Facebook Ad Hacks: A/A test








Testing should be a daily part for any marketer, as even small changes to a setup can harvest huge results. 

A/B testing is one of the most important things to improve performance of your ads and should always be practised in some kind of way, but there can be situations where it is known that a specific creative (combination of image/video, post copy, headline and CTA) is a good ad or a creative has been created by an agency or in-house designers without any variations. There are plenty of scenarios where this strategy can be applied. To make sure that this particular creative will go live and have the best possible chance to perform as intended, there are ways to artificially boost the initial performance.  

 I call it an A/A test hack, and it evolves 4+ identical ads in one ad set. As you don't want to miss out on the engagement from the losing ads, make sure that you are using the same post ID on all ads.

 By pushing 4 or more identical ads you let all ads have a chance of being shown, but not all ads will have the same level of positive engagement, and Facebook is known for favouring one ad over others in the same ad set. The trick is to only back the one ad with the highest bid. The bid you enter into Facebook and are willing to pay is only a part of the true bid.   



It is not uncommon that one ad in a good performing campaign has been allocated 80% of all the budget. 

True value of a bid: Bid x Estimated Action Rate + Relevancy Score


 Bid: There are different bidding strategies available on Facebook, but to keep it simple, this post will only focus on maximum bid. The max. bid is the most you are willing to pay for an action. in this case, let's say we can pay afford to pay £10 per purchase (purchase is an "action"). 

 Estimated Action Rate: This depends on the campaigns objective. Let's say we are optimising towards conversions. The action rate will be how likely the person seeing the ad is purchasing the product. This is different from user to user. For this example, we will say that the estimated purhcase rate for Person-1 is 0.2%.   

Relevancy Score: After 500 impressions an ad will be given a relevance score from 1 to 10, 10 being the best.

 Let's get the example:

 Ad 1: £10 x 0.2 + 4 = £6.00
Ad 2: £10 x 0.2 + 8 = £10.00
Ad 3: £10 x 0.2 + 7 = £9.00
Ad 4: £10 x 0.2 + 2 = £4.00

 In the above scenario, Ad 2, will win the most auctions, as it has the highest true bid (£10). Remember this is not what you are paying.

 Facebook use the Vickrey–Clarke–Groves auction model, which means that you will pay less than you bid (closer to the 2nd highest bid in the auction). 

 Once Facebook has deemed an ad to be a winner, pause the resisting ads. All that is left is an ad that will stretch your budget and get more delivery.

 Super simple, but so powerful. 

Comments

Post a Comment

Popular posts from this blog

Branded or non-branded Direct Response campaigns?

Image
Many people will probably look at the title and be perplexed, because why would you not brand your own direct response campaigns? Isn’t that the purpose; Not only to generate sales but build the brand to support retention? I would completely understand their point of view. On the other hand, there are huge advantages by running unbranded DR campaigns. First off, you are not limited to brand guidelines. Secondly, it’s easier to run Black Hat campaigns…. (Yes, I mentioned “BLACK HAT” and it sounds grim, but it doesn’t have to be).   If you have ever run a Black Hat initiative, you will know that they can be hugely successful (often only in the short run). They tend to have an expiry date just when the performance peaks.     Not Sticking to Brand guidelines Brand guidelines are in place for a reason, a very good reason to be honest. If Coca-Cola started using Pepsi colours in their marketing, this would confuse the consumer. If they started using a tone of voice ...
Read more

Are QR codes making a comeback?

Image
I remember playing around with QR codes back in 2016. The possibilities for this technology saw no limits. A few larger brands included it in their big-budget campaigns and then... nothing. It wasn't adopted by the people. Following, it was always kind of added as an after-thought on out-of-house advertising banners. Sometimes completely unimaginative with a link-out to the company's frontpage.  So what is the reason for this post? An ode to a fallen technology? ON THE CONTRARY Due to recent world events (read COVID-19) the technology has been dusted off.  The UK population is being re-educated on how to use QR codes by frequently being prompted by the NHS covid-app to use their phone to scan in public places. 👩‍🏫 According to a study by the European Journal of Social Psychology from 2010, it was found that it usually takes between 18 and 254 to form a new habit. It was also concluded that it takes about 66 days for a habit to become automatic 🔁 By 28 September 2020, 12.4 m...
Read more

Unlock the Power of Customer Journey Attribution: A Complete Guide to Google Analytics 4 Attribution Models

Image
Google Analytics 4 (GA4) includes several attribution models that can be used to attribute conversions and revenue to different touchpoints in the customer journey.  The following are the attribution models available in GA4: Last Interaction : This model attributes 100% of the conversion or revenue to the last touchpoint before the conversion. First Interaction : This model attributes 100% of the conversion or revenue to the first touchpoint in the customer journey. Linear : This model distributes the conversion or revenue evenly across all touchpoints in the customer journey. Time Decay : This model gives more credit to touchpoints that occurred closer in time to the conversion. The touchpoints closer to the conversion will have more weight and the touchpoints farther away will have less weight. Position-based : This model gives 40% of the credit to the first and last touchpoints in the customer journey, and distributes the remaining 20% evenly across the middle touchpoints. Data-...
Read more