Why Armilar invested in Mediaprobe…

… and a few insights on how emotional engagement is shaping the media industry

Armilar Blog
Published in
9 min readJun 7


“The glamorous, decades-long, Upfront Week is finally back in New York!”
Two years after the pandemic interruption, this was the predominant mood among publishers, media agencies and brands’ marketing execs just before the Upfront Week in early 2022. The Upfronts can be summarized as corporate events promoted by video publishers (mostly Linear TVs and streaming platforms) to showcase their debuting shows while negotiating their advertising inventory with brands and media agencies.

While “glamourous” might be our initial feeling to describe the Upfronts (considering the hosting venues like the Radio City Music Hall or live performances by top showbiz artists), the fact is that “deal-making” is probably a more faithful expression. Just to get a glimpse of the Upfronts economic impact, the consulting company Media Dynamics estimates than +$30 billion in advertising deals may have been negotiated in the 2022 Upfronts edition across multiple platforms.

When signing such big advertising deals, brands are heavily relying on the following concept also summarised in the diagram below: by paying publishers to insert ads attached to their most relevant media contents, brands are increasing the odds that viewers would get engaged when seeing that ad/show and that would positively impact the brand. In a nutshell, brands will pay according to the expected engagement to be reached, with “engagement” working like a predictor of ad impact.

A diagram explaining that brands pay publishers to insert ads attached to their most relevant media contents, in order to increase the odds that viewers would get engaged when seeing that ad/show and that would positively impact the brand.
Diagram of brands’ advertising investment process

At first sight, having the Upfronts such a multi-billion economic impact, we might expect that a highly sophisticated, transparent, and comprehensive process is being used to measure and trade on engagement.

Well… that’s not exactly the case. Brands and their media agencies still hang on to the old-fashioned engagement metric of impressions (meaning the number of times that an ad is displayed to an audience). As you may check in the illustrative chart below, by modelling engagement mostly through the impressions metric, the Upfronts might then be classified as a single-currency (impressions) ad trading marketplace, in which:

  1. Advertisers pay publishers to insert their ads attached to premium and debuting shows (since ad placement is considered to be as important as the ad creative content itself);
  2. That payment is associated with brands securing some expected level of audience engagement, based on the number of impressions. This is usually contracted as a percentage of an audience to which a specific ad will be displayed (aka Gross Rating Point) and paid according to the publisher’s cost per one thousand impressions (aka Cost per Mille);
illustrative chart modelling engagement mostly through the impressions metric
Engagement model based on impressions (illustrative)

But this strategy is far from being perfect … modelling the engagement of an ad mostly based on impressions raises several concerns, namely:
“How can a rating agency know if an ad displayed was actually seen? (e.g., a viewer switched on the TV in the living room, but then went to the kitchen while the ad was being broadcasted)”

An improvement to this scenario could be to move from impressions to reach. This means using some type of camera and computer vision algorithm that could track the viewer’s eye gaze confirming if an ad displayed was actually seen. But, again, new questions continue to pop-up:
“Even if someone saw the full commercial, how do we know if that ad had any positive impact on the viewer? (e.g., What did the viewer feel after viewing the ad?)”

With so many challenges we may end up bringing back a famous sentence associated to one of the modern marketing pioneers, the US merchant John Wanamaker: “Half the money I spend on advertising is wasted; the trouble is I don’t know which half.”

The new wave of engagement metrics… and potential multi-currency systems

Several entities in the media ecosystem have been working on introducing new metrics to better model engagement. A relevant example is the case of the Interactive Advertising Bureau (IAB) having introduced three clusters of engagement metrics (Cognitive, Emotional, Behavioral). Under such a framework, ad trading could move from a single-currency (impressions modelling engagement) to a multi-currency system (multiple variables modelling engagement), as summarised in the chart below.

Introduction of three clusters of engagement metrics (Cognitive, Emotional, Behavioral). Under such framework, ad trading could move from a single-currency (impressions modelling engagement) to a multi-currency system (multiple variables modelling engagement).
From single (impressions) to multi-currency engagement systems

The basic proposed concept is that engagement is neither a single-metric concept nor a one-time event. IAB uses the notion of inter-connected gears, meaning that engagement may need to be evaluated over time through different lenses:

  • The cognitive cluster tries to capture changes in the viewer’s awareness. These metrics tend to be mostly used in Linear TV advertising (as is the case of impressions and surveys);
  • The behavioural cluster is linked with the viewer’s triggered actions. Digital platforms heavily rely on these metrics (e.g., click-through rates), yet they are difficult to apply in linear TV and in some structured digital streaming platforms (e.g., how to interact with an ad on a TV?)

So, what about the emotional cluster of metrics…?
… Theoretically, they are one of the most relevant metrics since they should capture the viewer’s real feelings about a media content/ad/brand…
…In practice, how is it possible to measure emotions… and connect them with other metrics to generate multi-currency systems used by brands to negotiate more effectively with publishers?

Emotional engagement… as relevant as challenging to measure

Building emotional engagement metrics to effectively determine what an audience is feeling requires the evaluation of someone’s physiological activity. In a highly simplistic approach, summarized in the chart below, when a viewer is subjected to a given external stimulus (e.g., media content or ad), there is a perturbation in the viewer’s Sympathetic Nervous System (SNS), tied to the body’s quick involuntary responses. This perturbation drives the viewer to …
… achieve a temporary emotional arousal state;
… trigger specific physiological responses such as sweat glands’ secretion and heart rate;
… activate specific memory (recall) components around the emotional arousal peak.

Illustration highly simplified of a viewer subjected to a given external stimulus (e.g. a media content or ad) — perturbation in the viewer’s Sympathetic Nervous System (SNS), tied to the body’s quick involuntary responses.
Viewer’s emotional arousal index (illustrative)

In more practical terms, if we are able to monitor and build a model for sweat secretion and/or heart rate response, then we may have a predictor for the viewer’s emotional engagement.

Even though significant research continues to be done on how to model emotional arousal from measures of heart rate using non-invasive techniques (e.g., optically measuring the volumetric variations of blood circulation at the skin surface), one of the most robust methods to predict emotional engagement still relies on capturing the hand’s sweating level through the electric conductivity at the skin surface, a technique called Galvanic Skin Response (GSR).

Well, but mastering the GSR technology is not the same as having a scalable commercial product to measure an audience’s emotional engagement applicable in the media industry. For instance:

  • How to measure GSR in the most non-intrusive, miniaturised and seamless way?
  • How to link/sync acquired GSR data with non-linear media feeds and panelists’ activities?
  • How to provide actionable tools and business insights (and not just emotional data)?

Introducing Mediaprobe… the emotional engagement platform

Mediaprobe was founded in 2016 (under the name MindProber), leveraging neuroscience, electronics and data science know-how developed under the leadership of the company’s core founders, Prof. Pedro Almeida (CEO) and Prof. Nuno Dias (CTO). The company has since developed a biometrics-based media emotional engagement platform, grounded on 4 product components, as summarised below:

  • A — [The HW] Mediaprobe equips paid panelists with a wireless handheld device, capturing the GSR signal among other inertial data. The device remains attached to the panelist’s palm of the hand using stickers and an elastic band, enabling its use even when doing light exercise;
  • B — [The App] Panelists also use a mobile app that: (i) collects the data offloaded by the HW device and the audio feed; and (ii) registers optional user feedback to the media contents being consumed. (note that GSR only provides an amplitude measure of emotional engagement, a signal which can be further augmented by collecting user inputs on the app);
  • C— [The Cloud backend] This is Mediaprobe’s data processing layer: (i) syncing different data signals; (ii) filtering the data according to the panelist’s activities; (iii) enriching the media feed with tags, particularly inserted ads; and (iv) converting the GSR signal into a proprietary emotional engagement metric (Mediaprobe’s Emotional Impact Score — EIS);
  • D — [The Emotional engagement webapp] Web-based business insights and analysis application, merging: (i) EIS and audience data; and (ii) tagged media feeds with broadcasting data.
Representation of Mediaprobe’s product components
Mediaprobe’ product components

So, what is it so special about Mediaprobe?
Let’s put it this way … yes, they do have a proprietary robust miniaturised HW; … yes, they do have an innovative SW stack to link media feeds, with ads, with emotional engagement; … but, bottom-line, Mediaprobe solves business challenges by driving better ad spending and media production! To provide a more detailed explanation, let’s use some of the questions raised above.

  • How to measure GSR in the most non-intrusive, miniaturized and seamless way?
    Mediaprobe uses a certified small wireless handheld device with optimized embedded electronics generating a stronger signal-to-noise ratio, avoiding the need for large and wired systems. Seamless handshaking and calibration processes also simplify the sensor setup and daily use. Some of Mediaprobe’s clients are collecting data from panelists when they are at home, commuting or when doing light exercise;
  • How to link/sync acquired GSR data with non-linear media feeds and panelists’ activities?
    Mediaprobe uses audio content recognition (or ingests publisher’s logs) to track media content. It then merges that feed with timestamped GSR and panelists’ activity profiles (estimated based on inertial data). This allows Mediaprobe to have as clients leading US linear TV broadcasters as well as top US/EU video & audio streaming platforms;
  • How to provide actionable tools and business insights (and not just emotional data)?
    Mediaprobe provides a business-oriented SW layer with (i) augmented media feeds incorporating tagged commercials; and (ii) an emotional engagement score as a baseline to compare the effectiveness of different media contents and ads.

The unleashing of new ad trading and content production use-cases

So far, we have been arguing that emotional engagement is key to solving several advertising challenges within the media industry … which is particularly relevant considering how media is increasingly dependent on ads. Take as an example the case of streaming platforms Netflix and Disney+ which both launched in 4Q22 ad-supported pricing tiers.

But we can take a step further and claim that measuring someone’s emotions to a piece of media content not only provides brands and publishers with a tool to trade ads based on effective impact, but it is also a tool to address other market needs such as media producers being forced to have more emotionally attractive shows, while publishers fighting to have more engaging productions feeds.

In the chart below, you may find how Mediaprobe is addressing several use cases around ad trading and content production, with an attractive value proposition for three key media players:

  • Content producers (think about movie studios or advertising agencies) may rely on Mediaprobe for scalable media pre-testing (instead of lab-based, survey-oriented ones), evaluating how different production elements (scenarios, script, etc.) affect second-by-second audience engagement and churn probability. This has the potential to generate more valuable contents to be sold to publishers;
  • Publishers may use Mediaprobe to do live testing of new broadcasting formats (e.g., new camera viewing angles in sports, new commercial breaks formats, etc.), creating more engaging events for viewers and more valuable feeds for brands to insert their ads on;
  • Brands may leverage on Mediaprobe’s engagement metric to access the effective impact of their ads and compare the expected engagement of different advertising spots. This could optimize their advertisement ROI.
Chart on Mediaprobe’s use-cases around ad trading and content production, with an attractive value proposition for three key media players
Mediaprobe’s use cases

But that’s not all! Beyond those more traditional use cases, Mediaprobe could also be the enabler of innovative concepts leveraging multi-platforms and dynamic personalised feeds. Take the following two publishing examples: (i) Activation of a 2nd screen (e.g., posting an ad on social media) may be triggered when a particular engagement is detected on Mediaprobe’s platform; (ii) For a given show, Mediaprobe may be used to adjust the ads attached to the show when broadcasted on Video-on-Demand platforms based on the real impact that the show had when broadcasted in Linear TV.

In short, we believe Mediaprobe has the capability to cause disruptive changes in this industry. Pretty soon, in an upcoming Upfronts edition, with the help of Mediaprobe, publishers will have more emotionally impactful shows with an increased ad inventory, while brands and media agencies will have a higher advertising ROI by using multiple engagement metrics — including emotional impact!

Authored by João Dias, Armilar



Armilar Blog

Armilar is Portugal’s leading venture capital funds manager, an independent VC with a 20-year-old high-performance track record and an international footprint.