What do we mean by “conversion”?
In its most simplistic terms, a “conversion” is the ultimate action we are trying to get users to take on our ads. For Facebook ads in particular, this is most often an action that happens on our website; anything from visiting a particular page, to signing up for our newsletter, to completing a purchase.
How are conversions tracked on your website?
To track website conversions, you’ll need the Facebook pixel installed on your site. For the purposes of this post, I’m going to assume you already have this set up; if not, you can view some very helpful directions here.
What does we mean when we say “optimising for conversions”?
When creating an ad set, Facebook asks to make an optimisation event choice. This choice tells Facebook’s delivery system (this is the system that essentially runs through thousands of data points to decide who Facebook should show your ads to, when they should be shown and where to show them and what result to try to get for the advertiser). For example, some advertisers may want to optimise for link clicks while others may want to optimise for video views. One advantage of implementing a pixel is that you can tell the delivery system to optimise not just for link clicks to your website, but for an action taken on your website.
Before you start optimising for conversions, you need to understand a few points:
The types of conversion events available to you. You have to pick a specific conversion to optimise for – you can’t just optimize for conversions in general. Picking the right one is critical.
Conversion windows. In addition to selecting a conversion, you have to select a conversion window. This tells Facebook which conversions to use for delivery optimisation. It’s really important here that you understand that this is different to attribution windows. Conversion windows are used for delivery optimisation… remember, this is what informs Facebook on where, when, who and what to show when advertising. This is different to the next point…
Attribution windows. Attribution windows are the reporting system’s equivalent of conversion windows. Get that? Attribution windows relate to reporting… not to delivery. But they do help you understand your delivery results. To understand the results of your conversion optimization campaigns, make sure you select the attribution window that aligns with your conversion window. This is also crucial to making sure you’re getting enough conversions within your conversion window for Facebook to optimise correctly.
Say what? I didn’t know Facebook needed a minimum number of conversions to optimise correctly?
That’s not all that surprising. In fact, I find a lot of Facebook advertisers don’t understand this point, which is concerning as it’s absolutely fundamental to a campaign’s success. But yes, Facebook themselves state that they need a minimum number of conversions per week for their system to have enough data points to optimise correctly.
Are you ready for that number?
It’s 50 conversions per week. Minimum.
Facebook’s systems are smart, but there are certain laws of statistical probability that it can’t get around. And this “50 conversions per week” rule? That’s one that they’ve identified they need to hit statistical confidence levels.
Directly from Facebook, within ad manager, it states this, but it’s in a tool tip that you might miss if you weren’t paying attention. But I’ve pasted it below so you can see their own words:
When optimizing for conversions, we recommend choosing one that happens about 50 times per week at a minimum. Our system needs that many to learn from. Our system needs to learn so it can deliver your ad to the right people at the right time to get you the best results.
Important: In order for a conversion to count towards your 50, it has to happen within your chosen conversion window . For example, if you have a 1-day post-click conversion window for purchase conversions and someone clicks your ad and completes a purchase 3 days later, that doesn’t count towards the 50.
So at this point, if you’re running conversion ads in Facebook, I’m going to encourage you to think really hard on this an answer: “Is my chosen conversion happening enough times in a week for Facebook to actually do their job properly?”.
If your chosen conversion event (let’s take purchases, for the sake of illustration) is not happening enough times per week, you really have two options:
- Take your chances, and still ask Facebook to optimise towards the purchase event. There’s nothing stopping you doing this, but if you do, you really need to understand that you’re asking Facebook to do this without them having enough data to really have a good go at it. You could get lucky… but it’s more likely you won’t see the results you’re hoping for. And if you don’t, the important part here is to understand why: it’s not that “Facebook ads don’t work”… it’s that you haven’t given them the best possible chance of succeeding.
- Your second option is to reconsider what you’re doing and take a longer term approach. Taking this tact will allow you the time to try to give Facebook more data points, and to consolidate their delivery optimisation options before jumping straight into the “purchase” event. I’ll admit that this is not usually something a business owner wants to hear… they want to get sales as quickly as possible, and I understand that. But you also have to weigh up the benefits of throwing money at a purchase campaign too early (and possibly getting some sales, but likely losing more on untargeted ads) vs taking a smaller hit upfront (while you gather data points) and then using those data points for a more optimised push later.
I grant you, it’s not easy… but Facebook are upfront with you about the fact that they need a minimum of 50 conversion events per week for their system to learn enough to optimise correctly. My personal suggestion is that you understand this, before making any decisions on your campaigns.
Ok… but I’m definitely not getting 50 sales a week yet, but I do want to (eventually) optimise towards sales. What do you suggest I do then?
There’s a few options you have here if you’re in a situation where your Facebook pixel does not have enough conversion data per week. They may sound counter-intuitive if you’re looking for sales (as in our example), but they’re all about providing more data to Facebook:
- Optimise for a more common conversion. Yes, it may be tempting to jump right to optimising for purchases, assuming that’s what you ultimately care about. However, purchases are the rarest type of conversion. By optimising for a more common one instead, you’re allowing Facebook more data points to work with and – if you’re getting the required 50 a week – you’re giving it the amount of data it needs to start learning and optimising. So, instead of optimising for purchases, consider optimising for “adds-to-cart” or “product views”. Both events are more likely to lead to a purchase than, say, someone engaging with a post on your Page, but may occur frequently enough that Facebook can consistently get you that result. Purchases may not.
- Optimise for landing page views: This conversion event drives higher quality traffic than link clicks do, since a landing page view only occurs when the page you want to send people to loads. (If someone clicks a link, but then closes out your page before it loads, that counts as a link click, but not a landing page view.) However, it’s still more common than most other conversion events. It’s a good choice if you do need to build up more website traffic and a customer base.
As Facebook gathers more data points to optimise more and more conversions to this “higher funnel” conversion point, you’re hoping that you will see a causal (not coincidental) increase in your “ultimate conversions” through the increase in volume of these “upper funnel conversions”.
Ultimately, Facebook’s delivery system can’t distinguish between causal and coincidental conversions, so if it’s mostly learning from people who are converting coincidentally, it’s going to get you more conversions that are probably coincidental. That isn’t a good use of your ads. Instead, if you choose a causal conversion event (one that you still value, and that must be completed on the path to ultimate conversion) then you’re feeding Facebook more qualified data, and allowing it the best opportunity to optimise to causal conversion events.
This isn’t new. It’s just marketing.
I know this whole post might seem confusing, but if you take it out of the Facebook context, it can be easier to understand. Because essentially, it’s all still about moving people further down the funnel.
All we’re doing here is saying “Ok, we don’t know enough about our customers to ask them to buy our product yet. But we probably wouldn’t do that in real life straight away either. Let’s start by introducing ourselves (optimising to a landing page view). Once we have enough customers who know us, let’s then show them our products (at that point, you’re optimising to a product page view). Once we’ve shown them our products, let’s get them considering buying them (optimising toward adds-to-carts). And once we’ve got them there, and we have enough people considering buying our products, it’s only then that we move on towards asking them to buy (optimising for purchases)”.
It’s important to understand that this may not be the case for everyone. For example, if you have a very low-ticket item that doesn’t require any consideration before purchasing, then you might be able to jump right to the purchase event if you can get 50 conversions a week. But for the majority of businesses, the consideration period might be longer and require more phases than you think. And if that’s the case, then it’s just… marketing.
While you’re here, why not subscribe to our newsletter?
Want more of these insights delivered to your inbox? Sign up to our newsletter using the form below. Don’t worry, we take privacy and spam very seriously, so we’ll only email you our new releases.