Are We Building CTV From the Wrong Blueprint?

CTV runs on digital infrastructure, but it is not an extension of desktop and mobile. If the environment is different, the architecture and the measurement probably should be too.

Last week I attended two sessions about CTV by IAB MENA, and I learnt a lot about the world of CTV.

I should probably start with a disclaimer. I am not a CTV expert.

My background is broader across programmatic, publisher monetisation, ad tech and supply-side architecture. But these days I am deliberately trying to go deeper into CTV because I find this whole area fascinating. I feel the same about DOOH and Retail Media, which I have already had the opportunity to work with.

What is so interesting is that they all increasingly use digital infrastructure, data and programmatic technology, but none of them are simply extensions of desktop and mobile advertising. The environment, the behaviour and the inventory are all fundamentally different.

So, increasingly, I find myself wondering whether the measurement and execution should be different too, and this is what really interests me. I also want a good understanding of whether we are actually designing these ecosystems around how the medium works or simply applying the digital advertising models we already know.

The more I learn about CTV, the more one question keeps coming back: are we building it from the right blueprint?

First, what exactly are we talking about?

CTV conversations can become acronym soup surprisingly quickly. So before questioning the architecture, it is worth separating a few things.

CTV, Connected TV, is essentially about the device environment. Basically, a television connected to the internet, either directly as a smart TV or through another connected device.

OTT, Over The Top, is about content delivery. It means content is delivered over the internet rather than through traditional broadcast infrastructure. OTT can therefore be consumed on a television, mobile phone, tablet or computer. They overlap, but they are not interchangeable.

Now, let us talk about the commercial and content models.

SVOD: Subscription Video on Demand. You pay a subscription to access content.
AVOD: Advertising-Supported Video on Demand. The viewer accesses content supported by advertising.
TVOD: Transactional Video on Demand. The viewer rents or purchases individual content.
Hybrid: Normally a service combining subscription and advertising models, for example a lower-cost ad-supported subscription tier. Like Netflix in the UK.
BVOD: Broadcaster Video on Demand. On-demand content originating from traditional broadcasters. It describes the broadcaster relationship more than one universal monetisation model.
FAST: Free Ad-Supported Streaming Television. FAST recreates some of the experience of linear television: scheduled channels that you tune into rather than selecting every individual programme, often with curated content and a familiar channel-style experience. The delivery infrastructure is streaming, and the service is funded through advertising.

(Source: IAB MENA, CTV & OEM Capabilities in MENA)

And this immediately raises my first reservation. FAST is not a buying method, and it does not automatically mean programmatic or RTB. This inventory can be sold in numerous ways: traditionally directly, or through guaranteed deals, PMPs or auctions.

FAST may look like linear television. That does not mean the ecosystem is linear.

This distinction matters a lot. Traditional television had a relatively understandable structure: content, channel, commercial break, distribution, viewer.

CTV and FAST can introduce considerably more participants, starting with the content owner, then the FAST channel operator, streaming platforms, ad servers, SSPs, exchanges, resellers and DSPs. Then the curation layer. Measurement vendors and identity providers are just some of the other players that can be involved.

By nature I want to understand what is behind the scenes and peel all the layers of the onion. So when someone tells me that FAST inventory is rapidly expanding, my immediate reaction is asking key questions such as:

Whose inventory?
Who actually controls the ad decision?
Who measures the exposure?
And how many different companies touch that impression before it reaches the advertiser?

Why? Because all of these relationships can have an impact on the economics, transparency and ultimately the value of the impression. We need to be clear on these things. We do not want to create another black box that raises questions around trust, right?

I believe this is a valid architecture question. Especially when supply becomes more fragmented, it becomes increasingly important. This is also why supply-path transparency matters and why tools such as sellers.json and schain exist.

(Source: IAB Tech Lab, sellers.json and SupplyChain Object)

Oh, and let us not forget about the OEMs.

Another area I find particularly interesting is the growing role of OEMs, Original Equipment Manufacturers. Think smart-TV manufacturers and operating environments. They are no longer simply manufacturing the screen. Some now control interfaces, content discovery environments and advertising inventory with home-screen placements, screensavers, pause ads and native ads.

So, this is where I think again. A home-screen placement is not the same thing as an in-stream video advertisement. One happens while somebody is navigating their television. Another appears during content they actively chose to watch. A pause ad is different again.

So even within CTV, we are talking about very different advertising experiences. Thanks to IAB and other working groups, the industry has been working to standardise these emerging CTV formats, and I hope this will bring us to a new era of CTV ads.

(Source: IAB Tech Lab, CTV Ad Portfolio)

But standardising the format is not the same thing as standardising its value, and that brings us to measurement, which is still a challenge in many ways.

Measurement. But measurement of what?

Probably this is the area where I have most of my questions. We often talk about one of CTV benefits being the ability to bring digital-style measurement to television. But what are we actually trying to measure?

Reach? Attention? Brand lift? QR scans? Frequency? Incremental reach? Conversions? Or what is the main focus and the real value we are trying to understand here?

Because those are completely different things, and trying to collapse them into one universal measurement model does not make much sense to me. Fragmentation and inconsistent signal quality remain real challenges across impression measurement, viewability, reach, frequency and attention.

(Source: IAB, Standardised Measurement Guide for CTV)

Are we trying too hard to make CTV behave like digital advertising?

Look at the language we have carried across: impressions, IDs, viewability, frequency caps, attribution and so on.

All useful, but there are so many other things that need to be considered to really understand the value of an ad exposure in CTV. An advertising exposure can influence someone who purchases later, somewhere else, using a completely different device. The fact that a video was technically played to completion does not mean anybody watched it.

So instead of continually asking "how do we make CTV measurable like digital advertising?", I would ask: what would we design if CTV advertising were built around the environment itself?

And perhaps in some CTV use cases, household-level measurement and decisioning may simply make more sense than trying to force everything into an individual user model.

Which is why attention becomes very interesting.

This is an area I want to understand much more deeply. Because if CTV is fundamentally different from desktop and mobile, perhaps attention can become one of the signals that helps us measure quality differently.

An impression tells me that advertising was delivered, and a completion tells me that playback reached the end. Neither necessarily tells me that a human being watched it. Attention asks something more interesting: was anybody actually paying attention?

This is why attention is increasingly being treated as an important complementary measurement layer rather than simply an experimental metric.

(Source: IAB / MRC, Attention Measurement Guidelines)

In a CTV environment this becomes particularly interesting because different approaches can look at signals connected to the viewing experience, including presence, co-viewing and attention, depending on the measurement methodology.

(Source: TVision, CTV Measurement)

In my opinion, there is a really important distinction here. Attention is not the same thing as identity targeting. I would not state with 100% confidence that "this person is attentive, therefore target this person". That takes us straight back into many of the identity assumptions I am questioning.

But I truly believe attention can help identify or prioritise media environments with a greater predicted likelihood of earning attention, and can be one of the signals helping us better understand the quality of an exposure. Attention data is already beginning to influence planning, PMPs, pre-bid optimisation and bidding models.

(Source: Adelaide, Attention and Programmatic Activation)

That is a very different philosophy from traditional behavioural targeting, and perhaps a more natural one for television.

A television ad break is not five unrelated display impressions

Television already taught us something useful: the entire commercial break matters.

An ad pod is essentially a group of ads playing consecutively within an ad break. OpenRTB 2.6 introduced more specific support for pod bidding, helping the sell side communicate more information about the ad break itself.

(Source: IAB Tech Lab, CTV Programmatic Guide / OpenRTB 2.6)

Technology adapting to the viewing experience, and I hope we keep it that way. Otherwise, we risk simply importing some of the challenges digital advertising has already experienced onto the television screen.

Live CTV streaming

In my view these are some of the most valuable television environments, although programmatic buying can be particularly challenging here. This comes from the nature of it. Audience levels can change very quickly, large numbers of concurrent streams may need to be supported, latency matters, and forecasting becomes harder. The infrastructure therefore needs to be able to react and scale accordingly.

The fact that the industry is developing dedicated live-event programmatic standards reinforces the same point again.

(Source: IAB Tech Lab, Live Event Ad Playbook)

When the environment changes enough, the infrastructure has to change with it. This is not a nice-to-have. In many cases, it becomes a must-have.

We call everything "premium" when we talk about CTV

CTV inventory is frequently described as premium, and this is sometimes absolutely correct. But a television screen alone does not automatically make an impression premium. A lower-quality FAST environment with excessive advertising, weak metadata, repetitive creative and an unclear supply chain does not magically become premium because the viewer owns a 65-inch television.

To me, premium needs to come from a combination of content, audience, environment, experience, transparency and measurement. The screen, of course, contributes to that value, but it does not define it.

We love to use QR codes. They are useful. They are not an attribution strategy.

CTV also creates a completely different interaction challenge. So QR codes, interactive overlays and other formats become bridges between the TV and another device. I like that, and I am actually obsessed with QR codes and how creative we could be by using them.

But equally, I have to admit that scanning a QR code is an engagement event. It is just not proof of the entire value of an advertising exposure. Perhaps this thinking is inherited from digital advertising: our constant desire to make one observable action explain the entire customer journey.

AI still needs good architecture

Obviously AI will increasingly sit across all of CTV and the whole digital advertising ecosystem, and I see great potential in using the tech where it makes sense and where it has been properly tested and verified.

I want to stop here and say that I do not think we need to be in a race with AI. The competitive advantage does not come from simply having AI. It comes from having useful, efficient and trusted AI, supported by strong data and infrastructure.

But AI does not eliminate the architecture problem. If inventory taxonomy is inconsistent, identity is misunderstood, supply paths are opaque and measurement signals are weak, then AI still has to work with those limitations. Technology can automate existing fragmentation just as easily as it can help solve it, so the architecture underneath still matters.

This is also why I am interested in DOOH and Retail Media

The more I explore these areas, the more I see a common thread. CTV, DOOH and Retail Media sit somewhere between digital infrastructure and environments that behave very differently from traditional digital media.

Retail Media introduces stores, commerce data, loyalty relationships, product availability and closed-loop measurement.
DOOH introduces physical location, movement, audience modelling and one-to-many exposure.
CTV introduces households, content, co-viewing, attention and long-form viewing.

They can all use programmatic technology, but programmatic technology should be the enabler. It should not automatically become the blueprint.

So what would a CTV-native blueprint look like in my eyes?

I have to be honest, I do not have the answer. And I probably need to do more research on the topic and interview CTV experts. But I am curious, I naturally like to question how things fit together, and this is exactly why I am exploring it.

We can take the best parts of digital: automation, addressability, data, optimisation, accountability.

And the best parts of television: content, context, storytelling, attention, shared experience.

Then design the system around both and rethink some of the following as something genuinely native to the environment: identity, measurement, targeting, frequency, pricing.

CTV is great, but the mindset around it probably requires a little more openness to thinking beyond the traditional digital playbook. And I am sure, with this industry resilience, we will keep moving in that direction.

For me, the question is not whether digital technology belongs in CTV, because of course it does. The question is whether we are adapting that technology enough to the environment itself.

~ written by Sophie Toth with the help of an AI editor ~

Sources and further reading

Next step

If this describes an asset base you already control, the next step is a diagnosis rather than a technology decision.

Start a Transformation Conversation More Insights