FB post iOS 14 Case Study: Exceeding $250K in Monthly Revenue

Facebook performance has become very unsteady after the iOS 14 update and the change in the Facebook view-through conversion window from 28 days to 7 days. Or is it not? Below are a few Facebook ads observations post iOS 14 updates and the change in the conversion window. And, how I used these observations to my advantage in crossing the store sales to above $250,000 per month for the FIRST TIME EVER.

Ad set’s unsteady performance

Before iOS 14, most of the ad sets had pretty stable performance over days. Some of the soft metrics that I used to look at were
  • ad frequency
  • first-time impressions
  • audience overlap
  • CPC
  • product view to add to cart ratio etc.
Post iOS 14, I realized that besides looking at the soft metrics, there are other beyond indicators that should also be looked at and are not available on the Facebook ad Interface. Some of those indicators were
  • Facebook preference of different ad sets over different days
  • Interest vs LookAlike audience
  • Facebook performance during different hours of the day etc.
That was so because I found out that the ad set which was performing before the iOS 14 update, the same ad set performance had become unstable post iOS 14 update.

Addressing unstable ad set performance

To counter the above problem, I would use 2-3 ad sets targeting the same product or category under separate campaigns with different audiences and then wait for the FB to find the winner for the day and then pausing the rest of the competing ad sets. This is not CBO as ad sets are under separate campaigns. On some days, FB would prefer none of the ad sets and on some days, all of the ad sets will perform better.

Day-parted performance

Post iOS 14, I observed that the Facebook performance varies throughout the day and follows mostly one common pattern. The pattern can be summarized as:
  • if an ad set had a good performance from midnight till 8 to 9 in the morning, it will continue to perform for the rest of the day in most cases.
  • The performance slows down a bit from morning till 3-4 in the afternoon and then the performance picks up again till 9.
  • Post 9, the performance hardly improves with some exceptions such as weekends (mostly Friday & Saturday).

Day Parting Strategy

To leverage the above learning, I usually reactivate all of my ad sets at midnight with spend>0 on the previous day and check for their performance at around 8-9 the next morning. I pause the non-performing ad sets and continue with the performing ones.

Cruising or Surfing Technique on Facebook

I, even, employ the cruising/ surfing Facebook Ads technique. Keep increasing the budget of the performing ad set by at most 10% every hour. I have observed that increasing the budget above 10%, say 15% or 20%, sometimes upsets the ad set performance so I find 10% as a safer option, and instead of half an hour, I increase the budget every hour as long as the ad set keeps performing. In some cases, some ad set starts with a daily budget of $300 at midnight and ends with the daily budget of $600. 9’o clock will be the last hour of increasing the budget with rare exceptions.

Facebook Ads Slow Surfing Technique

Previous day performance & budget reset at midnight

I observed that the everyday midnight the learning resets for most of the ad sets except the ad sets which performed really well the previous day. You will also notice that Facebook will keep pushing the dollars on the ads, in most of the cases, that performed the best the previous day.

My Strategy

I do not reset the budget of a performing ad set at midnight because, in the most likely event, that ad set is going to perform the next day as well.  However, I also consider midnight as an anchor point where the learning resets for most of the ad sets. Like I mentioned in the previous example where an ad set starts at $300 at midnight and ends at $600 by end of the day. For such an ad set, I do not reset the budget at midnight to $300 or to a lower or higher value.
I leave them as it is and most likely they perform over the next days as well. But the ad set which was paused earlier in the morning the previous day, I sometimes restart them with a reduced budget to test them again. Overall, I give a new ad set 2-3 trials before permanently pausing them. So, using midnight as a reset point worked well for me.

Day of the week performance

After the iOS 14 rollout, the performance is not the same all days. On some days of the week, the performance is mostly good such as Friday or long weekends and some days are mostly slow such as Mondays. So, on Friday I will let my ad sets cruise more and on Mondays, only a few winning ad sets will run with a lower daily budget.

Budget distribution over the day of the week

One method that has really worked for me is a varied budget distribution across days. For example, on a good day, I let the account spend up to $3000 to $4000 and on a bad day, the account spends as low as $1000. 
Variable performance on different days

Interest vs Lookalike audience

For me, interest with a larger audience size has been performing much better than the Lookalike audience post ios 14 updates.

Lowest cost vs Highest Value

Optimizing for highest value vs lowest cost bidding strategies has given a more stable and better performance.

Reduced FB ROAS after iOS 14 update

This has been the most discussed and questions topic on the internet about Facebook ads after the iOS 14 update. I did not observe any click-through ROAS drop after the iOS 14 update. Yes, there was definitely a drop in the view-through ROAS because of the reduction of the view-through conversion window from 28 days to just 7 days.
However, when I compared the FB click-through conversion on other platforms such as Hyros, GA or segment, etc. The performance was similar. In fact, on some occasions, I observed that the FB was under-reporting when compared to other platforms. So, I would encourage you to use different reporting platforms and attribution methods to properly evaluate the performance of FB.
Below I am attaching a screenshot of the ROAS reported on FB vs Hyros which include the months of pre and post ios 14 rollout and FB viewthrough conversion window change and I am sure you will find it very interesting
Hyros vs FB Data

Boosted Learning Trick

The trick of boosted learning for Facebook ads has really worked well for me post iOS 14 updates. The trick is to planning to send some email campaigns at the midnight. If the email campaign generates good sales, some of the purchases also get attributed to some of the ad sets and the ad set learning gets boosted for the rest of the day.

Facebook ads ios 14 what to do

I am going to outline a few pointers that have really helped me uping the performance of the FB ads

Configure Facebook API for your shop

There are a multiple reasons to configure Facebook API for your shop. At the highlevel, you can think something Facebook has two ways to know about the conversion events that occur through the ads manager. One is browser pixel and another one is server to server. Browser pixel fires when a user does some action in the browser such as page view, add to cart etc.

But Facebook is experiencing “Signal Loss” for example –

  • Browsers are becoming less reliable for consistent conversion tracking due to 7 day cookie expiration on webkit (i.e. Safari) browsers, browsers blocking all tracking by default, etc
  • iOS14 impact placing even more importance on the 8 conversion events being accurate
  • AdBlockers are growing in usage with some estimates stating that up to 27% of users have ad blocking enabled
    Conversion discrepancies between your Shopify orders and Facebook purchases. It’s not uncommon for most FB accounts to be short ~ 10-20% of actual orders on your Shopify store due to issues above
  • Integrate Facebook API helps with more accurate data attribution
  • There are multiple resources available out there that will help you in configuring Facebook API for your shop or still, you cannot do it, hire a professional

Start using a third-party data attribution tool

Even with the Facebook API configuration, you will only enhance your Facebook data but it would not help with attribution on other platforms. That’s why you need a third party tool which can attribute different data to different platforms. Some of such tools are Hyros or Anytrack. Even at the basic level, configure your GA to count the conversions as accurate as possible through UTM parameters. So, that when you login to GA, you can look at the performance of different tools through different attribution model

Follow FB instructions

Facebook has been rolling out a lot of instructions lately and you can find them under the resource center such as domain verification, configuring 8 aggregated events in the Facebook events manager post ios 14 update. Or, in one line, follow all the FB instructions that you see on the FB ads manager dashboard

Do not ignore view-through or first-click or LTV based attribution

By default, many of the data reporting or attribution tools do reporting based on the last click attribution model which in some cases undermines the capabilities of the Facebook as an ad platform as branded Google ads, Google shopping campaign, direct traffic, organic traffic, email campaigns gets attributed more when we only look at the last click model.

So, time and often also look at other attribution model such as first-click or view through and weigh the importance of Facebook as an ad platform as Facebook really helps with reaching out to new audience faster. Also, Google coming out with EVC (Enhanced View Through) Conversion attribution model, I think It is good idea to not to ignore view-through conversions as video is going to be the future of the advertisement and marketers will slowly shift away from static images.

Leave a Reply

Your email address will not be published. Required fields are marked *