Welcome! The Marketplace Theory series on MarketMaker deep-dives into the different characteristics of marketplaces, and how these differences lead to different marketplace dynamics, and require different product interventions.

What is Time to Deliver, and why does it create so many challenges?

In most marketplaces, the timeline of a transaction can be broken down into three stages:

• Time to Match: How long it takes for a buyer to be matched to a seller. On Amazon, for example, this is the time from when a buyer starts searching for something to the time they check out. On Uber, this would be the dispatch time - the time between a rider ordering a ride, and when a driver is matched to them

• Time to Deliver: This is the time gap from when the match is made, to when the buyer actually receives the good or service and can start using it. On Amazon, this is the shipping time, e.g. 2 days if you used Prime. On Uber, this is the time it takes for the driver to drive from their current position to pick you up, typically about 5 to 10 minutes.

• Realization of value: This is the time in which the good or service is finally consumed, and thus when the value of the transaction is realized.

  • On the demand or buyer side, this is best thought of as “Time of active consumption”. For example, if you bought a pair of hiking shoes on Amazon, and you hike 7 days a year and expect the shoes to last for 5 years, the time of active consumption is 35 days. For Uber, this would be how long the ride lasts, typically about 15 to 45 minutes.

  • On the supply or seller side, rather than thinking of this as a time dimension, it’s often more helpful to think of this as a reward function. For example, for an Amazon seller it would be the amount of money they made from that sale; for an Uber driver, it would be how much they earned from that trip.

Many marketplaces (understandably) invest a lot in their matching experience, but it’s a mistake to forget about time to deliver, as this is exactly the spot where a lot of things get borked 😱:

• On Airbnb, a guest may have to change their vacation dates due to a sudden work emergency

• On Amazon, a buyer may cancel their purchase

• On Taskrabbit, a Tasker may forget that they had a job scheduled that day

• On Wonolo (a temp staffing marketplace), an inventory delivery for the warehouse may be late, and so the entire shift is canceled

Your technology needs to gracefully handle these changes, or enable users to quickly flag that something has happened. Some kind of fee or compensation may also result, since an agreed-upon contract fell through. Lastly, since this also breaks the match between the two halves of your marketplace, and you will often need to re-match the sides, for example, by opening the Airbnb property up to other potential guests.

Beyond the user experience with your tech, there’s also a lot of annoyance or change cost to your users - think of the Airbnb host who now has to reschedule when the cleaners come, or the Wonolo worker who may have to defer car repairs until next month. Pay attention to that emotional annoyance cost when designing a solution.

How may we solve this?

When should we even bother to solve this?

If you have a very short time to deliver, it may not even be worth investing in this, to be honest. You should consider prioritizing product work here if (1) you have a long-ish time to deliver (on the order of days), and (2) the root causes are relatively concentrated

Uber and Lyft, for example, have a time to deliver of maybe only about 10 minutes - and that’s just not a lot of time for things to go wrong. Sure the driver can make a wrong turn, or maybe even get into a car accident, but 99% of the time things probably go smoothly, so product work there is not particularly valuable.

If root causes are concentrated, that also makes a systemic product solve better, versus the alternative of simply pointing your users to the support team. For example, if the top 4 root causes drive 80% of cases, that’s a good sign to invest; if your top root cause drives just 3% of volume and over 100 root causes drive 80% of cases, you’ll need to develop dozens of different features to make a noticeable dent on the issue. This concentration dynamic probably drives some of the examples we see - e.g. on Amazon, buyers changing their mind about a purchase probably drives a significant percentage of order cancellations, so it was worth it to build out a self-serve order cancellation flow, even though time to deliver is fairly short.

3 possible product solves

We’ll focus here on solutions that keep your marketplace working - some kind of match or transaction still takes place. There’s a lot of other solutions, e.g. building self-serve cancellation flows, but those tend to be more obvious and won’t be discussed here.

1) Reminders and intention checks

This is exactly what your dentist does when they call you to see if you’re coming next week - users are sent a reminder, or the platform confirms that their intention to complete the transaction remains the same. A couple of examples:

• Airbnb sends a reservation reminder email campaign about 1 week before your stay

• On Amazon subscribe and save (a case where the original match / subscription order may have happened months in the past!), Amazon sends a weekly reminder email for you to adjust your deliveries

• eBay asks sellers to renew their listings every 30 days, which checks that the seller is still active

• At Wonolo, we built SMS reminders for workers for upcoming jobs, and workers could reply via SMS if they could not make it

2) Cancellation fees, a.k.a. aligning monetary incentives

Cancellation fees are another great way of making your match stick, and they have the additional benefit of ensuring that users (both buyers or sellers) are compensated fairly in case of an exception. Fees make your users put some skin into the game, and that monetary pull naturally drives better marketplace behavior. Some examples:

• Amazon will sometimes tell you that it’s too late to cancel your order, and you might be charged anyway (side note: since returns are free, I wonder if this really has any marketplace effect)

• Airbnb and most other hotels will have free cancellations if done far in advance (where the host/hotel has a good chance of finding another match), but will charge a cancellation fee if the cancellation happens too close to the date of stay (where the host/hotel may not be able to find a replacement guest)

• In an example from a slower market with a bigger price tag, remember that Elon was supposed to pay a $1 billion fee if he backed out of buying Twitter

3) Overbooking

Overbooking is when you match a higher quantity than what the buyer requested. For example, airlines will sell 275 tickets for a 250 seat plane, because they know some travelers will drop out as the departure time approaches.

One big caveat is that overbooking is only useful if your marketplace has perishable demand - if the trip/stay/flight/job needs to happen on that exact date and time, and cannot be time shifted. Uber rides, Airbnb stays, and even car rentals are all perishable - a Getaround booking for a certain car from 1pm to 2pm is completely different than a booking for that same car, on the same date, but from 3pm to 4pm. In a non-perishable marketplace, buyers or sellers simply absorb the incident - Amazon, for example, would never send you two pairs of hiking shoes if you bought just one but they were worried that you got the size wrong (that example does make me wonder if Amazon should actually start doing that, though).

You’ll also need some way to deal with any shortages, and to do so in a way that doesn’t permanently damage your relationship with your users. This is why we see overbooking happen more in airlines, where flyers can be tempted onto a later flight in exchange for a $200 flight voucher; but why we don’t see it for cases like hotels, since it would be a pretty crummy experience to be told that you have no lodging for the night in a strange city.

Startups often convert of time to match into time to deliver/realization of value

One final interesting point I’ll note about the 3-stage breakdown is that this can be a way to think about how a startup disrupts incumbents. Startups often take something with a long time to match or a bad match process, and exchange that for a longer time to deliver or lower realization of value.

Before Uber, you would spend 10 to 30 minutes frantically trying to wave down a cab before finally getting one. Sure, the time to deliver is shorter since the taxi is right there, but you probably hated the whole experience of finding a cab, and especially the anxiety of not knowing if you would even get one or how long it would take. Uber greatly reduced the time to match and the anxiety of matching, in exchange for a longer delivery time, but delivered a much better experience overall.

Similarly, buying an item online from Amazon or another retailer represents a reduction of time to match (versus schlepping yourself to a store), in exchange for a longer delivery time. Buyers also get a better selection online, or can price compare. Like the Uber example, this represents a trade between time to match and time to delivery; and like the Uber example again, the buyer is not just trading off the time of each segment, they’re also receiving other benefits such as a reduction in anxiety or frustration, or better selection.

A more interesting and recent example is TipTop, which offers to instantly buy any unwanted electronics you have, rather than having to find your own buyer on your local online second-hand marketplace of choice. This example offers a bit of a different spin - while it still reduces time to match (and reduces the annoyance/frustration of the match process), it instead reduces the value realized by the seller, instead of increasing the time to deliver.

Lastly, even though time to deliver may increase, that time should not be thought of entirely as a deadweight loss for buyers, since some of that time is multi-purpose or inactive. When you request a ride home from a restaurant, for example, you likely do it while still sitting at the table and finishing talking with your friends; when you wait for an Amazon package, you’re not sitting by the door for two whole days. That can make the tradeoff more attractive for buyers.

If you’re looking to build a new product, it might be very worthwhile to think about how you can use this tradeoff dynamic as a wedge against incumbents.

Thanks for reading! Subscribe below to make sure you get each new article from MarketMaker. I also welcome comments, corrections, and feedback - connect directly with me on LinkedIn for that.