As highlighted in the previous article, comprehensive, granular and accurate data about locations, audiences and behaviours is of paramount importance for advertisers. This data enables them to effectively plan their marketing activities and choose the right message, for the right audience, at the right time and location. These days, there are many data sources available in the Digital-Out-Of-Home industry for that purpose. Let us break them down below and analyse challenges and opportunities associated with each of them.
The number of people walking past their ad is one of the most important reference points for an advertiser to assess the advertising potential and actual performance. People count information is available from many sources, with varying levels of accuracy and granularity. Many environments have long been collecting this data for their own planning purposes. Many shopping malls or retail outlets are using people counting sensors, either infrared or camera-based. In most cases this data is good enough to understand patterns in visitation and average occupancy, although it normally comes with some challenges for a potential use in advertising.
First, this type of data is prone to over-duplication due to return visitors counted again and again. Second, this data may be hard to access as not all location owners are happy to provide it to third parties. Third, the people count figures collected at the store entrance may not directly correspond to the number of times a screen could be seen, if it is not installed right at the gate. And ultimately, oftentimes, there are different technologies installed at different locations, and some locations may not have sensors at all, and so the overall uniformity of the data may greatly vary.
Still, in some environments, people counting data can be used to estimate the total addressable audience of a campaign, if available.
Vehicle traffic measurement
Vehicle traffic can be efficiently used as a proxy for people count for road-side billboards, drivethru environments and car parks. This data can be collected using a variety of methods and technologies. For road-side, the historical vehicle counting information may be available from local councils or road authorities. It may or may not be easy to access. This data is sometimes used by DOOH industry associations for modelling purposes.
In some car parks, the data may be available from landlords or operators, if they count entering and leaving cars using cameras or boom gates.
Other popular methods involve cameras installed overhead – on the billboards or other fixtures – to get real-time counting and qualifying data, such as vehicle class, or even make and model. In some jurisdictions it is even possible to collect license plate information to detect repeat visits, map routes and perform deduplication of counts across multiple locations. This is a powerful but expensive method with a potential of giving a lot of data besides basic counting – speed, colour, demographic information based on the license plate, elements of the socioeconomic profile and preferences based on the vehicle class and model.
There are also ways of tapping into some common navigational services, like Google Maps, to get the speed, density and count data for a particular location.
Satellite data is also available from some providers and is based on the satellite photos of particular outdoor locations for the purpose of vehicle traffic estimation.
Vehicle count figures provided by any of those methods cannot be directly and easily translated into the number of people. To get that translation done, providers, publishers and brands are using multiplication factors reflecting the average number of people driving a car. Such coefficients are based on surveys, statistics and are usually static, as they do not change often due to the difficulty of their accurate estimation. They may also significantly differ by geography, location and time of day.
If done right, the vehicle traffic measurement data can provide very valuable metrics and insights about road-side digital media.
Anonymous Video Analytics
One of the most powerful way of collecting data about people in the “viewable” area of a digital screen, those with a potential to see it and those who actually saw it, is leveraging video sensors equipped with a smart Anonymous Video Analytics software solution capable of performing Anonymous Audience Measurement. The strength of this approach is in the richness of the data, including the total audience and watchers, conversion ratios, demographics, dwell and attention time, mood, distance and many other data points extracted from visual cues. For privacy-protecting reasons, that strength is counter-balanced by the inability to identify unique and repeat watchers, so the data may contain multiple detections of the same people.
Public and private datasets
Generally available aggregated data about citizens, their socioeconomic and demographic profiles, suburbs they live in, and other particulars can provide a great overlay on top of any dynamic or real-time data collected via other means. Detailed census information is collected and published every few years by governments. This data can be used to more granularly plan campaign execution and better address target audiences.
Similar to this, some brands can utilise their own customer datasets collected in their CRMs and loyalty databases, if this is done in a fully controlled manner, in accordance with the local privacy legislation and with an expressed consent from customers. This extra data, for example a known phone number of the customer driving past a brand store or digital billboard showing the brand’s campaign, may help to achieve another level of personalisation and thus enable the advertiser to better engage with their audience.
DOOH Publishers maintain their own private datasets about each screen and location in their network, along with their potential audiences. This is required to develop their pricing policies, work with media agencies and brands on tailored media plans, and is also used for programmatic trading. This data is often used to plan effective context-aware and personalised campaigns.
In some environments (transport, entertainment venues, cinemas, dine-in restaurants, clinics etc.) it is possible to accurately match the audience using entry tickets or bookings. It is the same with queue management systems in some banks and institutions. Although this may not provide a lot of information about visitors’ personal profiles (subject to privacy regulations), still their accurate numbers, the fact that they are at the premises at a certain time and the known purpose of their visit make it a powerful piece of data that can be used for targeting and personalisation. One challenge here is the limited applicability of this approach, only at some types of venues. Also, venues can be big enough to make it hard to patrons’ behaviour once they are inside, leave alone their interaction with a particular screen or content.
When customers use their loyalty cards to get a discount or a perk from their brand of choice, they willingly share their personal details, purchase history and location. This provides a strong opportunity for deep personalisation, although only for that small proportion of the audience using their loyalty cards. On the other hand, if an advertising campaign is targeted at the brand’s loyal customers, the number of loyalty cards presented at the store can be a good indicator of the campaign’s performance.
Completed transactions, such as payments for goods and services may serve as a good indicator of the number of visits into a store, however, this information is mostly limited to a location where the transaction has happened. On the other hand, the transactional data (indexed sales volume, average basket size, number of purchases in a category) can greatly help with the evaluation of the campaign’s performance.
In some cases, when automated collection of data is complicated or there are more sophisticated questions which cannot be directly addressed by such data, brands, publishers and agencies may decide to run a survey on a sample of the target audience, for example, to analyse how likely the screen or content is to be seen, or what exactly people liked or disliked about the ad. Although surveys may provide a lot of insights about the audience sentiment or behaviour, of course, they are very limited by their nature and can be efficient with relatively small samples only.
Mobile app/location data
One of the most popular methods of data collection for campaign creation, planning, buying and post-buy analysis assumes using mobile app data for the purpose of granular analysis of detailed demographic profiles in geofenced areas. Geofencing is a process of defining specific areas of interest on a geographical map to filter averaged “personas” reported by mobile app data providers along with their approximate locations, dwell time and travel routes. The popularity of this data source is due to the richness and coverage of the data it provides. One of its unique advantages compared to most other sources is that it can provide Unique Reach and Frequency (unique visitors and their repeat visits) data in a privacy-compliant way. However, the coverage and accuracy of this dataset have been significantly impacted over the last few years by the gradually tightening privacy regulations and privacy enforcement led by the leading mobile phone manufacturers. This trend is expected to continue into the future. Another potential drawback of this data source for some campaigns comes from its historical nature, making it hard to use it for real-time personalisation.
Telecom carrier data
Unlike mobile app data, the telecom carrier data is highly accurate and difficult to disrupt. It provides less information about people’s personal profiles, however, the collection and distribution of this data for audience measurement purposes is strictly regulated and may be illegal in some jurisdictions due to the fact that the providers may not have the legal or technical framework in place to guarantee complete privacy safety. Also, due to its very nature, this data only includes information about subscribers of a particular telecom service provider.
Passive WiFi or Bluetooth sensors
Another efficient way of collecting the Unique Reach and Frequency data is using the latest generation of mobile phone sensors. This technology allows to detect a unique signal pattern of each mobile phone and calculate the number of total, unique and repeat visitors with a relatively high accuracy, close to real time. Newer-generation sensors are able to address, to some extent, the challenges of MAC address scrambling and spoofing – measures introduced by mobile phone manufacturers in a bid to enforce users’ privacy. The technology provides ways to adjust the detection area and calibrate the output to account only for relevant traffic, but this data doesn’t provide data on demographic profiles and those who does not carry a mobile phone. On the other hand, it can be configured to analyse people’s journeys between sensors and thus enables a way to measure attribution, all in a privacy-compliant way.
Social WiFi/hotspot services
These days, most free WiFi service providers are able to collect rich information about their users – people who are willing to use their account details in one of common social networks to quickly get access to a free WiFi service. Providers are then able to compile, analyse and on-sell this data to advertisers. Despite the valuable information this method delivers, one of its major drawbacks is a relatively low number and proportion of visitors subscribing to it.
Campaign data from CMS
Any modern content management system provides information about played content, or Proof-Of-Play reports, historical or in real-time. Using this data along with any other data sources about the audience at each location provides a powerful way to analyse campaign performance at multiple levels – per play, per clip, per location and overall – for the campaign based on the audience data from each point in time and location.
The huge variety of data sources available for the planning, execution and performance evaluation of DOOH campaigns makes the data journeys of all industry players a challenge. Every company working in DOOH is faced with many choices about which data to use, how to integrate various data sources and how to effectively use it to their advantage and to meet the requirements of advertisers and other stakeholders.
Integration of multiple the data sources presents many opportunities. It is the integration that is able to address drawbacks of some solutions, provide more granularity, increase accuracy and introduce new metrics similar or better than what is available in other advertising channels. Proof, correlation, trends identification, root cause analysis, predictive analytics – all this is and can be enabled using existing and emerging ways to measure audiences in the physical world.
In the next article we will discuss the metrics and KPIs enabled by those data sources and their integration and correlation, metrics that are critically important for the brands on their DOOH and multi-channel advertising journey.