An important Marketplace Metric: Search to Fill

On a recent trip, I experienced something on the Uber app that I don’t recall seeing in recent times. After a minute or so of waiting, I saw this on the app:

My assumption is that in a high density area like Boston, it is quite rare that Uber is unable to match a rider with a driver. And with incentives like Surge Pricing, it is a matter of time and the right surge incentive to get a driver to accept the ride (I eventually was able to get a ride for ~$90!)

“Search to Fill” (or “Fill Ratio”) is a critical metric to track in a marketplace. This metric is simply measured as the percentage of searches or requests that result in completed transactions.

For example,

  • On Trusted or AirBnB or Pared, this is the percentage of search sessions that result in bookings.
  • On Threadflip, this was the percentage of search sessions that resulted in a purchase.
  • On or Thumbtack, this is the percentage of job postings that result in a candidate getting hired.
  • On Uber or Lyft, this is the percentage of ride requests that result in rides.

Generally the higher the fill ratio (at critical mass), the better. But the ratio helps marketplace operators really understand which side of the marketplace to focus on.

Type of Marketplace

Once you have been able to categorize your marketplace, you are able to understand how you can leverage various aspects to increase your marketplace’s Fill Ratio. Josh Breinlinger has a great post on marketplace categorization:

- Double Commit Marketplace

Examples:, oDesk (now Upwork), Thumbtack.

A typical workflow: Buyer posts job. Buyer invites candidates AND Candidates apply to job. Interviews happen. Buyer makes a hire.

- Buyer-Picks Marketplace

Examples: Trusted, YourMechanic, Airbnb

A typical workflow: Suppliers enter availability. Buyer can see available suppliers. Hires a supplier without discussion.

- Supplier-Picks Marketplace

Examples: Uber, Lyft, Doordash, Postmates, Pared

A typical workflow: Buyer posts job. Approved suppliers see available jobs. Supplier claims job.

Measuring Search to Fill

At Trusted, we measure Search to Fill religiously. Here are just some of the Looker charts we look at regularly:

  1. Search to Fill Overall (Monthly & Weekly)
  2. Search to Fill by Hour of Day/Day of Week past N weeks (rolling)
  3. Number of Unique Search Sessions by Parents (Monthly & Weekly)
  4. Search to Fill broken down by Geo (Monthly & Weekly)
  5. Search to Fill Lead Time Overall past N weeks (rolling)
  6. Custom Charts that display specific Days or Hours (or both) when Fill Ratio is known to be low (this informs our dynamic pricing)

And some painful charts:

  1. Unfilled Searches Overall by Hour of Day/Day of Week past 4 weeks
  2. Unfilled Searches by Geo by Hour of Day/Day of Week past 4 weeks

Time to fill

Using Josh’s nomenclature, in Double Commit and Supplier Picks Marketplaces, the Time to Fill is a very important factor. Especially in marketplaces where multi-tenancy is common (for example, Uber Drivers may also drive for Lyft), this number is critical to ensure liquidity, retention and satisfaction in the marketplace.

In a “Supplier Picks” marketplace, where the supply side is largely fungible, what could be the reason that the fill ratio wouldn’t be 100%? On Uber or Lyft, it could be that your driver is too far away which causes the rider to cancel. It also could be that the driver or the rider’s rating is low enough to demotivate the other party from engaging.

Pivoting on Geography (or other variables)

While simply looking at the overall Search to Fill can help you understand the health of the business, it may not give you an accurate way to find out what operators need to tweak in order to reach a high fill ratio.

At Trusted, we pivot our Search to Fill numbers on geography as well as by other factors that inform the search criteria, such as location, age of children, pets, rating, number of ratings, allergies etc. Similarly at AirBnB, this could be location, number of rooms, amenities desired etc.

Improving Search to Fill

The first step is to really understand your marketplace and the variables that drive the fill ratio. Let’s take AirBnB as an example, a certain market could have lots of 1 bedroom places for which fill ratio is looking strong. But if the demand for 2 bedroom places is high, how could that inform pricing?

1. (preemptively) Informing the supply about the opportunity

If you have the requisite supply, informing your supply about the requests on the demand that are going unmet could be a good way to increase Search to Fill. Here is an example of an email we launched to incentivize Trusted’s dormant caregivers to work (we credit AirBnB for this idea)

2. Dynamic Pricing

In addition to seeing where high demand is, Uber and Lyft drivers are incentivized with surge pricing. They can see which zones have higher multiples (or surges), presumably since demand there is high.


3. Offering Alternatives

Marketplaces like Opentable and Resy offer suggestions when your exact requirements cannot be met. Recommendations are a great way to ensure a user’s search has some great alternatives.

As Li and D’arcy point out however, offering such alternatives largely depends on the type of marketplace:

“homogeneous supply” marketplaces typically hit an asymptote in network effects, where the value to users eventually plateaus with greater market depth. For instance, if there were 6 Lime scooters on a city block near me, this is no more valuable than if there were only 4 or 5 scooters available for me to use in my vicinity — user value is unchanged despite the addition of more supply. On the other hand, for heterogeneous marketplaces, there is no asymptote because every node on the supply side is different and potentially can add greater value. In the Airbnb example, a user’s tastes may be quite specific, so every additional listing on the platform is useful to see.

Search to Fill should be top of mind for anyone building a two-sided marketplace.

Below are some amazing articles that share the best marketplace metrics and other KPIs to look at.

Anand Iyer

Special thanks to Li Jin, Andrew Chen, Dave Lu for reviewing this post