Tuesday, 1 September 2009

Getting more for your Facebook Ad Money

Customer development is a concept that I wish I had learned about earlier. To suggest that you can test your customer bases response before, and during development and release products multiple times a day with basically no money would have been unheard of only a few short years ago. Testing your market before you jump in is one of the key concepts behind customer development and I thought I'd post a little about my experiences with the "$5 a day" approach to customer testing.

I've been using Facebook Ads for a while now to sell a free service I've worked on (and must get around to updating). This project was always intended as an experiment of sorts to gauge customer interaction and demand but one of the most interesting results came not from the project but from the advertising itself. Namely, how to get the most out of my advertising dollar. We ended up with a per-dollar click-through-rate improvement of  over 200% and we haven't yet started optimizing our ads themselves.

I'm going to assume most people are familiar with Facebook advertising. It, like Google, works with a daily budget and a maximum bid per click system. Also like Google, Facebook makes recommendations on how much you should bid to get your ad placed. We had decided that we were going to stick to a budget of $5 a day I wanted to make the most of our cash.

When we started out, we started with the CTR recommended by Facebook. With our 8 ad and demographic variations,  we landed around 30 clicks a day. It was obvious that as our $5 budget was being hit consistently every day then there must be some period(s) of time in which our ad was withheld due to our budget cap - After all, Facebook can't know ahead of time how many clicks we were going to get on a given day. This assumption was validated in ad traffic patterns as well. It was common to see 4-10 clicks in a period of 5 minutes followed by a long period without any clicks.

This pattern continued around the clock so as to suggest Facebook was breaking our daily budget down into smaller time periods and using up our ad budget in each of those time slots. Based on the way the bidding system worked, I have to assume that lower bids are presented to users after the more lucrative higher-paying ads had used up their daily budgets in each of the time intervals (at a guess - hourly or half hourly). I figured it was easy to put this to the test. Creating a larger set of ads and allocating each of them a maximum bid much lower than the recommended value specified by Facebook - in our case as low as $0.04 for a recommendation of $0.30 - I found that, dependent on the keyword, I could pick up much lower cost per click rates and while impression volumes were much lower, the larger set of ads could make up for this.

[caption id="attachment_43" align="aligncenter" width="444" caption="Facebook ad clicks before and after changes"]Facebook ad clicks before and after changes[/caption]

The end result has been an average of 67 clicks per day this week on an average of $4.10 (0.06) compared to 27.2 clicks for $5 (0.18 per click) two weeks prior to the changes. Thats 200% better cost effectiveness! Of course, I can only hypothesize but my theory is that each ad is picking up the PPC advertising scraps after the higher paying ads are hitting their budget each hour. I assume this trick won't work so well for the more popular keywords. In my case I am assuming no other advertisers are advertising in the space between 0.04 and 0.30 and therefore I have the opportunity to reap the ultra-cheap rewards if I'm willing to take a back seat to the bigger paying advertisers and wait for them to eat through their budgets.  The chances that this might work with more popular keywords is naturally lower but you can try to avoid those keywords and hey, for the struggling startup, anything that can save a few dollars is worth a try. Right?