A guest post from David McKinney – CEO/Founder of Jammbox.
DIRTY HACKS FOR STARTUP FOUNDERS
The aim of the dirty hacks series is to help you launch your startup faster, cheaper & dirtier. Everything is rough, crude, and fast. Just what you need for your startup.
Part 1: How to use Facebook to size a market
Ok. So you’re assessing a new startup idea for potential, or you’re putting together your first pitch for an investor. You need to know the market size, right? Not only should it be at least 1 gazillion people so the VCs think you’re hot, but you need to quantify it in terms of TAM, SAM, and SOM. So how do you do it? Some people trawl through market reports or pay for specific market segment data. But this doesn’t actually give you what you need. You don’t get anything that means much in the real world. Instead, here’s a hack to get a rough feel for the market that you want to enter. It’s free, and it is easy. All you need to do is dig into the Facebook Ads platform. You’ll be done before you can type N-i-e-l-s-e-n on your iPhone.
Background
Facebook ads can be great for highly targeted advertising, but they also give us a quick way to quantify specific market segments by using a combination of the location, demography, and likes information of Facebook users. We can use this information to build useful market segment profiles and calculate the size of a potential market. What you get is an actual number of real people who have specifically stated that they like your kewords. You won’t get these data from a market report.
How to do it
- Login to Facebook
- Click on the “Advertising” link in the footer of your profile
- Click the “Create an Ad” button
- Complete the Design your Ad form (use any website, it doesn’t matter)
- Start changing the settings on the “Targeting” screen, and you’ll see your Estimated Reach update dynamically
- Choose Locations
- Choose Demographics
- Choose Likes & Interests (previously known as Keywords)
- Choose Connections
- Choose Advanced Demographics.
- Now you have your final Estimated Reach.
- NB: Don’t delete your settings once you’ve got your Reach. Save it as an ad campaign so you can run it when you start testing your MVP. (Use that $50 free ad credit voucher that Facebook sent you).
- Now get out of the building and go do some customer development
Caveats
Using Facebook for sizing markets is rude and crude. There are all kinds of inherent data biases, so be mindful when drawing conclusions from the data. For example:
- Facebook users are not a random, representative sample of the general population. This is a key point if you want to make some kind of inference about the greater population. While Facebook does have a very large sample size (500+million users) it does not accurately represent the world’s population, nor does it necessarily represent your market at all. You must ask yourself if the Facebook user base accurately reflects your market. That said, sub-sampling for results within the Facebook demographic is valid, where you believe that your market has strong crossover with the Facebook population. For tech startups, it is probably correct to say that some segments within the Facebook market are an excellent representation of your market. In terms of sizing your overall market, let’s be clear that you can’t extrapolate from Facebook to another market. It is entirely flawed to say that 65 536 out of 500 million Facebook users like a product, therefore 65 536 *12 = 786 432 out of 6 billion humans will like that product.
-
Facebook “likes” are also not random or representative. “Likes” data are based on what people say they like, yet many users don’t use the likes field, or they use inconsistent keywords, or their likes are biased selections. (Conversely, data such as location, gender, and age typically have a much better signal to noise ratio).
-
We can only pull broad data sets using the current implementation of the Facebook Ads platform. This is because the Likes & Interests field uses the Boolean OR operator, instead of AND. For example, entering “iphone, mp3” will return the Estimated Reach for “iphone or mp3”, not “iphone and mp3”. Just get smart with your likes terms and you can get great data. NB: When Facebook roll out the AND operator, this hack will improve 10x.
-
I’m not even going to start talking about null hypotheses, Type I and Type II errors, and all the other statistical funkiness. This is a dirty hack. Your stats professor will not give you an A+.
-
Lastly, once you’ve pulled your data from Facebook, you should also read this: http://cdixon.org/2010/04/03/size-markets-using-narratives-not-numbers/
Conclusion
A final note: market numbers don’t actually mean anything. Instead you need to go and talk to real customers. Use the data from Facebook as a proxy for your market, and then go and test your assumptions with real customers. If you haven’t got them already, go and buy the insanely good Customer Development book by Brant Cooper and Patrick Vlaskovits, the brain tweaking Four Steps to the Epiphany book by Steve Blank, and the piping hot Lean Startup Book by Eric Ries. And if you aren’t following @hnshah and @kissmetrics you should be.
What are your experiences with using Facebook as a dirty hack to size a market?