Blog Hashbrown

Performance Analytics in OOH - II

The science behind our methodology to measure performance impact of OOH campaigns.


Guru Cingh

The First Attempt

In order to conduct an exercise of OOH performance analytics, Hashbrown Systems selected three locations in Delhi, namely, SouthEx, Lajpat, and Aashram. These locations were selected because, (1) billboards were mounted at these locations, (2) these billboards were viewed by significant flow of traffic throughout the day and in the night, and (3) we had video footages available for these locations




Primary Research and Traffic Data Capture

We deployed video cameras to capture clear vision video footages of traffic in the selected locations. Footages were taken at 3 to 4 different time slots, such as, early morning, afternoon, evening, and late night, for all the 3 locations. 2-minute video footages were considered for analysis. Depending upon the traffic trends during different time slots in a day, duration (in hours) of heavy, medium, and low traffic volume are calculated and extrapolated to calculate total number of views in 24 hours.

Other Input Data

Once we gathered primary data, we formatted the same for consistency and quality. In our attempt to build a model to quantify the impact that makes comparison possible and leads to more effective and efficient business decision making, our expert data analysts conducted secondary research to gather & process other essential data, as listed below:

    ● RTO (Regional Transport Office) data for target districts for the last 10 years

    ● Cost of billboards

    ● Literacy rates for a region

    ● CPI (Consumer Price Index)

    ● Working / non-working percentage of population

    ● Percentage of viewers who would be interested in an Ad., as provided by Brand(s)

    ● Demography data of a region

    ● Percentage of interested demography, in an Ad., as provided by Brand(s)



Development of any analytics model requires one to make well-informed assumptions, some of which that we considered are listed below:

    ● Traffic is uniform on all weekdays and on weekends

    ● Traffic is uniform on all weeks

    ● Classification of traffic: Heavy, Medium, Low

   ● Percentage of Heavy traffic: From camera count, Medium: 75% of Percentage of Heavy traffic, Low: 30% of Percentage of Heavy traffic

    ● Buses in region are taken per 10 Sq.Km. area, calculated using the total number of buses in Delhi

    ● Every new vehicle on road in 10 years

   ● The % of small/medium and luxury cars is taken from the proportion of actual camera count estimate

    ● 10% of the illiterate population is impacted by the hoarding

    ● Working population on road is 90% and rest of the population is considered non-working

   ● There is a new vehicle on road every ten years, and therefore, total count of vehicles registered with the local RTO over a period of ten years is    considered for data analysis.

   ● Data is categorized into four vehicle types, namely, Two Wheelers, Cars (further categorized into Small, Medium, Luxury), Buses, and Others     (e- Rickshaw, Autorickshaw, Truck)

    ● The maximum possible change in purchasing power is 0.2, i.e. 20%

    ● Total heavy traffic hours in a day: 7 hours; Total medium traffic hours in a day: 8 hours; Total low traffic hours in a day: 9 hours

Data Analysis & Results

The first step in processing the data was to consider average of the camera count of traffic. This means that since the video footages of varying durations were captured at various time slots, it was necessary to standardize the data format. Therefore, 2-minute video footages were considered, and footages for total number of Heavy, Medium and Low traffic hours were extrapolated to account for the traffic flow for 24 hours. Subsequently, percentage of viewers during Heavy, Medium and Low traffic hours were calculated.

Our analysis was further modified by accounting for the data of ‘desired percentage’ and ‘interested percentage’ provided by Brands, and ultimately resulted in the number of views that a billboard Ad. gets, in a location, as given below:




Following the above exercise, a pure rating was calculated for number of views for a day per rupee spent, for an Ad. on a billboard. The rating was further improved to account for various parameters relevant to the Brand, some of which are – percentage of working/non-working population, percentage of interested demography in an Ad., and others.


The model, which we attempted to build, could be extended to any number of locations across the city. This model could be useful in measuring the actual impact of out-of-home advertisements as it provides an in-depth analysis of designated areas of a given city, where advertisers would be willing to invest.


More Hashbrown Stories

Hashbrown Systems Case Studies

Bubna Advertising

Our first case study briefly analyses the first outdoor monitoring and compliance system for the largest outdoor agency in India by volume.

Case study


Our OOH Audit & Monitoring System uses Machine Learning techniques and a uniquely crafted allocation model to optimize fund allocation for 88 billboard locations, a breakthrough in the Out-of-Home Advertising & Marketing industry.

Case study

Spotlight - Brand Sales & Distribution

An overview of digital transformation that employed cloud computing, data analytics, machine learning and location intelligence to create a constantly connected and data driven enterprise.

Case study

Building Digital Infrastructure for the Physical World

A triumphant tale of putting IOT to work for Out-of-home media owners and advertisers.

Case study