10 Things I Learned At The On-Demand Conference

Last week I had the chance to attend the first edition of the on-demand conference organised in San Francisco by Semil Shah, Pascal Levy Garboua and Tradecraft, and here are 10 things I learned:

1. San Francisco is currently an extraordinary open-air lab for the on-demand economy: hundreds of startups, investors and service providers are competing for a piece of the action — a whole ecosystem has sprung up in a mere few years, with unparalleled breadth and depth from anywhere else in the world.

2. Broadly defined as any service provided by a human in less than an hour and ordered through a mobile interface, the on-demand model is being applied to every imaginable vertical, from food delivery and transportation to healthcare, education or home improvements.

3. The barriers to entry keep on coming down, with most building blocks of on-demand apps being offered through easy to implement and cheap API-based SaaS products (Parse for mobile PaaS, Stripe for payment, Checkr for background checks, PlaybookHR for contractors management, Sinch for in-app communication, Algolia for Search, Lob for direct mail, OnFleet for delivery, etc.).

4. The more urgent and frequent the need, the more relevant the on-demand model, which explains why food and transportation have been the fastest growing categories to date. Healthcare (doctors consultation, prescription drug delivery, etc.) and enterprise services are being tipped by VCs as the next categories where very large companies are likely to emerge. Services with in-built virality (eg. sending parcels/flowers to someone w. Shyp/Bloomthat) or an inherently social component (taxi transport often has multiple passengers and/or happens in a social setting) will benefit from a strong scale effect.

5. Many subverticals will be too small and with too infrequent a usage to support VC backed companies, but could nevertheless make for good seed investments or lifestyle businesses.

6. Having said that, sizing the market can be a very difficult exercise as, if done well, these services will expand their underlying markets by making them so much more convenient to access.

In 2014, Uber generated $500m of gross revenues in San Francisco, its most mature market, vs. a local taxi market of c. $350–400m before it launched. And it’s supposedly on track to treble again this year.

David Bladow, CEO of Bloomthat, mentioned that they have been experiencing a similar phenomenon — by making it so much easier to get flowers delivered, they’ve turned it from a “special occasion” indulgence into a much more frequent and broader gifting habit.

7. The nature of the legal relationship between these platforms and the workforce that powers them is far from settled and remains a major threat to the model. Ongoing lawsuits are seeking to reclassify Uber and Lyft drivers from 1099 contractors to full-time employees, and similar questions are arising everywhere else, whether in the UK around the so-called 0-hour contract or in France with the auto-entrepreneur status.

8. Competition for the on-demand workforce is becoming extremely fierce, with hourly rates going up and retention a key issue. Employing workers full-time rather than as contractor is already being used by companies like ManagedByQ as a way to both attract better and more loyal people and to offer a better service. As Dan Teran, its CEO and co-founder, puts it:

in the on-demand economy your workforce is your User Interface.

9. City-by-city expansion (and sometimes even neighborhood by neighborhood) is extremely challenging and operationally complex. Startups shouldn’t try to sheepishly copy the Uber playbook of a roll-out of local teams in every city, which requires massive funding, but instead develop their own, finding the right balance between centralised vs. decentralised and in-house vs. outsourced services.

10. New interfaces may improve / transform the space again: voice (eg. Siri), button (eg. Amazon Dash), beacons (eg. Estimote), motion (eg. M8 chip), messaging (eg. FB messager/WeChat), text (eg. Magic), Slack channels…