Shooting fish in a barrel — The Netflix ad platform experience.

You may have recently heard that Netflix is releasing an ad-friendly, lower-priced tier of their subscription, as well as an ad platform for advertisers to run said ads. Netflix is a channel with massive potential for advertisers, with more than 220M subscribers worldwide. So, this sounds like a pretty great opportunity for a lot of brands. Right? Well, maybe not.

To understand why let’s take a look at some of the features of the platform. When Netflix announced the new pricing tier and platform, they gave us a closer look at what to expect from the ad platform. Prepare yourself, though, because it isn’t pretty. It’s honestly quite strange that they came right out with all of this detail considering how far from ideal it is for modern advertisers, but let’s dig in anyways.

A quick summary should work sufficiently to shock you:

Targeting options restricted to:

  • Geo-targeting (United States, Canada, etc.)
  • Genre-targeting (i.e. horror, comedy, drama, etc.)
  • That’s it

Something that dawned on me only as writing this is that there are huge implications in targeting children here. With no demographic targeting, there is no way to segment certain age brackets. It’s possible that they aren’t going to allow targeting of kid-friendly genres, or there will be strict review processes and guidelines in place to effectively ‘protect’ this segment. I suppose time will tell on this one.

Anyways, if you think that sounds bad:

  • No machine-learning optimization of ads
  • No meaningful measurement (you can ask a rep for an export of the data, lol)

Okay, it gets worse:

  • $65 CPM to start (they’ve confirmed this goes up after initial testing)
  • $1M minimum investment

Despite being a leader in digital media, they’ve taken a very ‘traditional advertising’ approach to their ad platform, which is incredibly disappointing. Netflix has a wealth of data on its users, so why can’t we unlock that? Perhaps down the road we will get more of that critical ‘data’ and ‘analytics’ we pesky advertisers are looking for, but until then, the rabbit hole actually gets a bit deeper…

After looking at the press releases from Netflix about this, something didn’t seem quite right. I mean, $65 CPM is high, but if the return on investment is high, that would be fine too, right? But what if the platform’s numbers aren’t actually all that great? Does the $65 CPM still make sense? I started crunching the numbers to better understand how this ecosystem might function. I discovered that the true cost to advertisers is much higher than you probably think.

Let’s get into the meat and potatoes of it.


Let’s start with my hypothesis:

There is a ‘real’ CPM (separate from what Netflix announced), based on several factors, that is much higher than the $65 presented.

This hypothesis comes from a hunch that ad viewership on Netflix will be as low as on traditional TV. With traditional TV ads, many people flat-out mute their TVs, leave the room or go on their phones. Compared to digital channels, a served impression is still delivered to a person in their feed - they literally see the ad - but choose not to engage with it. In contrast, with traditional TV, most of the impressions you pay for are literally not seen by the person they are served to.

I realize that this is just the nature of the channel, as there is no way for the platform to make people watch the ads, but from a digital marketer’s perspective, many of us will assume that we are getting 1,000 impressions for that $65, which is not the case.

To uncover the ‘truth’ about some of the numbers Netflix was reporting and to attempt to validate the hypothesis, I had to devise a method of calculating my figures. To get to where I needed to go, I had to find data on things like the average TV show and movie duration, the number of ad slots Netflix was going to place in various formats, figure out how many ads could be delivered per hour, Netflix subscription rates, TV ads retention rates, platform adoption rates, streaming hours per day before and after COVID, as well as a few other things to help piece it all together. Needless to say, there’s a lot to get through here, so let’s get to it.


(skip to THE GOODS if you don’t want to fall asleep)

Data can be sexy, right? Well, maybe conceptually, or if I had some incredibly elaborate, colour-coded spreadsheet to share with you, but I don’t. Instead, you’ll get the unsexy version – lots of numbers and no bright colours (I’m sorry).

So, where to begin? Let’s start at the top. To calculate the total potential impressions Netflix could serve, we have to figure out what types of placements they plan to run. Thankfully, Netflix told us exactly what the plan was at launch: 2-minutes of ads as a pre-roll for movies and 4-minutes of ads as interstitials in TV shows, with only 15-second and 30-second formats. With that, we can start crunching the numbers.

The average TV show length is 32 minutes (a 30-minute TV show is actually only 22 minutes long, and a 60-minute TV show is actually only 42 minutes long), and the average movie length is 95 minutes. With this, we know that the total average length across all mediums in Netflix’s library is 63.5 minutes in duration, or 1.06 ‘assets’ (shows or movies) viewed per hour. Using what we know about the ad lengths allowed at launch, we can determine that there will be an average of 5.51 ad placements per hour.

Alright, the first number is down: 5.51 ad placements per hour.

What’s next?

Next, we need to figure out the average hours of streaming time per subscriber per day. This is relatively easy. To calculate this, we just have to figure out the average viewing time before COVID and during COVID because these two numbers are dramatically different. It’s worth noting that this calculation will actually leave us with an overly optimistic number because COVID restrictions and regulations have decreased dramatically, which will naturally lead us to streaming numbers closer to pre-COVID, but these exact numbers aren’t out yet, so I am giving them the benefit of the doubt. So, for now, we calculate the average number of hours viewed by subscribers per day as 2.19. This is an average of the total hours per day pre-COVID (1.18) and the total hours per day during COVID (3.2).

Okay, the second number is down: 2.19 average hours streamed per day per subscriber.

With that, we can calculate a few important numbers. The average ads per person per day and per month.

Multiplying the average ad placements per hour by the average number of hours streamed per day, we get 12 (rounded down because you can’t include half an ad) ads served per hour per subscriber. To get our average ads per person per month, we just multiply that number by the average days in a month (30.437), which leaves us with 365 (again, rounding down) ads per subscriber per month.

Phew, alright, we’re almost there. 365 ads per subscriber per month.

What’s next? The last step is to take a look at Netflix’s subscriber numbers, as well as a few supplementary metrics.

First, Netflix’s current subscriber count is approximately 220,000,000. Next, we need the number of people who are likely to subscribe to the new, lower-tier version of the subscription. To get this, we need to take our total subscribers and multiply it by the adoption rate. Netflix, itself is estimating that approximately 18% of its current subscriber count will adopt this service by the end of 2023. With that, we can then calculate that about 40,000,000 subscribers will receive ads.


To tie this all together, we need to add one final key piece of information: viewership rates.

When we look at traditional TV placements (the closest thing we have to compare), we see pretty terrible retention rates when it comes to ad viewership. In fact, only an estimated 24% of people who watch TV even let ads play. This means that when ads appear on TV during their shows, 76% of audiences outright mute the TV, leave the room, or start to look at their phones until the show reappears.

You might be thinking: “You can’t use that to calculate a ‘true’ CPM because CPM doesn’t care whether you’re engaged with the ad or not,” to which I would reply, “you’re right.” It doesn’t, but if you think about impressions on digital platforms, the audience still sees the ad, they just decide not to engage with it. In the traditional format, audiences are literally not seeing your ad, despite it being delivered to them. Of course, there is no way for Netflix to discern whether or not the ad is being watched, but it does definitely imply that approximately 76% of the impressions you are paying for are effectively being flushed down the drain. In other words, 76% of the $65 CPMs that you’re paying are for impressions that will never been seen.

Again, the reason I mention this is that advertising has evolved, and the majority of digital advertisers are looking at Netflix’s new ad platform through the lense that the word ‘impression’ should be very literal - If I am paying for an impression, it should be seen by the person I pay to have it delivered to, otherwise what am I paying for? The opportunity to maybe serve an impression? If this is the case, many people will be surprised by the performance this drives, and therefore my goal is to clarify this $65 metric by putting it into terms that most people will be comfortable with.

So, I will continue running the calculations here with this in mind: only 24% of subscribers in this tier will see the ads you pay for.

There are a lot of calculations that I run with this data, so I will drop a very pretty screenshot of an excel file here and walk you through it.

Using the 24% ad retention rate metric, as well as the other metrics discussed above, we can estimate that the true CPM for advertisers is closer to $271.

Again, while you may be paying $65 for 1,000 impressions, you are technically only getting 240 of those impressions for that $65. If we recalculate this to be an actual 1,000 impressions, the cost is closer to $271 on average.

Also, I am acutely aware that I could have just taken the retention rate and CPM and run this calculation in a much less roundabout way, but what’s the fun in that? Now I get some idea of how much ad inventory Netflix has, how many advertisers they can cater to before saturation, etc. The “fun” stuff, I guess?

To make all of this worse, this doesn’t even account for other factors. Let’s not forget that there are no real targeting options, which means a reasonable percentage of that 24% will be outside of your target gender, age, or interest groups, etc. Further, there is no machine learning built into this, which is actually mind-blowing. Without that, your ads won’t be able to optimize themselves over time, improving performance with more dollars spent. This is a recipe for burning cash.

The entire point that I am making here is that there are a lot of inherent issues between what we, as advertisers, expect and what Netflix’s traditional advertising approach can deliver. When a media giant like Netflix says they are bringing opportunity to advertisers, it’s easy to get excited, but this is a far cry from anything that demands celebration.

Even traditional advertising is becoming smarter – take Decibel, for example. They are a self-serve audio ad platform that lets you run ads on the radio (among other placements), with deep targeting and customization options, and even provides analytics and potential attribution solutions. It is not only possible to innovate on the traditional media format, but it’s happening right before our eyes. So, when it comes to a company like Netflix, with a $107B market cap, you have to ask: “What’s the excuse?”


One thing that I should probably acknowledge is that this ad platform doesn’t need to be for everybody. It could be built in a more industry-conscious way, but at the end of the day, eCommerce brands aren’t likely to be Netflix’s target audience in the first place.

So, who would use the platform? Well, I think there is still plenty of opportunity for brands here. Take UberEats, for example. At a certain level of scale, simply getting in front of people becomes the goal over direct-return initiatives, and I can’t personally think of a better outlet for UberEats than Netflix. Imagine the opportunity to run ads for snacks right before a movie starts - a true theatre-like experience in your own home.

That said, it’s disappointing because this platform could have been much more available, which would have been a win for the industry. We will just have to wait it out and see how the launch goes. Who knows, maybe Netflix will come to their senses and add in some of that wealth of data they have for targeting, build in some sort of optimization and analytics, and lower the barrier to entry.

I’m not holding my breath, though.

Share this post