Data Driven Out of Home(OOH) Media Advertising

About the industry

The out of home (OOH) media advertising industry has been around for centuries. It provides brands with an opportunity to connect with consumers in the physical world through various channels that include billboards, street furniture, transit advertising and digital billboards. And with the advent of technology, the industry has undergone a significant transformation in recent years.

One of the main advantages of OOH advertising is its ability to reach a mass audience. For example, billboards on a busy highway can reach thousands of commuters every day, providing brands with a great opportunity to reach a large audience in a specific area.

Another advantage of OOH advertising is its ability to create a strong emotional connection with consumers. The campaigns can be designed to be visually striking, with large and colourful displays that can grab the attention of consumers. This is especially true for digital billboards, which can display videos, animations and interactive content to create a memorable experience for the viewers.

OOH advertising also provides brands with an opportunity to reach consumers in the "real world" and in real-time. Unlike online advertising, which can be easily ignored or blocked, OOH advertising is difficult to ignore and it reaches consumers when they are most likely to be engaged and receptive to marketing messages.

?

Data Sources and metrics

As can be imagined, one of the key challenges with OOH is being able to track the number of views for its assets. But before we talk about this, we need to understand some industry standard metrics/keywords.

One of the key metrics used in the OOH advertising industry is reach. Reach refers to the number of people who have been exposed to an advertising campaign. Another important metric is frequency which refers to the number of times a person is exposed to an advertising campaign.

There can be different reasons for advertisers to focus on either of these metrics. The philosophy behind them is really quite straightforward. Reach depicts how many unique people have seen the advertisement. However, in our uber busy lives, it is very easy for us to forget stuff we see only once. This is where frequency comes in. To make sure that the brand recall aspect of the advertisement is at an optimum level, frequency signifies how many times we expect each individual to have seen that particular advertisement.

?

There are a lot of third party data sources that the OOH industry relies on for coming up with these metrics. Here we will talk about 2 such data sources and how they are leveraged in the OOH industry to come up with targeted advertisements.

Some of the most popular third party datasets that are leveraged for coming up with valuable insights are based on banking and telecom data. By further partnering with retail stores, there are further segments that are created to ensure that the advertisements are displayed at relevant locations.

Based on these datasets, the best results are obtained by using both the datasets for their strengths. While the banking based data has the segmentation information based on share of wallet spend analysis, where it is found lacking is in its recency. However, the strength of the telecom data is usually in its recency and (anonymised) tracking accuracy. The best way in this case is to combine both the results from both of these datasets using techniques like conflation.

?

Optimization challenges and Problems

At a very broad level, clients for OOH assets make 2 kinds of request- location based or fixed metric based advertisement.

The location based requests are usually limited to clients who have physical stores and need to advertise for their product/offerings in the vicinity of the specified locations. For instance, if we are talking about KFC, they might want to advertise at 100 billboards close to all of their stores in Sydney. The key challenge here is the overlap factor. In this example, if there are 500 KFC stores, how do we select the 100 panels from the 1000 available panels within a 2 km radius of all of these stores. We need to ensure that not only we maximise the total reach but also achieve a frequency factor which is at an optimum level.

Fixed metric based advertisements are an altogether different challenge. In this case, the clients request for maximising their reach, frequency, number of panels etc. A lot of times, there are clear budget restrictions in place as well. In this scenario it becomes more important to leverage the reach and frequency data to be able to come up with the best selection of billboards.

While making the selection of the billboards, there are two kinds of effects that we need to keep in mind. The first effect is what we call ‘incremental’ effect. This is the phenomenon where we see a decline of the reach as we keep the advertisement going in a particular location. To illustrate this further, imagine a billboard X on your way to the workplace. The first day you see the advertisement, the reach increases by 1. However, because of the way the reach is defined, any more views are not going to increase the reach. So theoretically, it is in the best interest of the advertiser to advertise at a different (albeit similar in terms of footfall) location on the next day. However, due to the nature of the business and the costs associated with putting up an advertisement, there is a minimum period for which the booking can be made. As such, when we calculate the reach, we ensure that we keep this effect in mind.

?

The second one is what we call the ‘interactive’ effect. If a client buys 2 advertisements right next to each other, we expect the same people to see both the ads. As such, even though on an individual level, the panels might have a good reach, it might be counterproductive since the total reach for both the panels combined will not be cumulative but rather a function of other factors like how close the panels are to each other or what direction they are facing etc.

Most of the third parties I have highlighted above provide their own proprietary api that we can use to get the total reach based on a given panel combination. These third parties have various studies and models that they use to come up with these numbers. For instance, every year they take a small number of carefully curated collections of shopping centres. They do a study to come up with the number of views and use this data to extrapolate the same for similar shopping centres (Note: It is expected that over time various IoT devices might be deployed at various locations to make this process a lot more streamlined). However, there are multiple concerns when using these APIs. The first one is the time factor. If we have a collection of 1000 panels and we wish to select 100 from it, we will have to go through all the 1000C100 combinations and run them through the api to see which combination came up with the best reach or frequency. Also, using these api comes with a cost. Selecting these panels is a very iterative process with a lot of cancellations, swaps etc.

Another major challenge is the optimization process itself. If we continue with our 1000C100 assumption, our optimization will select the best 100 in the first stage. However, it is imperative that for the second client who is looking for the next 100 sets of panels, the remaining selection will not be as lucrative as it needs to be. While it might seem like the first come, first serve rule is applicable here- in this case, it is very highly likely to turn off prospective clients which is a major drawback of this technique. In this case, there needs to be a fine toothed balance between the selection that is done now while keeping the options open for any requests that are likely to arrive in the future.

?

Food for thought

There are various ways these issues can be overcome. For instance, instead of selling for panels, we sell for reach. What that means is that we promise to sell a specific number of reach to our prospective clients when we take the bookings. This way any adjustment that needs to happen remains a backend process.

However, this will come with its own set of unique challenges. Moving from selling panels to selling reach is a very unique shift in the way the business has been done traditionally. Also, this approach is very heavily reliant on a transparent communication of the campaign performance to inspire confidence in the clients that they are indeed getting the results that they paid for. It will need a very tech heavy paradigm shift in the way that the OOH industry operates.

With the increasing competition from online advertising and other forms of digital marketing, the OOH is geared towards overcoming these challenges by increasingly leveraging the power of data to improve targeting and personalization of advertising campaigns.