Whenever people talk about Facebook Ads (or even more generally “Can Facebook justify their $50bn valuation?”) the conversation always tends to swing around to how poorly Facebook Ads perform.
That drives me crazy.
For the last four months Facebook Ads have been the single best source for paid traffic for my startup CoderStack – I pay less for better converting traffic.
This wasn’t always the case. When I started out running Facebook adverts I saw low click-through rates (CTRs) and expensive costs per click (CPCs). Initially I was left with the same impression as everyone else: this doesn’t work.
But part of me didn’t quite believe that, so I set out to read everything I could on Facebook Ad optimizations, and when I discovered that wasn’t much, I started putting time (and money) into running experiments. After about two months worth of experiments my ad performance had increased dramatically (literally a hundred-fold). I’d discovered that with a little work Facebook Ads can be hugely profitable.
I gave a talk on this topic at the London ProductCamp in February and more than a few of the audience were surprised at how much difference tweaking ad copy and targeting can make. I think what convinced people at the talk more than anything was the hard-data, when I said “I’m paying 1p/click” that’s when the audience really started paying attention.
Unfortunately there’s very very little hard data out there to convince people about how effective Facebook Ads can be, while I was willing to share my real-world performance data at an unconference I’m hesitant to publish it in public where my competitors could use it to compete against me.
Given the inherent commercial nature of advertising everyone else seems to have the same opinion on secrecy so very little real hard data gets published, and most of the data that gets published tends to be from ad campaigns that haven’t done well (as the information doesn’t give competitors any advantage).
To rectify that (and also to test my optimization skills) I decided to run an ad campaign from scratch for one of my side projects (my webcomic Theory of Geek) and publish the data.
In the interest of openness I should declare at this point that I got approval from Facebook prior to publishing this data, but that they didn’t see this data beforehand, they just knew I was writing an article about Facebook Ads and wanted to publish ad campaign data.
Here’s the screenshot of the Facebook Ads Campaign five days after I started it (I’ve blurred out some of the ad names but none of the other details have been changed):
Photo: Imran Ghory
The important column to look at is CPC as that’s how much we’re paying per ad click, the worst performing advert was at £0.58 and the best performing was at £0.003. That’s a difference of 200x. For those of you who aren’t familiar with buying ads, a CPC of below £0.03 is practically unheard of on any other ad platforms, and what we’re getting on Facebook is 10 times cheaper than that.
Of course you could argue that as I’m advertising a webcomic I’m going to get a much higher CTR (and thus lower CPC) than if I was advertising something more “serious”.
For my software developer job board to target software developers I am paying more than in this ad campaign, my best ad for CoderStack was 3x more expensive than my best ad for Theory of Geek. But despite that I’m still paying an order of magnitude less on Facebook than I’m paying on any other ad networks.
I’d also like to comment on how this entire ad campaign and all the optimizations were done with a budget of under £20. Even if you’re running a bootstrapped startup, you’d have to be crazy not to at least experiment with buying Facebook Ads given the low cost of experimenting and the high potential upside.
I plan to write some more articles on Facebook Ads, covering the practical side of doing ad optimizations and also what I think Facebook should be doing in order to make their ad platform better for advertisers, if you’d be interested in reading them then follow me on twitter @imranghory where I’ll post links to my articles as I write them.
UPDATE: Just to clarify a couple of points that people have asked about:
- My ad campaign targeted only the US and UK markets, CPC was slightly lower for the US than for the UK due to a higher CTR.
- Overall engagment seems to be fairly high, on par with organic search traffic. Bounce rate was 35%, and non-bounce visitors viewed an average of 4.5 pages. For CoderStack the figures are in the same sort of ballpark.