Facebook promoted posts – we’re a believer!
With the recent flooding in southern Alberta (BC, and Ontario), many of our policyholders were inquiring if home insurance protects against flood damage. In addition to responding to those individual inquires, we thought it would be worthwhile distributing the answers more broadly. Social media was a logical channel.
For us, social media generally means LinkedIn, Twitter, Google+ and Facebook. When you post a “status update” to Facebook, it uses an algorithm to determine whose newsfeeds the update is going to show up on. The more people that interact with your post (like, add a comment, etc.), the more people Facebook will show that post to.
Over 1,200 people have “liked” us on Facebook, but only around 200 people actually see our updates in their newsfeeds. Obviously this really limits our reach, and makes it hard for us to get our message out to our “fans”. Not to worry though, Facebook has a solution to this, and not surprisingly, it costs money. 😉 Now that Facebook’s a public company obviously there’s a lot of pressure to monetize their massive user base, and one way they’re able to do this is through “promoted posts”. It’s a pretty simple concept: if you want your post to be seen by more people, you can add a budget to your post. How far that budget goes depends on how many other people are paying to promote posts at the time. That means the number of possible “impressions” are divided amongst the number of promoted posts to be displayed, and whoever bids more gets a higher percentage.
On a whim, we decided to give promoted posts a try, for this particular flood post. Initially, I just added $20 to our Facebook account, and fired up Google Analytics (Real-time). Almost immediately, I saw a trickle of traffic coming from Facebook to the link we were promoting. The post got its first few “likes” and “comments”. Since I’d promoted the post on a whim, we hadn’t really prepared ourselves to respond to the incoming comments. It was Friday evening, and Daniel was out of the office, doing the grouse grind.
Comments are great because it means that the post is then likely going to show up in the newsfeed of the commenter’s friends. That said, I admit I started to panic a little, when some of the comments being posted had a negative tone. At Square One, we really try to be open and transparent, so Daniel raced back from the Grind, and took the time to respond to each comment. Almost magically, the tone quickly turned around. It’s amazing how people respond when they realize their dealing with a real, live human, not just a faceless company.
With the initial $20 spent, we took a look at whether the promoted post was offering good “value”. Generally online ads are measured as a cost per thousand views (CPM). As you can see from below, for the initial $20 spend, our cost was $2.94 per 1,000 views:
All views are not created equal however! We’ve done quite a bit of online advertising, and we’ve gotten the cheapest CPM with the Google display network. Our average CPM with Google is around $0.60. That would be great if the “engagement” with those ads wasn’t so weak. I think most of us are now conditioned to completely ignore display ads on most web pages. Our best measure for engagement is whether someone clicks on the ad. With Google, our click through rate is about 0.16% – yes, that’s only a fraction of 1%! As you can see above, with the Facebook promoted post, we were initially seeing a click through rate of 2.6%, and it got even better over time.
Based on the metrics noted above, we figured we were getting pretty good value with our promoted post, so we figured we’d let it ride. It was a bit addicting to watch the traffic stream into our site. We increased our budget for the promoted post several times, and ended up capping it at $250 per day (for 3 days). It was incredible to watch the traffic to our site spool up almost immediately after increasing our budget. It gives you a sense of the massive number of people using Facebook at any given time. For example, for awhile we were averaging over 300 visitors from Facebook on our site at a time. Given that only around 5% were clicking on the post, that means Facebook was showing the post to over 6,000 people at the same time. Keep in mind, the post was only being shown to people in BC, Alberta and Saskatchewan!
Here’s the promoted post’s stats at the conclusion of the campaign:
We actually found that Facebook underestimated the number of clicks that were sent our way. All told, Google Analytics counted 17,768 clicks from Facebook to our flood damage page. This translates to an incredible 7.8% click through rate (based on the 229,175 “impressions” Facebook indicates the post received)! At $0.04 per “click”, it means that traffic was super cheap to generate. (In comparison, our Google display network ads work out to around $0.75 per click). Our site is targetted to a pretty specialized audience – people in BC, Alberta and Saskatchewan looking for home insurance. So, it’s not normally a high traffic site by any stretch. Nonetheless, 17,000 unexpected visitors over a 3-day period gave our server a bit of a workout. It also made our regular daily site traffic look like a rounding error:
There you have it, all the gory details of our “promoted post” experiment. If you’re as obsessed with monitoring and optimizing website usage as I am, I hope you found this post interesting.