All of this is critically important to advertisers, if they want to:
- Know what results you’re getting
- Optimize to get even better results
- Get the lowest cost per click, cost per fan or even highest return on investment possible
This article will discuss one third-party platform that makes that easier. But not everyone will be able to take advantage of this platform, so I’ll also give you some tips anyone can follow to improve their Facebook advertising results.
I was on a Facebook advertising panel at AF Expo with Rob Leathern, founder and chief executive officer of XA.net, and my ears perked up when he started talking about multivariate testing (MVT).
That’s something we’ve seen multiple companies provide for search marketing advertisers, but I hadn’t yet heard of a platform that made this possible with Facebook ads. XA’s Facebook platform Optim.al does.
So, I interviewed Rob to find out more about Optim.al and what it might be able to do for you. And he gave me access to Optim.al to run my own tests, which I’ll discuss below.
More Conversions, Clicks and Fans
With Optim.al you can optimize your Facebook ads to improve conversions, clicks or connections (fans) – that means it helps you optimize all the key Facebook advertising strategies, from ecommerce sales to fan growth.
“Optim.al is a workflow tool,” says Rob, “but more than that, it’s meant to help you understand Facebook data and audiences, and leverage insights from that data.”
Optim.al also provides conversion tracking, which Facebook eliminated from their ad platform in September 2010 (after debuting it just seven months earlier).
Conversion tracking lets you see which ads gets you more sales, leads, or whatever other website-based result you’re looking for; and which ads don’t work. You can also set it up to report the exact revenue from your conversions so you can find the ROI. Optimizing ads according to ROI is gold.
Optim.al is the Facebook Platform. XA.net, the company behind the Optim.al product, has served 70 billion (non-Facebook) display ad impressions. Optim.al was released in October, 2010.
Currently, it only works with Facebook advertisers who spend more than $10,000 per month. It also works with a few select start-ups at a lower monthly minimum.
Bigger Profits and Bigger Audiences
What kind of improvements do businesses see from testing and tracking conversion rates with optim.al?
- From best to worst ad, just by testing creative with Optim.al, they’ve seen a 700 percent increase in click through rates and conversion rates. Some ads completely miss the mark. Finding that sweet spot and best ad is not just about creative, but the combination of the right creative and right audience provides the biggest gains.
- But another problem is to increase audience size while maintaining profits. It’s great if you’re getting a high ROI from an ad, but if that ad is only reaching 2,000 people, that’s not so great. The optim.al platform has also been able to increase audience size by 400 to 500 percent while maintaining ROI.
- Some optim.al users are getting better results Facebook ads than they do with paid search (i.e. Google and Yahoo/Bing).
- One large company started with a $1.00 per fan goal. They used sponsored stories to drive the cost down from 75 cents to 43 cents per fan in 10 days.
I found in my first multivariate test that my ad cost per click was 15 percent cheaper with optim.al (I tested it against another campaign in which I created similar ads via my normal process). I think this 15% savings is partly due to the power of MVT, and partly because it forces you to test more creative.
Even if optim.al did nothing else, the fact that you have to test more images and copy leads you to get better results.
In fact, I was totally surprised which image did the best. On reflection, I’m not surprised that I was surprised — because I’ve seen over and over in Internet marketing that even experienced marketers’ are unable to accurate predict what customers will go for.
Testing is critical. Creativity is required, or you’ll have nothing to test. But if you don’t test, you won’t get as good results. If you don’t test and your competitor does, you’re at a disadvantage.
I asked Rob if there was a simple way to explain how you can test so many creative variations via just 18 ads? He replied:
Multivariate testing eliminates redundant combinations. It designs experiment in a statistically correct way. The mathematical model removes unnecessary versions you might otherwise create.
Is Facebook the Best Creative-Testing Platform?
I asked Rob what he thought about Facebook ads versus display ads in general. He pointed out that the ad environment in Facebook is more homogenous; the ad is always shown in a standardized set of locations, so there are fewer variables to account for.
You also get more robust demographic data. You can create lots of Facebook ad versions without a designer, so overall you can test more ideas.
The upshot is that Facebook is the best place online to test marketing creative. You can create smaller scale experiments with optim.al, then expand your buying based on those experiments.
Rob adds, “Some companies find that they can test people’s response to copy and imagery on Facebook, then use those in offline collateral and display ads, and that these work much better than average.”
That makes Facebook advertising the marketing laboratory from which all the best marketing creative is deployed. That could be quite a shift in strategy and process for some companies, but it makes sense.
Trends in Social Advertising
When asked about the latest trends in advertising, Rob said:
We are so early in this process. There’s so much development to come in optimizing advertising. Facebook continues to improve ability for people to run ads on their system. The pace of development is quick, and quickening. But people still have little experience in this area. Things change daily and weekly in how Facebook looks at it and how people want to use it.
Many TV ads use only URLs from Facebook in their ads now. There’s so much data in Facebook to target, but there’s so much room for more data, especially users 35 and up, because they share only half as much data about themselves as the average 18-to-24 year old does. There’s been large growth among older users, but we’re still in the first inning of social media advertising.
I’m interested in how can we get more synergies between Facebook advertising and the rest of the display advertising world. We’d love to see likes in regular display ads, for example in an ad on Yahoo.com. Everyone wants to extend the reach of Facebook data outside the walls of Facebook.
What If You Can’t Afford optim.al?
The takeaway here is to test more ads.
- Test four to six images
- Test several different headlines
- Test several ways to say your body copy
- Test multiple demographics and precise interests
The hard part to mimic is testing all that with just 18 ads. But I guarantee you that if you run just two or four times as many ads as you currently are, you’ll find better performing ads.
The fundamental lesson I learned was not to assume I knew which image was best. Run some images you don’t think will work, too.