“Taking the Wrong Lessons from Uber” — C(osts).R.E.A.M. in On-Demand

Addition to Sarah Tavel’s post and Dustin Rosen’s addendum

I recently came across Sarah Tavel’s post on “Taking the Wrong Lessons from Uber” and Dustin Rosen’s addendum. Being an ex-finance and startup operations guy, I wanted to shed a bit more depth to the subject, but from the business side. I work at OrderAhead (YC’11)*, an on-demand delivery service and I believe the delivery game will be won in unit economics.

It has been a gold-rush the past couple of years for startups to copy the “Uber” business model. However, many haven’t considered the underlying cost structure. Furthermore, not all on-demand services are created equal: different industry services have different underlying operational dynamics. Just take a read on why HomeJoy failed. So, what does it take for an on-demand company to succeed?

My two largest takeaways from Sarah’s and Dustin’s posts are:

The challenge I see with so many of these services is that most often, 1) they are new costs, and 2) they don’t fundamentally recast cost structures like Uber did — instead, many of them are an arbitrage on the cost of wealthy people’s time vs the less wealthy.” — Sarah Tavel


“In a way, a better price can be seen as a feature. In my estimation, if your price is 10% better, this is the equivalent of a 2x feature, 20% = 3x, 40% = 5x, etc.” — Dustin Rosen

So if product and price drive the success of on-demand, and price reduction is inversely related to increase in on-demand product experience…how low can on-demand go? In other words, what do costs look like?

C.R.E.A.M. — COSTS Rule Everything Around Me

Let’s look at Sarah’s first point: “They are new costs.” One of the most successful restaurant delivery companies I know of is Grubhub Seamless. The company started out as an online food ordering company connecting people and companies with take-out restaurants (read: in-house delivery capabilities). Eventually the company IPO’d and is now trading around a market cap of $1.6billion.

Scalable software + restaurants with delivery capabilities = no new material costs for customers.

However, Grubhub Seamless’s recent entry into on-demand delivery may have been fueled by heavily-funded startups like Postmates and Doordash, or acquired companies like Eat24 (Yelp) and Caviar (Square). These models have new costs associated with them.

With the old Grubhub Seamless, or any take-out restaurant, you would pay a small delivery fee (usually $1 if memory from my banking days serve me right) and a % tip. However, with most on-demand restaurant delivery services, you’re adding new costs to the customer: the delivery fee ($5–6 or more depending on order size, location, etc.) and the additional mark-up on food (anywhere 1–3%).

Credit: Costkrt

Let’s move to Dustin’s point on price as a feature. If an x% decrease in price to consumer yields some multiple return on better product experience, how low-cost can these on-demand delivery companies go? Let’s simplify the math behind on-demand delivery costs (note: numbers are purely fabricated, do not reflect the company’s, and is only for illustrative purposes). Here are a some assumptions:

  • Minimum wage in San Francisco is $12.25; let’s round that to $14 to account for gas and wear & tear (likely what contracted drivers will compare their opportunity cost of working other jobs to)
  • Most companies strive for deliveries under an hour (assume 1 order per hour)

Let’s also assume a customer orders ~$80 base cost worth of food. Here are the additional cost assumptions:

  • 3% food mark-up = $2.40
  • 9% tax = $7.40
  • 10% tip (post-tax) = $9.00
  • Delivery fee = $5.00 (assume 100% passed along to drivers; some companies charge a higher delivery fee and skim a % from the driver’s pay)

This totals to $103.80! The delivery fee + mark-up ($7.40) equate to an additional 9.3% of the original order, on top of the 10% tip you’re paying. This ultimately becomes, as Sarah describes, “arbitrage on the cost of wealthy people’s time vs. the less wealthy.” The select wealthy will pay for the service but the less companies charge, the fewer drivers will be attracted to the platform. However, also note while many startups start serving the 1%, this purely serves as an entry way into a broader market (e.g. Uber black cars to Uber Pool).

So where can you shave 10% off of the price for the consumer? The $5 delivery fee and the $9.00 tip go to the driver. Cutting 10% here means ultimately paying the driver $12.60, or slightly above minimum wage (read: expect an increase in driver acquisition cost). And while the company still makes a small gross profit from food mark-ups on the delivery, this assumes no refunds, delayed orders, issued credit, free promos, etc. which will all add to the cost of that delivery. Thus on a per order level, it’s tough to improve your product even by “2x” through decreasing costs by 10%.

Ultimately, companies can achieve profitability through 1) optimizing for costs, 2) increasing AOV or 3) increasing efficiency of deliveries/hour. And this is exactly where OrderAhead has been focusing for a while now — optimizing for unit economics and profitability, not growth at any cost. It’s definitely a challenge, but one that will reap big rewards if solved.

Summary: Many startups tried copying the Uber business model into other industries without thinking about cost structures. They ended up creating new costs for the consumer and are unable to reduce pricing to build a better experience. Unit economics are the key to surviving in the on-demand space.

If you enjoyed this read, please recommend ☺
Comments? Suggestions? Questions? Write me a note below!

Want to connect? Timothyxchen@gmail.com

Follow me on Twitter at @timothyxchen for more stories like this on the Art of Business

*Disclosure: All comments are my own views and not an official reflection of the company’s.