Fabien-Pierre Nicolas (VP of US Marketing at SmartNews) talks with Esther Shatz about one of the most questionable growth channels out there: TV advertising.
Questions to ask yourself when considering TV:
1. Are we struggling to find a new mobile channel that’s ROAS positive?
2. Are we spending a good amount (i.e. 1 million-plus in the US)?
3. Do we understand the target audience enough and can we express the value proposition of the product in 15 seconds?
4. Do we have a significant chunk of our audience over 40?
5. Do we have a celebrity endorsement or well-known IP?
Every channel that you’re running for performance, you can test and “dip your toes”. TV is no exception but in the US you need to have $200k+: for TV creatives ($40k with a cost-efficient agency) and testing permutations of creatives/channels/dayparts.
Depending on your product, your creatives/channels/dayparts are different. For Calm, evenings were working very well because they’re about relaxing and sleeping + there are low CPMs. For SmartNews it was about the high attention on news channels and they were more interested in high response rates than high CPMs.
SmartNews has 2 attribution logics in parallel:
1. Baseline 2 weeks before and 2 weeks after
2. Using TVSquared (integrated with Adjust) to which you pass all your organic installs to then give a fractional value to say whether a user comes from TV or not (based on ingesting timestamps +- 5 min)
You can measure the age and gender coming from TV ads through Apple Search, even though it’s just a sample.
Every 3 months, Smartnews was re-establishing a new baseline by pausing TV campaigns (2-3 weeks “blackout” period).
A lot of agencies are not awesome at doing 15-second ads, because they typically hire talent that’s good at writing 30-second and 60-second ads. So the SmartNews team had to become script writers.
It’s performance marketing so you need to formulate several different hypotheses and measure the response rate (ratio of impressions to installs). Example: why news matters, something feature-driven, 2 about politics leveraging a polarized environment.
If you buy inventory the week before (instead of quarterly), which is heavily-discounted, you will have weeks where you can easily spend $200-300k but others where you won’t be able to spend all your budget.
Select agencies that understand your KPIs. A lot of agencies falsely claim they understand what mobile apps do. Have the agencies submit a proposal and provide you high-level media planning, and use a matrix to evaluate them.
During the test, ask your agency to have a weekly sync call to have a feedback loop with your management.
If you’re working with a global company that already has experience, make sure you tap into that.
In the US there’s a split in age: 30-40+ audience is still very much on linear TV, but the younger audience is only on it for specific events exclusive to specific TV channels (e.g. superbowl, political debates, etc.).
Don’t hesitate to bring other people like UI/UX designers, product team, etc. to get feedback on the TV creatives.
Streaming is limited (several services with no advertising) and fragmented. It is inefficient and has extremely high CPMs: compared to linear TV, you will pay 5x to 10x more for impressions.
For 18-40 audience, streaming also competes with influencer marketing in terms of video views and capability to measure the uplift. In many verticals influencer marketing might become more important.
Questions to ask yourself when considering TV:
1. Are we struggling to find a new mobile channel that’s ROAS positive?
2. Are we spending a good amount (i.e. 1 million-plus in the US)?
3. Do we understand the target audience enough and can we express the value proposition of the product in 15 seconds?
4. Do we have a significant chunk of our audience over 40?
5. Do we have a celebrity endorsement or well-known IP?
Every channel that you’re running for performance, you can test and “dip your toes”. TV is no exception but in the US you need to have $200k+: for TV creatives ($40k with a cost-efficient agency) and testing permutations of creatives/channels/dayparts.
Depending on your product, your creatives/channels/dayparts are different. For Calm, evenings were working very well because they’re about relaxing and sleeping + there are low CPMs. For SmartNews it was about the high attention on news channels and they were more interested in high response rates than high CPMs.
SmartNews has 2 attribution logics in parallel:
1. Baseline 2 weeks before and 2 weeks after
2. Using TVSquared (integrated with Adjust) to which you pass all your organic installs to then give a fractional value to say whether a user comes from TV or not (based on ingesting timestamps +- 5 min)
You can measure the age and gender coming from TV ads through Apple Search, even though it’s just a sample.
Every 3 months, Smartnews was re-establishing a new baseline by pausing TV campaigns (2-3 weeks “blackout” period).
A lot of agencies are not awesome at doing 15-second ads, because they typically hire talent that’s good at writing 30-second and 60-second ads. So the SmartNews team had to become script writers.
It’s performance marketing so you need to formulate several different hypotheses and measure the response rate (ratio of impressions to installs). Example: why news matters, something feature-driven, 2 about politics leveraging a polarized environment.
If you buy inventory the week before (instead of quarterly), which is heavily-discounted, you will have weeks where you can easily spend $200-300k but others where you won’t be able to spend all your budget.
Select agencies that understand your KPIs. A lot of agencies falsely claim they understand what mobile apps do. Have the agencies submit a proposal and provide you high-level media planning, and use a matrix to evaluate them.
During the test, ask your agency to have a weekly sync call to have a feedback loop with your management.
If you’re working with a global company that already has experience, make sure you tap into that.
In the US there’s a split in age: 30-40+ audience is still very much on linear TV, but the younger audience is only on it for specific events exclusive to specific TV channels (e.g. superbowl, political debates, etc.).
Don’t hesitate to bring other people like UI/UX designers, product team, etc. to get feedback on the TV creatives.
Streaming is limited (several services with no advertising) and fragmented. It is inefficient and has extremely high CPMs: compared to linear TV, you will pay 5x to 10x more for impressions.
For 18-40 audience, streaming also competes with influencer marketing in terms of video views and capability to measure the uplift. In many verticals influencer marketing might become more important.
Questions to ask yourself when considering TV:
1. Are we struggling to find a new mobile channel that’s ROAS positive?
2. Are we spending a good amount (i.e. 1 million-plus in the US)?
3. Do we understand the target audience enough and can we express the value proposition of the product in 15 seconds?
4. Do we have a significant chunk of our audience over 40?
5. Do we have a celebrity endorsement or well-known IP?
Every channel that you’re running for performance, you can test and “dip your toes”. TV is no exception but in the US you need to have $200k+: for TV creatives ($40k with a cost-efficient agency) and testing permutations of creatives/channels/dayparts.
Depending on your product, your creatives/channels/dayparts are different. For Calm, evenings were working very well because they’re about relaxing and sleeping + there are low CPMs. For SmartNews it was about the high attention on news channels and they were more interested in high response rates than high CPMs.
SmartNews has 2 attribution logics in parallel:
1. Baseline 2 weeks before and 2 weeks after
2. Using TVSquared (integrated with Adjust) to which you pass all your organic installs to then give a fractional value to say whether a user comes from TV or not (based on ingesting timestamps +- 5 min)
You can measure the age and gender coming from TV ads through Apple Search, even though it’s just a sample.
Every 3 months, Smartnews was re-establishing a new baseline by pausing TV campaigns (2-3 weeks “blackout” period).
A lot of agencies are not awesome at doing 15-second ads, because they typically hire talent that’s good at writing 30-second and 60-second ads. So the SmartNews team had to become script writers.
It’s performance marketing so you need to formulate several different hypotheses and measure the response rate (ratio of impressions to installs). Example: why news matters, something feature-driven, 2 about politics leveraging a polarized environment.
If you buy inventory the week before (instead of quarterly), which is heavily-discounted, you will have weeks where you can easily spend $200-300k but others where you won’t be able to spend all your budget.
Select agencies that understand your KPIs. A lot of agencies falsely claim they understand what mobile apps do. Have the agencies submit a proposal and provide you high-level media planning, and use a matrix to evaluate them.
During the test, ask your agency to have a weekly sync call to have a feedback loop with your management.
If you’re working with a global company that already has experience, make sure you tap into that.
In the US there’s a split in age: 30-40+ audience is still very much on linear TV, but the younger audience is only on it for specific events exclusive to specific TV channels (e.g. superbowl, political debates, etc.).
Don’t hesitate to bring other people like UI/UX designers, product team, etc. to get feedback on the TV creatives.
Streaming is limited (several services with no advertising) and fragmented. It is inefficient and has extremely high CPMs: compared to linear TV, you will pay 5x to 10x more for impressions.
For 18-40 audience, streaming also competes with influencer marketing in terms of video views and capability to measure the uplift. In many verticals influencer marketing might become more important.
Notes for this resource are currently being transferred and will be available soon.
Transitioning from gaming to apps meant changing business model and consumers. In gaming he worked with high LTV but at SmartNews the LTV is much lower.
You change the way you approach your marketing, your stack, your team
2 types of bucket for KPIs:
Past 7 days, the retention curve is fairly flat. This is what they optimize against.
SmartNews is focused on Japan and the US. Over 50M installs, over 20M monthly active users.
Been running in both the US and Japan as a channel to keep growing.
[💎@06:07] Questions to ask yourself when considering TV:
They didn’t have #5 but fell into all the other buckets.
[💎@07:59] Every channel that you’re running for performance, you can test and “dip your toes”. TV is no exception but in the US you need to have $200k+: for TV creatives ($40k with a cost-efficient agency) and testing permutations of creatives/channels/dayparts.
This budget should give you the ability to measure the incrementality over your baseline of organic users for let’s say 3 weeks.
[💎@09:32] Depending on your product, your creatives/channels/dayparts are different. For Calm, evenings were working very well because they’re about relaxing and sleeping + there are low CPMs. For SmartNews it was about the high attention on news channels and they were more interested in high response rates than high CPMs.
SmartNews started with a pretty clean slate when it comes to offline, in both the US and Japan.
[💎@11:15] SmartNews has 2 attribution logics in parallel:
1. Baseline 2 weeks before and 2 weeks after
2. Using TVSquared (integrated with Adjust) to which you pass all your organic installs to then give a fractional value to say whether a user comes from TV or not (based on ingesting timestamps +- 5 min)
[💎@12:34] You can measure the age and gender coming from TV ads through Apple Search, even though it’s just a sample.
They also noticed there was a re-engagement effect: TV helped get a lot of users back into the product.
[💎@13:15] Every 3 months, Smartnews was re-establishing a new baseline by pausing TV campaigns (2-3 weeks “blackout” period).
From an awareness perspective, there has been a lasting impact: they increased awareness by +50%.
Unlike for games, that brand awareness for a news app can pay long term when people are considering a news app way after having seen the ad.
No How Did You Hear About Us survey: they did not change the onboarding.
Messaging “news from all sides” was already included in the product, app store page and creatives.
There was an interesting feedback loop: they could tell which reviews (good or bad) came from users that saw a specific TV ad.
Joined with creatives for 3-months worth of TV tests.
[💎@17:25] A lot of agencies are not awesome at doing 15-second ads, because they typically hire talent that’s good at writing 30-second and 60-second ads. So the SmartNews team had to become script writers.
[💎@18:13] It’s performance marketing so you need to formulate several different hypotheses and measure the response rate (ratio of impressions to installs). Example: why news matters, something feature-driven, 2 about politics leveraging a polarized environment.
Had a hard time picking a winner and ran the creatives for 3-4 months but they were also not running huge budgets.
[💎@20:30] If you buy inventory the week before (instead of quarterly), which is heavily-discounted, you will have weeks where you can easily spend $200-300k but others where you won’t be able to spend all your budget.
Do not try TV ads if you don’t have 3 or more creatives. Make sure you have support from your C-level because it is a high touch channel.
[💎@23:00] Select agencies that understand your KPIs. A lot of agencies falsely claim they understand what mobile apps do. Have the agencies submit a request for proposal and provide you high-level media planning, and use a matrix to evaluate them.
Get your analytics ready (TVSquared).
[💎@24:44] During the test, ask your agency to have a weekly sync call to have a feedback loop with your management.
When you get started make sure you discuss the follow up budget more quickly and plan for success: think ahead.
Fabien also thinks they could have worked more closely with Japan beforehand (measuring baseline, going dark, etc.).
[💎@25:37] If you’re working with a global company that already has experience, make sure you tap into that.
[💎@26:15] Don’t hesitate to bring other people like UI/UX designers, product team, etc. to get feedback on the TV creatives.
[💎@28:09] In the US there’s a split in age: 30-40+ audience is still very much on linear TV, but the younger audience is only on it for specific events exclusive to specific TV channels (e.g. superbowl, political debates, etc.).
Netflix is the biggest streaming service and does not have advertising. Same for Amazon Prime Video and a big portion of Hulu subscribers. That’s a lot of eyeballs not available for purchase.
[💎@29:35] Streaming is limited (several services with no advertising) and fragmented. It is inefficient and has extremely high CPMs: compared to linear TV, you will pay 5x to 10x more for impressions.
[💎@30:35] For 18-40 audience, streaming also competes with influencer marketing in terms of video views and capability to measure the uplift. In many verticals influencer marketing might become more important.
Influencer marketing gives you less control. But there is also a diversification in terms of “products”: 5-10 second intro or midroll, influencer content not really mentioned, and we should start seeing more “produced” creatives etc.
A lot of influencer marketing platforms have not really done the analytics work to be able to measure incrementality.
#1 tip: always be testing. 80% on tested channels but 20% on testing new channels (ask platforms for credits!).
Transitioning from gaming to apps meant changing business model and consumers. In gaming he worked with high LTV but at SmartNews the LTV is much lower.
You change the way you approach your marketing, your stack, your team
2 types of bucket for KPIs:
Past 7 days, the retention curve is fairly flat. This is what they optimize against.
SmartNews is focused on Japan and the US. Over 50M installs, over 20M monthly active users.
Been running in both the US and Japan as a channel to keep growing.
[💎@06:07] Questions to ask yourself when considering TV:
They didn’t have #5 but fell into all the other buckets.
[💎@07:59] Every channel that you’re running for performance, you can test and “dip your toes”. TV is no exception but in the US you need to have $200k+: for TV creatives ($40k with a cost-efficient agency) and testing permutations of creatives/channels/dayparts.
This budget should give you the ability to measure the incrementality over your baseline of organic users for let’s say 3 weeks.
[💎@09:32] Depending on your product, your creatives/channels/dayparts are different. For Calm, evenings were working very well because they’re about relaxing and sleeping + there are low CPMs. For SmartNews it was about the high attention on news channels and they were more interested in high response rates than high CPMs.
SmartNews started with a pretty clean slate when it comes to offline, in both the US and Japan.
[💎@11:15] SmartNews has 2 attribution logics in parallel:
1. Baseline 2 weeks before and 2 weeks after
2. Using TVSquared (integrated with Adjust) to which you pass all your organic installs to then give a fractional value to say whether a user comes from TV or not (based on ingesting timestamps +- 5 min)
[💎@12:34] You can measure the age and gender coming from TV ads through Apple Search, even though it’s just a sample.
They also noticed there was a re-engagement effect: TV helped get a lot of users back into the product.
[💎@13:15] Every 3 months, Smartnews was re-establishing a new baseline by pausing TV campaigns (2-3 weeks “blackout” period).
From an awareness perspective, there has been a lasting impact: they increased awareness by +50%.
Unlike for games, that brand awareness for a news app can pay long term when people are considering a news app way after having seen the ad.
No How Did You Hear About Us survey: they did not change the onboarding.
Messaging “news from all sides” was already included in the product, app store page and creatives.
There was an interesting feedback loop: they could tell which reviews (good or bad) came from users that saw a specific TV ad.
Joined with creatives for 3-months worth of TV tests.
[💎@17:25] A lot of agencies are not awesome at doing 15-second ads, because they typically hire talent that’s good at writing 30-second and 60-second ads. So the SmartNews team had to become script writers.
[💎@18:13] It’s performance marketing so you need to formulate several different hypotheses and measure the response rate (ratio of impressions to installs). Example: why news matters, something feature-driven, 2 about politics leveraging a polarized environment.
Had a hard time picking a winner and ran the creatives for 3-4 months but they were also not running huge budgets.
[💎@20:30] If you buy inventory the week before (instead of quarterly), which is heavily-discounted, you will have weeks where you can easily spend $200-300k but others where you won’t be able to spend all your budget.
Do not try TV ads if you don’t have 3 or more creatives. Make sure you have support from your C-level because it is a high touch channel.
[💎@23:00] Select agencies that understand your KPIs. A lot of agencies falsely claim they understand what mobile apps do. Have the agencies submit a request for proposal and provide you high-level media planning, and use a matrix to evaluate them.
Get your analytics ready (TVSquared).
[💎@24:44] During the test, ask your agency to have a weekly sync call to have a feedback loop with your management.
When you get started make sure you discuss the follow up budget more quickly and plan for success: think ahead.
Fabien also thinks they could have worked more closely with Japan beforehand (measuring baseline, going dark, etc.).
[💎@25:37] If you’re working with a global company that already has experience, make sure you tap into that.
[💎@26:15] Don’t hesitate to bring other people like UI/UX designers, product team, etc. to get feedback on the TV creatives.
[💎@28:09] In the US there’s a split in age: 30-40+ audience is still very much on linear TV, but the younger audience is only on it for specific events exclusive to specific TV channels (e.g. superbowl, political debates, etc.).
Netflix is the biggest streaming service and does not have advertising. Same for Amazon Prime Video and a big portion of Hulu subscribers. That’s a lot of eyeballs not available for purchase.
[💎@29:35] Streaming is limited (several services with no advertising) and fragmented. It is inefficient and has extremely high CPMs: compared to linear TV, you will pay 5x to 10x more for impressions.
[💎@30:35] For 18-40 audience, streaming also competes with influencer marketing in terms of video views and capability to measure the uplift. In many verticals influencer marketing might become more important.
Influencer marketing gives you less control. But there is also a diversification in terms of “products”: 5-10 second intro or midroll, influencer content not really mentioned, and we should start seeing more “produced” creatives etc.
A lot of influencer marketing platforms have not really done the analytics work to be able to measure incrementality.
#1 tip: always be testing. 80% on tested channels but 20% on testing new channels (ask platforms for credits!).
Transitioning from gaming to apps meant changing business model and consumers. In gaming he worked with high LTV but at SmartNews the LTV is much lower.
You change the way you approach your marketing, your stack, your team
2 types of bucket for KPIs:
Past 7 days, the retention curve is fairly flat. This is what they optimize against.
SmartNews is focused on Japan and the US. Over 50M installs, over 20M monthly active users.
Been running in both the US and Japan as a channel to keep growing.
[💎@06:07] Questions to ask yourself when considering TV:
They didn’t have #5 but fell into all the other buckets.
[💎@07:59] Every channel that you’re running for performance, you can test and “dip your toes”. TV is no exception but in the US you need to have $200k+: for TV creatives ($40k with a cost-efficient agency) and testing permutations of creatives/channels/dayparts.
This budget should give you the ability to measure the incrementality over your baseline of organic users for let’s say 3 weeks.
[💎@09:32] Depending on your product, your creatives/channels/dayparts are different. For Calm, evenings were working very well because they’re about relaxing and sleeping + there are low CPMs. For SmartNews it was about the high attention on news channels and they were more interested in high response rates than high CPMs.
SmartNews started with a pretty clean slate when it comes to offline, in both the US and Japan.
[💎@11:15] SmartNews has 2 attribution logics in parallel:
1. Baseline 2 weeks before and 2 weeks after
2. Using TVSquared (integrated with Adjust) to which you pass all your organic installs to then give a fractional value to say whether a user comes from TV or not (based on ingesting timestamps +- 5 min)
[💎@12:34] You can measure the age and gender coming from TV ads through Apple Search, even though it’s just a sample.
They also noticed there was a re-engagement effect: TV helped get a lot of users back into the product.
[💎@13:15] Every 3 months, Smartnews was re-establishing a new baseline by pausing TV campaigns (2-3 weeks “blackout” period).
From an awareness perspective, there has been a lasting impact: they increased awareness by +50%.
Unlike for games, that brand awareness for a news app can pay long term when people are considering a news app way after having seen the ad.
No How Did You Hear About Us survey: they did not change the onboarding.
Messaging “news from all sides” was already included in the product, app store page and creatives.
There was an interesting feedback loop: they could tell which reviews (good or bad) came from users that saw a specific TV ad.
Joined with creatives for 3-months worth of TV tests.
[💎@17:25] A lot of agencies are not awesome at doing 15-second ads, because they typically hire talent that’s good at writing 30-second and 60-second ads. So the SmartNews team had to become script writers.
[💎@18:13] It’s performance marketing so you need to formulate several different hypotheses and measure the response rate (ratio of impressions to installs). Example: why news matters, something feature-driven, 2 about politics leveraging a polarized environment.
Had a hard time picking a winner and ran the creatives for 3-4 months but they were also not running huge budgets.
[💎@20:30] If you buy inventory the week before (instead of quarterly), which is heavily-discounted, you will have weeks where you can easily spend $200-300k but others where you won’t be able to spend all your budget.
Do not try TV ads if you don’t have 3 or more creatives. Make sure you have support from your C-level because it is a high touch channel.
[💎@23:00] Select agencies that understand your KPIs. A lot of agencies falsely claim they understand what mobile apps do. Have the agencies submit a request for proposal and provide you high-level media planning, and use a matrix to evaluate them.
Get your analytics ready (TVSquared).
[💎@24:44] During the test, ask your agency to have a weekly sync call to have a feedback loop with your management.
When you get started make sure you discuss the follow up budget more quickly and plan for success: think ahead.
Fabien also thinks they could have worked more closely with Japan beforehand (measuring baseline, going dark, etc.).
[💎@25:37] If you’re working with a global company that already has experience, make sure you tap into that.
[💎@26:15] Don’t hesitate to bring other people like UI/UX designers, product team, etc. to get feedback on the TV creatives.
[💎@28:09] In the US there’s a split in age: 30-40+ audience is still very much on linear TV, but the younger audience is only on it for specific events exclusive to specific TV channels (e.g. superbowl, political debates, etc.).
Netflix is the biggest streaming service and does not have advertising. Same for Amazon Prime Video and a big portion of Hulu subscribers. That’s a lot of eyeballs not available for purchase.
[💎@29:35] Streaming is limited (several services with no advertising) and fragmented. It is inefficient and has extremely high CPMs: compared to linear TV, you will pay 5x to 10x more for impressions.
[💎@30:35] For 18-40 audience, streaming also competes with influencer marketing in terms of video views and capability to measure the uplift. In many verticals influencer marketing might become more important.
Influencer marketing gives you less control. But there is also a diversification in terms of “products”: 5-10 second intro or midroll, influencer content not really mentioned, and we should start seeing more “produced” creatives etc.
A lot of influencer marketing platforms have not really done the analytics work to be able to measure incrementality.
#1 tip: always be testing. 80% on tested channels but 20% on testing new channels (ask platforms for credits!).
