Wednesday Jun 18, 2014 by Pat Kinsel - Venture Partner, Polaris Partners
Market sizing is an incredibly important task for both entrepreneurs and investors. It helps you decide which markets to prioritize and helps you better understand how your application [or someone else’s] is fairing in a particular area.
At Spindle, our application needed to be enabled and launched city by city. This is an increasing trend. Uber, Lyft, Drizly, Washio, Spoonrocket, and countless other services need to launch their transportation, alcohol, laundry, or meal delivery service incrementally. Like Spindle, expanding these services requires significant effort in each city. It’s incredibly important to prioritize which cities you tackle first. And once off the ground, it’s equally important to know how much of the market you’ve captured.
When we were trying to create our prioritized city list at Spindle, we looked at every possible piece of data: population, population density, income, inbound migration [Marc Cenedella of The Ladders told me that inbound migration equalled new people looking for jobs/things to do/etc - good tip!], age, and other demographic information. We had a mountain of things to consider, but still felt we had no real insight.
Although we knew how many people lived in a particular city and how many people were of a certain age, the data was disjoint. We lacked the ability to perform complex queries that would consider multiple factors. Applications are often directed toward very specific sets of early adopters and we needed more granular data.
Enter Facebook Ads Manager.
While experimenting with buying ads, it struck me that Facebook’s Ad Manager is the key to market sizing. Facebook’s penetration is so great that it’s safe to assume they represent the US internet population. They know which devices people use and which interests and behaviors people have. What’s more, it’s all easily queryable. Facebook will tell you how many people match your criteria in each city. Because these people are actually acquirable via Facebook ads, I believe this is a tremendous way to prioritize cities.
Let’s say you’re building a dating application that enables people to meet while trying restaurants they both want to visit. You’re located in San Francisco, but think there might be a better city to launch your application. You’re early version is iOS only.
Visit Facebook’s Ad Manager and select App Installs.
Under Audience > Platform select iOS only
Under Interests select Family & relationships > Dating
Under Behaviors select Purchase behavior > Buyer profiles > Foodies
For age, let’s say you require 18 minimum
* There are other interests and behaviors you can select, but these are sufficient for this exercise
Now that you’ve identified the type of people you’re targeting (dating foodies), where should you launch?
Under locations, entering a few cities will reveal:
Now you know which cities have the most people that match your criteria. San Francisco and San Diego are roughly equivalent, but I can guarantee you the cost of acquiring a user in San Francisco is higher than in San Diego.
Once you reduce your list to a few cities, the next step is to run some test ads in each city. Within about 24 hours, you’ll be able to estimate the cost per install for each city. From there, you’ll know which city has the most people and is cheapest to pursue = your launch city.
Finally, these numbers give you a way to benchmark your success. If your dating+dining app launches in San Diego and has 2,000 active users… you’re doing much better than you might have realized without this data.
I should also note this data is valuable for entrepreneurs as they communicate with VCs. If your application is focused on a niche like in this example, you will need to contextualize the number of users you have. Saying you have “2,000 users” or “half of the targeted early adopter market” in San Diego are very different things. From there, you’ll need to experiment with expanding out of the targeted niche into broader audience segments and into launching in other geographic areas.