The Problem

Beginning on April 26, 2021, Apple began enforcing App Tracking Transparency (ATT) on all devices using iOS 14.5 and above. What this meant in the most simple terms was that advertisers would need iOS users’ permission to track their activity and behavior across other apps and websites used on their devices. Previously, app developers could track this activity by default. 

The implications for advertisers were significant because:

1. Prospecting depends on having accurate information about user interests and behaviors. Fewer data points means less accurate interest/behavior targeting.
2. Remarketing often depends on having accurate information about actions taken by users on other apps and websites. More opt-outs means smaller remarketing audiences and less accurate lookalike audiences based on those remarketing audiences.
3. Reporting on conversions for opted out users becomes more difficult.

Most advertisers (and many people at Meta) anticipated 50% of users would opt out of tracking. Unfortunately for advertisers, the opt-out rate was closer to 95%. So if you were an advertiser who got a significant percentage of traffic from iOS, you were about to be in a world of trouble.

Most advertisers expected a 50% opt-out rate. In reality, it was closer to 95%.

Meta knew this and had more than $100 billion in advertising revenue at stake. So a solution was needed, and quickly.

An Incomplete Solution

Meta’s multifaceted solution to ATT (although likely the best they could do) was imperfect at best and created a lot of angst among advertisers.

In a nutshell, the solution was to try to provide as much information as possible to advertisers without sacrificing the privacy of users. They tried to achieve this by:

1. Using Aggregated Event Measurement to report on only the highest priority event when a user is opted out.This means that if an opted out user performed a ViewContent, AddToCart, and Purchase event within a 72 hour period, only the Purchase event would be reported (assuming that is the highest priority event).
2. Eliminating the ability to see conversion breakdowns (i.e. conversions by platform, device, age, and gender).
3. Delaying conversion reporting by up to 72 hours.
4. Providing modeled data to fill in gaps in conversion reporting.
5. Shortening the maximum available attribution window from 28-day-click/28-day-view to 7-day-click/1-day-view.

In particular, the shortened attribution window created havoc for advertisers, particularly the ones with high-price-point or long-consideration products. Most of the reduction in reported conversions seen by advertisers following iOS 14.5 was likely due to the shortened attribution window (now advertisers would only receive credit for clicks made 7 days prior to purchase instead of the previous default of 28 days).

Additionally, Aggregated Event Measurement has sometimes shown incomplete data. And modeled data had a problematic rollout, also with incomplete data (it’s much better now).

From a performance standpoint, advertisers saw their remarketing audiences shrink significantly and their lookalike audiences lose effectiveness. Unless you had an extremely helpful Meta rep or were an experienced advertiser with in-depth platform knowledge, it wasn’t clear what (outside of Meta’s official solutions above) could be done to mitigate the impact of ATT.

After April 26, 2021 (and particularly in June 2021), many advertisers looked at their results, saw a 50% decrease in conversions, and simply switched their campaigns off.

A Path Forward

It’s been 16 months since I got that call from Meta (and a year since Apple began enforcing ATT), and I’m now the Paid Social Director at Incubeta. As part of my job, I audit all prospective US paid social accounts. And it’s like going back in time. So many have obviously been switched off in frustration. Some in April-June 2021 and some a bit later. Most clients don’t fully understand the impact of the attribution window changing, and most have done little to mitigate the real performance impact of iOS 14.5 (as seen in their decimated remarketing audiences).

To complicate issues, turning off their Meta campaigns has often resulted in a downward spiral in other channels, particularly branded paid search. Many are desperate for a path forward, both from a paid social perspective and from a business perspective.

The good news is that the path forward is pretty straightforward. Here are the steps we take to help clients quickly mitigate the impact of iOS 14.5 and get back on track:

1. Education – The smaller attribution window is here to stay. If a client has a high price point or a product that requires long consideration, they are likely getting a large number of conversions outside of the 7-day-click/1-day-view window. This is something they need to come to terms with. It’s also possible to do an imperfect test on a platform like Pinterest that allows for 28-day attribution windows and directly compare the difference between 28-day and 7-day click windows. It’s not perfect, but it can provide some insight into when conversions occur.
2. Aggregated Event Measurement (AEM) – It is critically important to set up AEM on all of your domains (you must verify your domain first). This will allow you to tell Meta which events to prioritize. Failing to set this up will result in fewer reported conversions from opted out users.
3. Conversions API (CAPI) – Conversions API allows Meta to capture all website events at both the browser and the server levels. Events are then deduplicated to ensure there is no double-counting. This is extremely important as ad blockers and certain website features can interfere with the Meta Pixel. In an environment of scarce data, advertisers should leverage CAPI to ensure as many conversions as possible are being captured.
4. Broader Targeting – Due to the reduction in performance of lookalike audiences, it is often necessary to experiment with broader targeting for prospecting. This can mean broad interest-based audiences or fully broad targeting (targeting everyone in a country for example).
5. Wider Placements and Longer Lookback Windows – Due to shrinking audience sizes, it is often necessary to use wider placements, particularly in remarketing ad sets. For remarketing ad sets, we recommend using Automatic Placements. For prospecting ad sets, we recommend using Automatic Placements excluding Audience Network. It may also be necessary to expand your remarketing lookback windows to 90 or 180 days to increase the audience size so that it is large enough to be effective.
6. Simple Account Structure – Due to smaller audience sizes, it is necessary to build simplified account structures. Gone are the days of hundreds of active campaigns and ad sets in one account. We recommend as few campaigns as possible with no more than six ad sets per campaign and six ads per ad set. This will give you the best chance of leaving the learning phase. It’s important to keep an eye on frequency and first-time impression ratio as guideposts for ad sets that may need to be combined (ad sets with shrinking audiences will see frequency spike and first-time impression ratio plummet). 
7. Creative – Good creative has always been important, but it is particularly important post-iOS 14.5. Because audience sizes have generally shrunk, and different slices of the remaining audience members prefer to interact with different types of creative, it’s important to have a good mix of creative types in every ad set. Because we are retargeting social media engagers, the creative needs to be good (so that users engage with it and go into engagement retargeting pools). And because we are retargeting video viewers, we recommend overweighting video assets in the top of the funnel. Without great creative, the on-platform efforts recommended above will be less effective.
8. On-platform Signals / Owned Data – The audiences that have been most impacted by iOS 14.5 are the valuable site-based remarketing audiences that advertisers have long relied on to pump their ROAS. Audiences like site visitors and Add To Cart purchase. Those audiences are subject to large numbers of opt-outs and have become limited in size. The primary workaround is to retarget on-platform signals and owned data. We typically retarget the following on-platform and owned audiences (in addition to site-based audiences):

1. 1. Social media engagers
   2. Video viewers
   3. Lead form openers
   4. Instant experience openers
   5. Newsletter subscribers (owned list)
   6. Previous purchasers (owned list)

All of these audiences are not subject to iOS opt-outs and we now often see ad sets with these on-platform and owned audiences outperform those with site-based audiences.

Following these steps has allowed us to quickly increase reported ROAS on nearly all accounts that have come to us without these steps in place, sometimes by as much as 50%. Many have seen a significant increase in branded paid search volume as well. Businesses who don’t take these steps are leaving money on the table.

We Can Help

A year after the release of iOS 14.5, so many advertisers are still stuck believing paid social (particularly Meta Ads) no longer works. If you are one of them and you are interested in learning more about how we might help your business get back on track with a full-funnel paid social strategy, get in touch today.