Dennis Mink (SVP Marketing at Liftoff) discusses with Si Crowhurst (VP Creative Labs & Brand at Vungle), Gavin McNicholl (Global Head of Creative Labs at Vungle) and Thomas Zukowski (Creative Director at Liftoff) about how mobile creatives change post-IDFA, with less data granularity for tracking and targeting.
To make up for the limitation in targeting data, think about how the creative can “work harder” to ask people for their preferences and choice. Example: dating app serving an ad with an interactive unit where you can ask people questions.
A lot of what was recommended before in terms of creative production will still hold true: simplify your message and tie it directly to the product itself.
Pre-ATT, a heavy portion of the testing process was tied to incremental, single variable testing: taking the best performing creative and making minor adjustments to ultimately have multiple variations.
Without the possibility to attribute performance to a specific variation and with only campaign-level attribution, you can’t parse out the details/performance of iterations and you therefore need to introduce larger creative conceptual shifts. That’s a massive challenge.
You need to get a hold on any information you can: does a user interact, how many players play with the ad unit (for playables), do they complete it? You can then start to understand where the performance is coming from. This is Vungle’s Game Refinery plug.
Do not introduce a brand new creative strategy right now, as iOS 14.5 adoption takes weeks/months. And also, don’t introduce yet another variable during this transition period.
Do not ignore video as a format. A greater amount of spend distribution is now going towards video on SKAN. It makes sense because you are now reaching a bigger subset of users that you have not targeted based on user-level data, and video gives you the ability to give a more in-depth sales pitch.
Try combining several short videos (“stories”) together so you can test their order, etc. and figure out what works best. Vungle does this with their “Video stories” self-contained unit. Example.
There’s going to be a renewed attention to CTR and finding high-click creatives, which means you’ll want ad formats that have a high context.
Vungle’s “video banners” have been working very well because they allow to convey a bit more information.
To make impactful videos, get the message across very quickly and be very direct. Users seeing the ads are already on the device on which they’ll download your app, so show them what they’re missing. Example: showcase the goods, showcase the app interface, showcase the sample of products, etc.
For dating apps, highly-professional stock photography generally doesn’t work well and people respond better to things that are more realistic. Example.
At least in programmatic, there are usually not many differences in creative winners between iOS and Android (Liftoff have run over 60k A/B tests and about 90-95% of the times the winners are the same) and you can often combine your budgets to get learnings faster.
Don’t try to resist the changes that are coming. Get set up on SKAN and start getting learnings.
To make up for the limitation in targeting data, think about how the creative can “work harder” to ask people for their preferences and choice. Example: dating app serving an ad with an interactive unit where you can ask people questions.
A lot of what was recommended before in terms of creative production will still hold true: simplify your message and tie it directly to the product itself.
Pre-ATT, a heavy portion of the testing process was tied to incremental, single variable testing: taking the best performing creative and making minor adjustments to ultimately have multiple variations.
Without the possibility to attribute performance to a specific variation and with only campaign-level attribution, you can’t parse out the details/performance of iterations and you therefore need to introduce larger creative conceptual shifts. That’s a massive challenge.
You need to get a hold on any information you can: does a user interact, how many players play with the ad unit (for playables), do they complete it? You can then start to understand where the performance is coming from. This is Vungle’s Game Refinery plug.
Do not introduce a brand new creative strategy right now, as iOS 14.5 adoption takes weeks/months. And also, don’t introduce yet another variable during this transition period.
Do not ignore video as a format. A greater amount of spend distribution is now going towards video on SKAN. It makes sense because you are now reaching a bigger subset of users that you have not targeted based on user-level data, and video gives you the ability to give a more in-depth sales pitch.
Try combining several short videos (“stories”) together so you can test their order, etc. and figure out what works best. Vungle does this with their “Video stories” self-contained unit. Example.
There’s going to be a renewed attention to CTR and finding high-click creatives, which means you’ll want ad formats that have a high context.
Vungle’s “video banners” have been working very well because they allow to convey a bit more information.
To make impactful videos, get the message across very quickly and be very direct. Users seeing the ads are already on the device on which they’ll download your app, so show them what they’re missing. Example: showcase the goods, showcase the app interface, showcase the sample of products, etc.
For dating apps, highly-professional stock photography generally doesn’t work well and people respond better to things that are more realistic. Example.
At least in programmatic, there are usually not many differences in creative winners between iOS and Android (Liftoff have run over 60k A/B tests and about 90-95% of the times the winners are the same) and you can often combine your budgets to get learnings faster.
Don’t try to resist the changes that are coming. Get set up on SKAN and start getting learnings.
To make up for the limitation in targeting data, think about how the creative can “work harder” to ask people for their preferences and choice. Example: dating app serving an ad with an interactive unit where you can ask people questions.
A lot of what was recommended before in terms of creative production will still hold true: simplify your message and tie it directly to the product itself.
Pre-ATT, a heavy portion of the testing process was tied to incremental, single variable testing: taking the best performing creative and making minor adjustments to ultimately have multiple variations.
Without the possibility to attribute performance to a specific variation and with only campaign-level attribution, you can’t parse out the details/performance of iterations and you therefore need to introduce larger creative conceptual shifts. That’s a massive challenge.
You need to get a hold on any information you can: does a user interact, how many players play with the ad unit (for playables), do they complete it? You can then start to understand where the performance is coming from. This is Vungle’s Game Refinery plug.
Do not introduce a brand new creative strategy right now, as iOS 14.5 adoption takes weeks/months. And also, don’t introduce yet another variable during this transition period.
Do not ignore video as a format. A greater amount of spend distribution is now going towards video on SKAN. It makes sense because you are now reaching a bigger subset of users that you have not targeted based on user-level data, and video gives you the ability to give a more in-depth sales pitch.
Try combining several short videos (“stories”) together so you can test their order, etc. and figure out what works best. Vungle does this with their “Video stories” self-contained unit. Example.
There’s going to be a renewed attention to CTR and finding high-click creatives, which means you’ll want ad formats that have a high context.
Vungle’s “video banners” have been working very well because they allow to convey a bit more information.
To make impactful videos, get the message across very quickly and be very direct. Users seeing the ads are already on the device on which they’ll download your app, so show them what they’re missing. Example: showcase the goods, showcase the app interface, showcase the sample of products, etc.
For dating apps, highly-professional stock photography generally doesn’t work well and people respond better to things that are more realistic. Example.
At least in programmatic, there are usually not many differences in creative winners between iOS and Android (Liftoff have run over 60k A/B tests and about 90-95% of the times the winners are the same) and you can often combine your budgets to get learnings faster.
Don’t try to resist the changes that are coming. Get set up on SKAN and start getting learnings.
Notes for this resource are currently being transferred and will be available soon.
Si
There will be a limited amount of creatives, because of the campaign limitation. You will have to work harder on telling your story.
Gavin
Higher focus on creative strategy: how do you want to achieve your objectives? Scale, targeting new sources of users, etc.
Thomas
Biggest change is that advertisers were used to using targeting automation but with people opting-out, you won’t be able to leverage user-level data. You now have to find the right users without that head start.
We will still know a little bit about the source app.
Si
[💎@13:12] To make up for the limitation in targeting data, think about how the creative can “work harder” to ask people for their preferences and choice. Example: dating app serving an ad with an interactive unit where you can ask people questions.
What can we do to learn more about the context and the publisher that is serving the ad?
Gavin
We do have a few data points: the app you’re advertising for and the publisher where the ad is displayed. This is why Vungle partnered with Game Refinery: to get more info.
Thomas
A few things are not changing: we’ll still be serving ads in small formats on mobile devices, to people that have short attention spans which means you have a limited amount of time to convey the value of a product.
[💎@15:32] A lot of what was recommended before in terms of creative production will still hold true: simplify your message and tie it directly to the product itself.
Loss of post-install and even install data.
Thomas
It’s going to change everything about how we test ads.
[💎@17:40] Pre-ATT, a heavy portion of the testing process was tied to incremental, single variable testing: taking the best performing creative and making minor adjustments to ultimately have multiple variations.
[💎@18:22] Without the possibility to attribute performance to a specific variation and with only campaign-level attribution, you can’t parse out the details/performance of iterations and you therefore need to introduce larger creative conceptual shifts. That’s a massive challenge.
Gavin
Tests will have to be more impactful. You’ll need to test something very different.
[💎@20:00] You need to get a hold on any information you can: does a user interact, how many players play with the ad unit (for playables), do they complete it? You can then start to understand where the performance is coming from. This is Vungle’s Game Refinery plug.
Thomas
[💎@21:54] Do not introduce a brand new creative strategy right now, as iOS 14.5 adoption takes weeks/months. And also, don’t introduce yet another variable during this transition period.
Gavin
A slow transition makes the most sense.
Vungle has already been using variables in creatives. Example: text localized based on the device language. This allows some level of personalization.
Si
Take your top X performing creatives that you are running today and monitor them in a test environment against non-IDFA/tracking campaigns.
Thomas
[💎@25:14] Do not ignore video as a format. A greater amount of spend distribution is now going towards video on SKAN. It makes sense because you are now reaching a bigger subset of users that you have not targeted based on user-level data, and video gives you the ability to give a more in-depth sales pitch.
Gavin
[💎@27:28] Try combining several short videos (“stories”) together so you can test their order, etc. and figure out what works best. Vungle does this with their “Video stories” self-contained unit. Example below.
Si
Yes. Not sure how.
Denis
Like retargeting won’t be possible, incrementality tests might not be possible.
Gavin
For Vungle, the majority of focus is on user acquisition.
Thomas
Same.
Before, for ecommerce Liftoff would retarget users that have seen specific products. In the future they’ll have to work with publishers to identify the most popular products and “manually” focus on these.
Si
It can depend on the media buy/channels.
There probably won’t be a clear winner, it’s more going to be about the context in which the ad is shown.
Gavin
[💎@33:27] There’s going to be a renewed attention to CTR and finding high-click creatives, which means you’ll want ad formats that have a high context.
[💎@34:06] Vungle’s “video banners” have been working very well because they allow to convey a bit more information.
Thomas
Videos and playables are going to be more important because there is more time to show the value of an app.
On the other hand, as CPMs on video/playables go up, banners and interstitials might become more compelling.
Have all your bases covered and keep a diverse mix.
Gavin
Nothing shocking is coming up. There is a lot of historical data and right now things are looking pretty solid.
Thomas
It’s early but have been seeing great performance on SKAN-compatible campaigns because of the current price of CPMs. Main callout is how video has increased (10-15% increase).
Have seen higher CTR on video, with similar install rates on video vs. html formats.
“Video is just killing it at the moment”
Gavin
This is the reason the Game Refinery acquisition has been important: to get granular data on what about a video is performing, and also for which publishers/placements.
[💎@38:50] To make impactful videos, get the message across very quickly and be very direct. Users seeing the ads are already on the device on which they’ll download your app, so show them what they’re missing. Example: showcase the goods, showcase the app interface, showcase the sample of products, etc.
Thomas
“Showcase the goods”
It’s probably best to capture footage to showcase the actual content of the app, instead of creating a trailer.
Gavin
Identify why people get excited about your app and show that. Example: pikachu in AR for Niantic worked better than more sophisticated videos.
Problem-solving is a thing a lot of people play for, and you can highlight this in video in a number of ways. Example: video where you pause the image and use “coach drawings” (like in sports).
Thomas
The context of being in an in-app environment informs all decisions.
Even though you’re targeting broad, you don’t want to show too much.
Don’t rely too heavily on branding. Give a clear example of what the app UI is.
[💎@45:23] For dating apps, highly-professional stock photography generally doesn’t work well and people respond better to things that are more realistic. Example below.
Showcase the interface itself, lead with the product instead of making the brand the centerpiece.
Gavin
You need to be very direct. Branding is best kept for billboards where you’re trying to get people to associate your brand with something.
Thomas
[💎@48:29] At least in programmatic, there are usually not many differences in creative winners between iOS and Android (Liftoff have run over 60k A/B tests and about 90-95% of the times the winners are the same) and you can often combine your budgets to get learnings faster.
Si
Usually try to make creatives platform-agnostic.
The targeting on types of devices on Android has been quite useful.
Gavin
You usually see more difference based on the publisher/placement mix.
On Android, 360 video has been easier to unlock.
Thomas
Things are very different in programmatic vs. social media. You can usually get away with more of a narrative approach on social media because you’re more receptive to seeing people.
Gavin
On Vungle they have users that are playing games, so they know that playables are going to be successful.
Gavin
It’s still a very premium product.
Si
Technology/intelligence is very important for Vungle and they’re looking at the tools that can give more insights (e.g. Game Refinery).
Thomas
We can assume Android will follow suit but we can assume that you can test things on Android so far.
[💎@57:03] Don’t try to resist the changes that are coming. Get set up on SKAN and start getting learnings.
Si
There will be a limited amount of creatives, because of the campaign limitation. You will have to work harder on telling your story.
Gavin
Higher focus on creative strategy: how do you want to achieve your objectives? Scale, targeting new sources of users, etc.
Thomas
Biggest change is that advertisers were used to using targeting automation but with people opting-out, you won’t be able to leverage user-level data. You now have to find the right users without that head start.
We will still know a little bit about the source app.
Si
[💎@13:12] To make up for the limitation in targeting data, think about how the creative can “work harder” to ask people for their preferences and choice. Example: dating app serving an ad with an interactive unit where you can ask people questions.
What can we do to learn more about the context and the publisher that is serving the ad?
Gavin
We do have a few data points: the app you’re advertising for and the publisher where the ad is displayed. This is why Vungle partnered with Game Refinery: to get more info.
Thomas
A few things are not changing: we’ll still be serving ads in small formats on mobile devices, to people that have short attention spans which means you have a limited amount of time to convey the value of a product.
[💎@15:32] A lot of what was recommended before in terms of creative production will still hold true: simplify your message and tie it directly to the product itself.
Loss of post-install and even install data.
Thomas
It’s going to change everything about how we test ads.
[💎@17:40] Pre-ATT, a heavy portion of the testing process was tied to incremental, single variable testing: taking the best performing creative and making minor adjustments to ultimately have multiple variations.
[💎@18:22] Without the possibility to attribute performance to a specific variation and with only campaign-level attribution, you can’t parse out the details/performance of iterations and you therefore need to introduce larger creative conceptual shifts. That’s a massive challenge.
Gavin
Tests will have to be more impactful. You’ll need to test something very different.
[💎@20:00] You need to get a hold on any information you can: does a user interact, how many players play with the ad unit (for playables), do they complete it? You can then start to understand where the performance is coming from. This is Vungle’s Game Refinery plug.
Thomas
[💎@21:54] Do not introduce a brand new creative strategy right now, as iOS 14.5 adoption takes weeks/months. And also, don’t introduce yet another variable during this transition period.
Gavin
A slow transition makes the most sense.
Vungle has already been using variables in creatives. Example: text localized based on the device language. This allows some level of personalization.
Si
Take your top X performing creatives that you are running today and monitor them in a test environment against non-IDFA/tracking campaigns.
Thomas
[💎@25:14] Do not ignore video as a format. A greater amount of spend distribution is now going towards video on SKAN. It makes sense because you are now reaching a bigger subset of users that you have not targeted based on user-level data, and video gives you the ability to give a more in-depth sales pitch.
Gavin
[💎@27:28] Try combining several short videos (“stories”) together so you can test their order, etc. and figure out what works best. Vungle does this with their “Video stories” self-contained unit. Example below.
Si
Yes. Not sure how.
Denis
Like retargeting won’t be possible, incrementality tests might not be possible.
Gavin
For Vungle, the majority of focus is on user acquisition.
Thomas
Same.
Before, for ecommerce Liftoff would retarget users that have seen specific products. In the future they’ll have to work with publishers to identify the most popular products and “manually” focus on these.
Si
It can depend on the media buy/channels.
There probably won’t be a clear winner, it’s more going to be about the context in which the ad is shown.
Gavin
[💎@33:27] There’s going to be a renewed attention to CTR and finding high-click creatives, which means you’ll want ad formats that have a high context.
[💎@34:06] Vungle’s “video banners” have been working very well because they allow to convey a bit more information.
Thomas
Videos and playables are going to be more important because there is more time to show the value of an app.
On the other hand, as CPMs on video/playables go up, banners and interstitials might become more compelling.
Have all your bases covered and keep a diverse mix.
Gavin
Nothing shocking is coming up. There is a lot of historical data and right now things are looking pretty solid.
Thomas
It’s early but have been seeing great performance on SKAN-compatible campaigns because of the current price of CPMs. Main callout is how video has increased (10-15% increase).
Have seen higher CTR on video, with similar install rates on video vs. html formats.
“Video is just killing it at the moment”
Gavin
This is the reason the Game Refinery acquisition has been important: to get granular data on what about a video is performing, and also for which publishers/placements.
[💎@38:50] To make impactful videos, get the message across very quickly and be very direct. Users seeing the ads are already on the device on which they’ll download your app, so show them what they’re missing. Example: showcase the goods, showcase the app interface, showcase the sample of products, etc.
Thomas
“Showcase the goods”
It’s probably best to capture footage to showcase the actual content of the app, instead of creating a trailer.
Gavin
Identify why people get excited about your app and show that. Example: pikachu in AR for Niantic worked better than more sophisticated videos.
Problem-solving is a thing a lot of people play for, and you can highlight this in video in a number of ways. Example: video where you pause the image and use “coach drawings” (like in sports).
Thomas
The context of being in an in-app environment informs all decisions.
Even though you’re targeting broad, you don’t want to show too much.
Don’t rely too heavily on branding. Give a clear example of what the app UI is.
[💎@45:23] For dating apps, highly-professional stock photography generally doesn’t work well and people respond better to things that are more realistic. Example below.
Showcase the interface itself, lead with the product instead of making the brand the centerpiece.
Gavin
You need to be very direct. Branding is best kept for billboards where you’re trying to get people to associate your brand with something.
Thomas
[💎@48:29] At least in programmatic, there are usually not many differences in creative winners between iOS and Android (Liftoff have run over 60k A/B tests and about 90-95% of the times the winners are the same) and you can often combine your budgets to get learnings faster.
Si
Usually try to make creatives platform-agnostic.
The targeting on types of devices on Android has been quite useful.
Gavin
You usually see more difference based on the publisher/placement mix.
On Android, 360 video has been easier to unlock.
Thomas
Things are very different in programmatic vs. social media. You can usually get away with more of a narrative approach on social media because you’re more receptive to seeing people.
Gavin
On Vungle they have users that are playing games, so they know that playables are going to be successful.
Gavin
It’s still a very premium product.
Si
Technology/intelligence is very important for Vungle and they’re looking at the tools that can give more insights (e.g. Game Refinery).
Thomas
We can assume Android will follow suit but we can assume that you can test things on Android so far.
[💎@57:03] Don’t try to resist the changes that are coming. Get set up on SKAN and start getting learnings.
Si
There will be a limited amount of creatives, because of the campaign limitation. You will have to work harder on telling your story.
Gavin
Higher focus on creative strategy: how do you want to achieve your objectives? Scale, targeting new sources of users, etc.
Thomas
Biggest change is that advertisers were used to using targeting automation but with people opting-out, you won’t be able to leverage user-level data. You now have to find the right users without that head start.
We will still know a little bit about the source app.
Si
[💎@13:12] To make up for the limitation in targeting data, think about how the creative can “work harder” to ask people for their preferences and choice. Example: dating app serving an ad with an interactive unit where you can ask people questions.
What can we do to learn more about the context and the publisher that is serving the ad?
Gavin
We do have a few data points: the app you’re advertising for and the publisher where the ad is displayed. This is why Vungle partnered with Game Refinery: to get more info.
Thomas
A few things are not changing: we’ll still be serving ads in small formats on mobile devices, to people that have short attention spans which means you have a limited amount of time to convey the value of a product.
[💎@15:32] A lot of what was recommended before in terms of creative production will still hold true: simplify your message and tie it directly to the product itself.
Loss of post-install and even install data.
Thomas
It’s going to change everything about how we test ads.
[💎@17:40] Pre-ATT, a heavy portion of the testing process was tied to incremental, single variable testing: taking the best performing creative and making minor adjustments to ultimately have multiple variations.
[💎@18:22] Without the possibility to attribute performance to a specific variation and with only campaign-level attribution, you can’t parse out the details/performance of iterations and you therefore need to introduce larger creative conceptual shifts. That’s a massive challenge.
Gavin
Tests will have to be more impactful. You’ll need to test something very different.
[💎@20:00] You need to get a hold on any information you can: does a user interact, how many players play with the ad unit (for playables), do they complete it? You can then start to understand where the performance is coming from. This is Vungle’s Game Refinery plug.
Thomas
[💎@21:54] Do not introduce a brand new creative strategy right now, as iOS 14.5 adoption takes weeks/months. And also, don’t introduce yet another variable during this transition period.
Gavin
A slow transition makes the most sense.
Vungle has already been using variables in creatives. Example: text localized based on the device language. This allows some level of personalization.
Si
Take your top X performing creatives that you are running today and monitor them in a test environment against non-IDFA/tracking campaigns.
Thomas
[💎@25:14] Do not ignore video as a format. A greater amount of spend distribution is now going towards video on SKAN. It makes sense because you are now reaching a bigger subset of users that you have not targeted based on user-level data, and video gives you the ability to give a more in-depth sales pitch.
Gavin
[💎@27:28] Try combining several short videos (“stories”) together so you can test their order, etc. and figure out what works best. Vungle does this with their “Video stories” self-contained unit. Example below.
Si
Yes. Not sure how.
Denis
Like retargeting won’t be possible, incrementality tests might not be possible.
Gavin
For Vungle, the majority of focus is on user acquisition.
Thomas
Same.
Before, for ecommerce Liftoff would retarget users that have seen specific products. In the future they’ll have to work with publishers to identify the most popular products and “manually” focus on these.
Si
It can depend on the media buy/channels.
There probably won’t be a clear winner, it’s more going to be about the context in which the ad is shown.
Gavin
[💎@33:27] There’s going to be a renewed attention to CTR and finding high-click creatives, which means you’ll want ad formats that have a high context.
[💎@34:06] Vungle’s “video banners” have been working very well because they allow to convey a bit more information.
Thomas
Videos and playables are going to be more important because there is more time to show the value of an app.
On the other hand, as CPMs on video/playables go up, banners and interstitials might become more compelling.
Have all your bases covered and keep a diverse mix.
Gavin
Nothing shocking is coming up. There is a lot of historical data and right now things are looking pretty solid.
Thomas
It’s early but have been seeing great performance on SKAN-compatible campaigns because of the current price of CPMs. Main callout is how video has increased (10-15% increase).
Have seen higher CTR on video, with similar install rates on video vs. html formats.
“Video is just killing it at the moment”
Gavin
This is the reason the Game Refinery acquisition has been important: to get granular data on what about a video is performing, and also for which publishers/placements.
[💎@38:50] To make impactful videos, get the message across very quickly and be very direct. Users seeing the ads are already on the device on which they’ll download your app, so show them what they’re missing. Example: showcase the goods, showcase the app interface, showcase the sample of products, etc.
Thomas
“Showcase the goods”
It’s probably best to capture footage to showcase the actual content of the app, instead of creating a trailer.
Gavin
Identify why people get excited about your app and show that. Example: pikachu in AR for Niantic worked better than more sophisticated videos.
Problem-solving is a thing a lot of people play for, and you can highlight this in video in a number of ways. Example: video where you pause the image and use “coach drawings” (like in sports).
Thomas
The context of being in an in-app environment informs all decisions.
Even though you’re targeting broad, you don’t want to show too much.
Don’t rely too heavily on branding. Give a clear example of what the app UI is.
[💎@45:23] For dating apps, highly-professional stock photography generally doesn’t work well and people respond better to things that are more realistic. Example below.
Showcase the interface itself, lead with the product instead of making the brand the centerpiece.
Gavin
You need to be very direct. Branding is best kept for billboards where you’re trying to get people to associate your brand with something.
Thomas
[💎@48:29] At least in programmatic, there are usually not many differences in creative winners between iOS and Android (Liftoff have run over 60k A/B tests and about 90-95% of the times the winners are the same) and you can often combine your budgets to get learnings faster.
Si
Usually try to make creatives platform-agnostic.
The targeting on types of devices on Android has been quite useful.
Gavin
You usually see more difference based on the publisher/placement mix.
On Android, 360 video has been easier to unlock.
Thomas
Things are very different in programmatic vs. social media. You can usually get away with more of a narrative approach on social media because you’re more receptive to seeing people.
Gavin
On Vungle they have users that are playing games, so they know that playables are going to be successful.
Gavin
It’s still a very premium product.
Si
Technology/intelligence is very important for Vungle and they’re looking at the tools that can give more insights (e.g. Game Refinery).
Thomas
We can assume Android will follow suit but we can assume that you can test things on Android so far.
[💎@57:03] Don’t try to resist the changes that are coming. Get set up on SKAN and start getting learnings.
