What’s Mobex? It’s mobile context. And yes, it’s very different from context.

4 min readDec 20, 2021

We’re going to try and dispel some of the confusion surrounding contextual targeting in advertising and why it means different things in web and mobile environments.

Most of the conversation around contextual targeting has developed based on definitions borrowed from the web and force-fitting them into a mobile environment with the hope of solving the broader challenges facing the digital ad industry (e.g., privacy, data misuse, cookie deprecation, loss of identifiers, addressability). This has led marketers and advertisers to ask — what exactly is contextual targeting? And how can they use it to solve these broader challenges?

What is context? And why should you care?

To answer these questions we should first start with context and why it’s important.

According to the Oxford English dictionary, context means

the circumstances that form the setting for an event, statement, or idea, and in terms of which it can be fully understood.

So when we put something in context, it helps us better understand the meaning behind it.

For example, you wouldn’t try to sell ice-cream when it’s -20 degrees outside. You could — no one’s stopping you. But most likely it’ll be a waste of time. You’d greatly increase your odds if you tried selling mulled wine.

Why? Simply because you’re tailoring your offering to what the consumer is most likely to want given the circumstances. In this case, context refers to the fact that it’s cold and people are walking outside.

Thus, to live up to the age-old marketing goal of “Right Person, Right Message, Right Time’’, understanding the right context is absolutely critical. Getting the context right is often the difference between your message being considered as downright spam versus something actually relevant and meaningful for the customer.

So, how does Mobex differ from web context?

Now that we’ve understood the importance of context, let’s see how Mobex (mobile context) is different from the web (Note: we’re clubbing mobile web in the same group as web).

  1. Mobex is not limited to the confines of a web browser and can take advantage of sophisticated hardware to deliver rich multimedia experiences
  2. Users typically consume a whole range of content on mobile apps, often simultaneously (music, games, news, social media, utilities, etc.)
  3. Very low correlation between the type of content the user is consuming on a mobile app and their actual needs (e.g., simply because a user is listening to music or playing a game doesn’t mean they’re looking to buy sneakers)
  4. Mobile devices are used in various moments throughout the day

With these key differences in place, let’s now see how advertisers and marketers can leverage contextual signals and behaviours effectively in both web and mobile environments.

Context in a web environment

Ad targeting is based purely on the page content (including but not limited to text, audio, video or images). Not based on any user characteristics and therefore not PII-reliant (Personally Identifiable Information).

Web contextual signals

  • Vertical (e.g., Food, Dance, Travel), including sub-verticals (e.g., Vietnamese Food)
  • Content (e.g., keywords, images, emojis, GIFs, hyperlinks, videos)
  • Sentiment — content-based analysis (e.g. negative press and/or topic)
  • Device type (e.g., desktop, tablet, phone)
  • Domain & URL (e.g., .ai refers to AI companies, .org refers to non-profit organizations)

Web behavioural

These web contextual signals can be grouped together and observed over certain periods of time to create online behavioural cohorts that can be used by advertisers to reach relevant web audiences without requiring any identifiers. This is what Google is trying to achieve with FLoC and Microsoft with Parakeet.

Mobex: Context in a mobile in-app environment

Ad targeting is based on ID-less contextual data from the user’s device and/or the app.

Mobex signals

  • App-level (e.g., app category, app version, app ranking, app reviews)
  • User-level (e.g. session length, # of interactions/session, # of sessions)
  • Device-level (e.g., battery status, device storage, headphones in/out)
  • Sensor-level (e.g. accelerometer, GPS, gyroscope)
  • Content-level (e.g., listening to a rock song, browsing sports equipment in-app, playing level X in a game)
  • Live context (e.g., running, commuting, relaxing, in a car, at home, at a restaurant, housework, holding the device in hand)
  • Post-live context (e.g. post-run, back from holiday)

Mobex behavioural

Mobex signals can be grouped and observed over periods of time to develop behavioural cohorts (similarly to the web) that can be used by advertisers to reach relevant audiences without relying on identifiers.

These cohorts are based on real-world user habits and behaviours, with or without the presence of any identifiers. This is in stark contrast to the generation of traditional behavioural audiences which relies on identifiers to put similar people into groups and are often stored in centralised cloud storage warehouses.

Examples of mobex behavioural audiences include Joggers (uses app 2x a week while running), Frequent shoppers (uses app 4x a month while shopping), Bar magnets (uses app 2x a week while at a bar).

Hopefully, this has cleared some confusion about the differences between web context and Mobex and left you asking — “Ok great. Now how can I buy against these Mobex signals and behavioural audiences?”. If so, drop us a line!

Disclaimer: NumberEight specialises in developing Mobex software for mobile devices to detect the real-world context (e.g., running, commuting) from device sensors and packages them into ID-less audiences (e.g., frequent joggers, culture vultures). All computations happen on-device, thus producing a privacy-by-design solution.