Better was (until shutting down last October) using technology and “personal health assistants” to coordinate every aspect of a person’s interaction with the healthcare system — booking appointments, reviewing doctors, coordinating care between doctors, dealing with insurance, providing answers to common medical questions, and more.

Communication between a customer and their TEM is best done through a simple messaging interface in either a native mobile app directly through the phone’s SMS). This is partly because the experience feels simple and natural to the customer (there’s no learning curve), but of equal importance is that it lets the TEM switch between a human and AI in conversations without the customer noticing.

For TEM economics to work, you need the following:

1. a deep pain point. That is, a process so complex, time consuming, and/or stressful for customers that they would happily pay some someone else a bit extra to handle it for them. Find the deepest pain by looking for fields where even those who work in the industry get overwhelmed and confused (healthcare, customer service, government services, etc.).
2. a solution that provides substantial, concrete value (measured in time saved, money saved, stress reduced, and/or outcomes improved). The problem with all-purpose ‘personal assistant’ apps is their value prop is too vague and their abilities too unspecialized.
3. economies of scale, wherein the TEM handling the process on behalf of customers can do it ever faster, cheaper, and more effectively the more users it acquires. This model is scalable only if AI can take over an increasing amount of conversation and work (relative to human employees), continuously sharpening its understanding of each customer’s needs and the way to solve for them. Otherwise it’s just a tech-savvy agency.
4. high frequency of use over a short period time, or medium to high frequency of use on an ongoing basis. High frequency for a temporary length of time can work if that’s an intensive period with substantial value on the line (tour guide while traveling, coordinator when selling/buying a house, etc.) but the immediate ROI must be apparent to the customer.

What else fits these criteria?

There are a number of opportunities I see for new TEM startups to tackle important, high-value challenges in people’s lives.

• Personal finance — a TEM that coordinates your accounting, budgeting, student loans, taxes, etc. on an ongoing basis, plus gives you wise counsel on spending decisions, loan refinancing, choosing the right credit card, and perhaps even maximizing your “points” from different cards. (Penny seems to be targeting the basic spending guidance piece of this pie. Intuit’s Tada app does this for filing taxes.)
• Healthcare — Better (incubated by Social Capital and the Mayo Clinic) closed down in October, but I believe the opportunity they were chasing remains there and remains massive. Healthcare is very complex and finding just the right model to pull this off isn’t easy. You’ve got the patient, family of the patient, different medical providers who don’t talk to one another, and then insurance.
• Job hunting — you need a new job…where do you start? You message a TEM focused on recruiting and they send you opportunities that are a strong fit, as well as advice and an inside scoop on different companies. For active and passive job hunters alike. Maybe it even uses your social media accounts and phone book to suggest folks you know that can provide introductions or referrals. (OnboardIQ briefly operated a tool called Miley last year doing this for hourly on-demand startup jobs.)
• University admissions and enrollment — help me find the right program and figure out how to apply, plan for the SAT/ACT/MCAT/LSAT/GRE, navigate financial aid, choose the right student loans, etc. Rich families pay thousands (or tens of thousands) per year to have private counselors guide their child through the college admissions process on the traditional agency model…a large proportion of that could honestly be done by technology instead of humans. The college admissions process itself is a barrier to middle and lower class students’ educational pursuits since so much of it is about understanding the game.
• Government services (local, state, and federal levels) — have you ever interacted with government and not wanted to pull your hair out? Everyone has to navigate government, from local businesses figuring out regulations and licensing, homeowners going through zoning, families wanting to access local recreational (do you know how to reserve space at your local park?), questions about senior/disabled/veterans’ benefits, and staying updated on legislation that’s relevant to your concerns.
• Nutrition — our nutrition is one of the most important things in life, with a ripple effect across health, thinking, mood, and more. But few people have any idea how to piece together all the moving parts to fit their particular situation or know what actual research says about certain foods. Wouldn’t it be awesome to have someone who’s an expert always on call for nutritional guidance unique to you? If you are as stressed by cooking or even ordering food as me, maybe they just tell you what you’re going to eat and order it for you. Depending on how in depth the nutritional guidance is, this could step out of wellness and into healthcare territory…where it would need licensed nutritionists, but potentially get covered by insurance.
• Fitness — The lighter, cheaper version of having a $100/hr personal trainer multiple times per week. A mix of AI and actual trainers create unique workout plans, proactively give you feedback based on both self-reported results and data from workout apps like Strava, and are on hand to answer quick questions about regiment, nutrition, technique, etc.
• Apartment hunting — even with all the sites listing apartments this is an experience everyone dreads. Tell an app what you’d like and it curates just the few top finds, plus then coordinates paperwork, payment, and potentially ongoing interaction with the landlord.
• House hunting — A TEM that hunts for properties and coordinates the financing and closing. (Opendoor is approaching this although I wouldn’t quite classify it as a TEM since they actually just buy your house from you upfront then figure out the sales process to the next resident themselves.)
• Car buying — Hunting for a car is like hunting for an apartment, especially if you want a used car. It’s hours scrolling through sites looking at listings. It’d be better to just text the conditions of your search and let a service hunt for and vet qualified cars on the market so you don’t have to scour Craigslist.
• Dating — Dating apps have a hard time understanding your “type” because it’s a very human skill to gauge that. Perhaps that’s an opportunity for a TEM that uses a light touch of human match-makers empowered with AI. People might pay to receive ongoing “picks” from a tech startup that has additional curation from real humans who can instantly gauge aspects of personality, attractiveness, style, and “sketchy-ness” that are currently impossible for AI. (Update: I’m told Three Day Rule in LA is targeting this.)
• Enterprise/SMB use cases —TEMs that coordinate purchasing, travel, relationship management, legal, and many of the same financial considerations as noted above for individuals. A TEM that acts as an outsourced CFO for SMBs would be interesting. Also tool like Operator or Magic specifically dedicated to B2B purchasing could be incredibly helpful, especially when you think of all the business owners wanting to know what software tools to use (within their budget) for marketing, recruiting, onboarding, payroll, benefits, etc.

What’s the business model?

The business model for TEM companies is typically one of two: a subscription or a percentage cut of transaction value.

Many of the startups in this category launch doing everything for free, then implement a business model after they’ve gained traction and experimented with different approaches. This makes sense within a short window, but is risky to prolong. It’s a mistake to gauge “product-market fit” solely by whether people will use and praise a free service…most people are quite happy to let someone help them for free (i.e. it’s not hard to gain customers for a landscaping service by offering to do your neighbors’ landscaping for free.) The real question is whether people will actually pay for the service provided at a price point that’s profitable.

Subscriptions make sense when confronting an ongoing, medium- or high-frequency pain point. Customers see the value of the service on a daily or weekly basis; it eliminates something that’s a constant drag on life rather than the occasional big thing. Ongoing help in nutritional or fitness guidance, personal finances, looking for dates, passively job hunting, and coordinating healthcare fall into this category. There’s a high switching cost to leave one of these after all the personal data you’ve “invested” in them through regular usage, making them more defensible against competitors.

The value-based model makes sense when the task has high financial value and is a low frequency occurrence (but when it does occur it’s high-intensity within a short window). The need to help someone buy/sell a house, lease an apartment, or coordinate their vacation is infrequent enough that customers won’t maintain an ongoing monthly subscription. But if the TEM provides a lot of value when the need is present, that’s something worth paying for. A percentage cut of transactions can be lucrative here and acceptable to the customer given the amount of value provided relative to the transaction size.

Depending on the dynamics of each industry, there could alternative models, like health insurance companies potentially covering the cost of a TEM in healthcare or credit card companies covering the cost of a travel TEM as a perk for its members (that gets them to travel and spend more on their cards). Ads might find a way into TEMs focused on providing consumers with information or recommendations, like via sponsored recommendations included in the response to the customer. Flat rates can make more sense than percentage cuts of a transaction when it comes to infrequent processed where the final transaction isn’t conducted through the middleman.

How TEMs can ‘disrupt’ their industries.

These startups can transform their target industries by using 1) an information advantage and 2) collective bargaining power.

1. As TEMs scale to tens of thousands (and perhaps hundreds of thousands or millions of customers), they are collecting an enormous amount of data on those customers, their pain points and desires, and the bahavior of all the other entities in the industry that they interact with. This data provides a distinct information advantage that allows them to create new technology solutions that better serve customer needs. Those new solutions can either rapidly improve their core operations or serve as direct replacements for incumbent’s products. A large TEM naturally starts to encroach on the industry it coordinates a consumer’s engagement with because it sees all the things it could do better and has a large base of customers it can immediately shift over to a new product.
2. The larger a TEM’s customer base gets, the more power it has to pressure other players throughout the industry do what it wants, similar to how large health insurers use their membership size to negotiate the cheapest reimbursement rates with hospitals or how Amazon uses its size to get rock bottom pricing and better service from shipping companies. This bargaining power can force the hand of incumbents with carrots (bulk access to a large, organized base of target customers) or sticks (the threat of being blacklisted, with the TEM moving its customers to competitors that are willing comply).
   The exploitation of its collective bargaining power allows the company not just to reduce costs and increase revenue, but eventually to force the adoption of new standards in the industry. This is especially impactful in industries where the consumer doesn’t actually have a strong seat at the table because payment comes from another party (i.e. healthcare and government services both have terrible customer experiences and little incentive to change that).

Because of this, TEMs can be sleeping giants who appear harmless for years as merely a helper coordinating interaction between customers and industry incumbents, then suddenly become active participants in the market themselves once they’re too big to be quashed.

Moreover, as the field of artificial intelligence advances over the decade(s) ahead, TEMs will increasingly use AI while decreasing the use of human employees/contractors. The TEM companies launching now will have an advantage over new startups that launch in the future with solely AI, because as existing companies they will have years’ worth of data already informing their systems. Similar to how Amazon’s years of data on each customer’s behavior creates a deep, defensible moat against even the best funded potential competitors, TEM companies founded today will have heavily defensible positions in the market down the road when they switch from tech-enabled to purely tech.