Analysis / Mobile Location / Mobile Marketing

Targeting Mobile – Building Profiles

This article is part 3 of an ongoing series about Mobile Targeting. In the first 2 articles  we talked about the difficulties with tracking customers on mobile devices and the limitations with cookies.  We reviewed some of the techniques the mobile industry is using to track mobile customers for the purposes of advertising and identified that customers cannot be uniquely identified as they move between online and mobile.

You can find the first two articles here:

In this article we look at the problem of cross-domain tracking and the need for advertisers and ad networks to create behavioural profiles for users who move between online, mobile web and mobile apps.  We review techniques such as fingerprinting and how some innovative companies are using complex algorithms to analyse all manner of data to build up highly accurate profiles without using personally identifiable information (PII). Finally we look at the kinds of data that mobile advertisers want to track to make advertising more effective.


There have been numerous mobile advertising startups in recent years, touting new technology that improves audience targeting and solves the problem of cross-domain tracking.  Many propose to use techniques that have come to be known as ‘fingerprinting’, whereby hundreds of different pieces of anonymous data are analysed to create a profile that can be uniquely matched to a particular user.  This technology is also referred to with terms like device bridging and signature matching.

… if a particular device clicks on an ad for an app, then moments later a device with very similar characteristics is identified downloading that app, there is a very high chance it’s the same device.

These techniques are not 100% accurate, but services claim as much as 70 – 90% accuracy based on increasingly intelligent algorithms.  To illustrate, let’s look at a scenario where a mobile ad is served into a web browser environment.  Typically ad networks are unable to measure if mobile click throughs lead to application sales.  This is because there is no way of tracking  the user outside the mobile browser and into the app download.   However by collecting some characteristics of a device, it’s possible to make an intelligent match between the device that saw the ad and a device that downloaded the app. So for example if a particular device clicks on an ad for an app, then moments later a device with very similar characteristics is identified downloading that app, there is a very high chance it’s the same device.

Several startups in the last few months have announced a focus on connecting desktop cookies to social media accounts.

The obvious benefit of device matching is it’s potential for cross-device tracking. By collecting non personally identifiable information (PII), mobile ad companies can start to build profiles that check out across multiple devices.  These techniques don’t stop at matching device characteristics.  They are also building up databases of customers with particular behaviours, patterns of browsing and social media usage that can be algorithmically matched across environments. Several startups in the last few months have announced a focus on connecting desktop cookies to social media accounts.  They can then pick up these customers on mobile devices based on their social media accounts, thus tracking and connecting a single customer across multiple devices. By combining device bridging with social media matching, advertisers can deliver targeted messages on any device based on activities or connections within a social media account.

By collecting non personally identifiable information (PII), mobile ad companies can start to build profiles that check out across multiple devices.


The point of all this tracking, bridging, matching and identifying is to allow advertisers to deliver targeted and contextually relevant messages to customers who are willing and ready to receive them.

Once we overcome the challenge of being able to track in the mobile space, we then move onto the question of what it is that we want to track.  In mobile advertising this generally boils down to the following categories of information:

  • Device
  • Context
  • Audience
  • Location
  • Behaviour

The world of mobile audience targeting is loaded with terminology that is sometimes interchangeable, and often quite confusing, so lets break each of these down.

Device Targeting

In the early days of mobile advertising, when very little data was captured about mobile users, often the best we could manage was Device targeting. This was about the type of phone or browser a customer was using, and later, whether they were on WiFI or a 3G network. Although rudimentary, it was possible to make sure you were only advertising to customers who could actually access your site or download your product.

Contextual Targeting

On top of this base level, many ad networks started to group audiences by context. Contextual targeting is usually based on the publisher’s (or ad network’s) categorisation of their mobile sites and apps.  Simple contextual targeting categorises the content, rather than the customer, and allows advertisers to access customers when they are viewing particular type of content, such as sporting results, mobile games, social networks, women’s interest or auto content.  

Audience targeting

Mobile Audience targeting builds on device and context.   It is a broad term that encompasses a variety of personal but non-identifying details about customers.  Most commonly, in the digital space, audience targeting incorporates demographics such as age, gender, income, education, employment and interests.  Audience targeting can also include location details, but generally at the very broad address level.  eg. the customer lives in NSW.   It is still very early days for audience targeting in mobile.  To be effective, mobile audience targeting would rely on known data provided by a brand, media provider or publisher that has been collected from their existing or potential customers.    

Location Targeting

Geographic targeting, also known as geo-targeting, location targeting, LBS or proximity targeting goes beyond simple address data and involves the more transient nature of customer movement.  Geo-targeting means understanding a customer’s current location and its immediate relevance.  It usually involves positioning a customer using the GPS transmitter in their smartphone or other mobile device. Location targeting can also refer to Assisted GPS (A-GPS) which uses information from the WiFi or Cellular network to make assumptions about a customer’s location. 

Behavioural Targeting

Behavioural targeting identifies and tracks a customer’s activities such as viewing or searching for particular content, purchasing products or engaging with a site, app or game. This data is usually collected by tracking and monitoring mobile users as they move about the mobile and desktop web. Behavioural data could also be aggregated by analysing a customer’s geographic movements and comparing these to known physical locations, such as retail stores, sporting grounds, airports or restaurants.  

While device and audience data is generally fixed for the longer term, location, context and behaviour changes over time. Thus targeting criteria might include fixed data, such as Women over 35 who have an iPhone and live in Melbourne.  Alternatively, targeting might be based on conclusions drawn from historical data such as People who engaged with a social network while at the MCG on a Saturday during winter when Carlton was playing AFL rules.

Tracked, historical, location and contextual data can shed a great deal of light on a customer. Contextual data can be built into historical profiles that track a customer’s mobile content interests over time.  When this is matched with specific behaviours such as visiting certain retail stores or frequenting airports and coupled with demographic or historical contextual data that identifies interests or life stages, the aggregated knowledge can go a long way to creating a valuable audience profile.

By recording data about a customer’s physical location over time, along with their browsing history, app downloads, online purchases and social connections, a detailed profile can be built up without actually storing any Personally Identifiable Information.


This is where we get into the complex world of Audience Profiling. Audience profiling can mean different things to different people.  Generally an audience profile is understood to mean a group of potential customers that have been anonymously identified, tracked & allocated to a targetable group based on factors like interests, life stages or purchase intent. Some organisations do store personally identifiable information in their audience profiles, and this is generally a matter of policy and appropriate protection of privacy rights.

An audience profile is often a combination of first party data, historical data and contextual and location data that has been collected by a mobile ad network. For example many advertisers are interested in targeting groups like Mums, Business Travellers or House Hunters.  These audiences are identified by a variety of means.   By collecting first party data from publishers and combining it with historical contextual, location and behavioural data, certain assumption can be made. Sometimes an audience is known, such as expectant mums who have signed up to receive a weekly newsletter during pregnancy.  But even without that known data,  female customers aged 25 – 40 who regularly browse baby and pregnancy websites can safely be assumed to be hopeful or expectant mothers.

In our final article in this targeting series, we look at how some companies are creating pre-packaged and customised mobile audience profiles for advertisers to use.  We also investigate the practice of mobile retargeting, and how social networks like Facebook are making it work for them.  Finally we delve into a future where mobile wallets make it possible to target, track and retarget customers all the way through the sales funnel from awareness to purchase.


  • Until recently it has been difficult to track users between sites & apps and between online & mobile.
  • New techniques like fingerprinting, device matching & signature matching are used to collect and analyse data and build up anonymous profiles.
  • Publishers and ad networks use data about a customer’s device, context, location, audience and behaviour to deliver targeted advertising messages.
  • Audience profiles are created by aggregating data from multiple sources and tracking techniques, often without the need to store any PII.

The 4th and Final Part of our Targeting Mobile series is now live at Targeting Mobile – The Holy Grail 

Tags: , , ,