I wrote a piece on the various Facebook bidding options a few months back and it seems like high time to revisit the topic. Specifically I’m going to share some official best practices for how to squeeze out the most out of Optimized CPM (oCPM) for Facebook pixel conversions, which has been an area generating quite a few grey hairs since its inception.
First of all, let’s reiterate from a high level perspective what Optimized CPM is good for and how it works:
Just like in the case of regular CPM bidding you pay for the amount of times your ad has been delivered (its impressions), but the delivery is optimised by Facebook to meet an objective set by you. That objective can for example be maximum reach, Page likes, or as will be the focus of this article: conversions on your Facebook pixel.
There will always be a target value
When setting up your ad you are required to set a target value for the objective you want to optimise for. I know this fact has caused confusion for some advertisers as Facebook doesn’t ask you explicitly for this if creating your ad via the Facebook ad creation tool, giving the impression that this is somehow done automatically:
However, what is actually happening under the hood is Facebook setting a target bid for you, which will be the same as the suggested bid that’s shown if selecting the second radio button:
So, provided that you don’t change the suggested bid there is actually no difference at all between these two options.
With that quick recap it’s time to go through 6 best practices for working with oCPM for pixel conversions:
1. Find out if your conversion event is a good candidate for oCPM
In order for oCPM to work well there are two prerequisites needed:
Minimum 25 conversions with a 0.5% conversion rate based on 1-day click-through
A certain volume of conversions is needed for Facebook to have enough data to figure out which people are likely to convert. If your pixel does not meet this prerequisite there is the option of starting on CPC and then switching over to oCPM - more about that below.
A budget of at least 5X the target value
As for the budget, the optimisation algorithm needs to buy a certain amount of impressions before determining what constitutes a user likely to convert. If the budget is too small it will not be able to gather enough data before having to stop showing the ad.
2. Know your 1-day conversion rate
Checking your conversion rate for a given campaign, ad set or ad on a 1-day click-through basis requires some custom report pulling from Facebook. To check this for a given campaign, access the “Reports” section of the Facebook Ads Manager.
While selecting which columns to show you have the option to change the attribution window. You want the “1 day after clicking on ad” option.
To see your conversion rate, divide the amount of conversions with the amount of clicks. Remember that conversions should be at least 25, and try avoiding looking at too long periods of time. Don’t select a time period longer than a month.
If your conversion rate turns out to be greater than 0.5% you’re good to go. If it’s lower however, stick to CPC until you have managed to increase it. There’s of course also the option to optimise for an event further up the conversion funnel. More about that below.
3. Set your true bid and focus optimisation on improving quality
Since Facebook will optimise the delivery of your ad to get you conversions at the bid level you set, there is no need for changing the bid - unless how much you value a conversion changes. This leaves room for focusing more on the qualitative aspects of your ad: the creative, the target audience and the placement.
All these factors will impact the performance of your ad so don’t stop iterating - there will always be performance increases to be gained. Comparing apples with apples will be important however, so think about how you structure your campaigns. A good idea is to only have one target audience for each ad set, and then vary the creatives within the ad set.
4. Start slow and iterate your conversion event if necessary
If you’re struggling with getting your conversion event to meet the prerequisites in best practice number 1 above, it might be the case that your pixel is placed where there’s simply not enough action happening for oCPM to get the data it needs. If so an option is to move the pixel further up the funnel to an event which makes a good proxy for the conversion you really care about.
It is also a good approach to start with a bit of caution and not expose too much budget. Give just enough room for the optimisation algorithm to gather the data it needs and increase it as you’re starting to get conversions at a level you’re happy with.
5. Don’t overwork your bidding and budgeting
Regarding bidding and budgeting, there are two things to make sure you keep in mind:
Bid the amount you actually value your conversion to
Bidding higher will end up with too costly conversions for you and bidding lower will not maximise the amount of conversions you could get.
Don’t change your bids and budget more than 3 times per day
The optimisation algorithm needs time to tune in and if its instructions change too often it will have suboptimal conditions for doing so.
6. Keep your audience at the right size
Although oCPM works to find the sweetspots in your target audience, it’s still a good idea to bring your intel to the table when setting up your audience. That way you save the optimisation engine some time (and yourself some money) since you have eliminated audiences that wouldn’t convert anyway. What I’m saying here is that you shouldn’t set your audience to “everyone” just because oCPM will try to find the best converting segments.
That said however, you don’t want to keep it too small either. Why? Because you’ll end up with few or no conversions. Say your average conversion rate is 2% and your average CTR 0.5%. That means 1 out of 10,000 people will convert and with an audience of 100,000 you’d be looking at 10 conversions in total. Although this simple example doesn’t take frequency into account it hopefully gives a decent idea of why it might be a good idea to not starve the target audience too much.
You’re good to go!
I hope these best practices have been useful to you. If you have any experiences of your own that you’d like to share please don’t be a stranger and let us know in the comments!
Connect with Sven on Twitter: @svenhamberg