Changes to calculations and modelling

The April 2020 release (R34) incorporates a range of significant developments to the methodology. These affect the definitions and calculations made and, in turn, the audience numbers produced. Here we summarise the changes applied to the data models and algorithms – namely the introduction of spot ratings, application of seasonal impacts and the evolution of the 14-day algorithm.

Digital Spots

Route presents a world-first in its measurement of digital out of home advertising. This comes with the introduction of a new methodology enabling analysis of digital spots. (More info on the effect of spots can be found here).

Key components in the new measure are:

1. Analysis of individual digital spots. The system previously apportioned reach and audience for digital frames (and mechanical scrollers) proportionate to the share of voice/loop. The new system is more sophisticated. It now takes into account the length of time people are exposed to screens and the specific spot schedule. Being armed with the spot duration (in seconds) and also the time between playouts on screen, we can calculate a more precise audience estimate.  The audience of a spot is derived from the expected overlap and frequency of contact with the spot schedule, standardised within 15 minute intervals.

This means we now differentiate between campaigns of different spot durations that run on the same share of voice. A 5 second spot appearing every 30 seconds and a 10 second ad broadcast every 60 seconds will have different audiences. We provide audience measures for any spot duration from 1 second to 1 week

With more detailed data available to make better calculations, it’s now possible to account for the likelihood of people seeing the same campaign multiple times in a single exposure. For instance, if we know a person has been exposed to a screen for 60 seconds and that the schedule is showing 5 second spots repeating every 30 seconds, we can establish that people are able to see the campaign more than once per exposure to the screen. This was not possible in the past.

2. Introduction of a new 14-day benchmark reach algorithm. With a full complement of participants’ travel data collected over 14 days, we can move to a new benchmark period. Previously, benchmark reach (the campaign reach ‘jumping off’ period that is used to project longer term campaigns) was based on 9 days. We now have our full sample with two full weeks of travel data on which to base our benchmarks. This gives additional reliability to the data and better confidence in projecting to longer term reach builds.

3. Non-linear visibility curves have been introduced by environment, replacing the pre-existing linear curves. The new curves better reflect behaviour observed in the Travel Survey.

4. The time frame of a Realistic Opportunity To See (ROTS) has changed from 300 seconds to 6 seconds. This means that there can be a new impact every 6 seconds rather than every 300 seconds as currently, though visibility will be lower for those impacts.  The change was made in order to identify variations in behaviour and types of contact that the 300 second interval was too blunt to capture.  Statistical analysis identified 6 seconds as the best option to capture these variations without generating an unnecessarily large number of contacts.  The change in definition is significant for ROUTE’s outputs, usually increasing impacts, though with considerable variation by environment, frame type and size.  Shopping environments have seen the most marked increase in impacts as a result.

5. Seasonality has been introduced, applying a seasonal factor by month to impacts, reach is unchanged by month. Seasonal factors are applied depending on the underlying data. In some instances, seasonality is fixed at environment level (e.g. motorway service areas, outdoor shopping centres). In others there is variance in factors being applied at point of interest level (POI) – such as stations or malls. More information on what the seasonal data looks like can be seen here.

6. Ipsos have developed the ‘Contact Redistribution’ statistical model. It is used to expand the Travel Survey sample to support more granular analysis at frame level. It increases the number of contacts available in the sample for analysis purposes. The expansion is done by redistributing contacts from frames which have them to frames that don’t, using a detailed set of rules that take into account frame characteristics, audience profiles geographical proximity, etc

7. Introduction of MST smoothing to represent true travel habits. New multi-sensor-tracking meters record locations, no matter where participants go. Data from the new devices is more precise than ever – accurate to one metre. They take readings every second and uniquely, they work indoors and out. We’re now able to track people for longer and further each day. Outputs from the sensors result in smooth journeys that overcome noise in the GPS readings. This means we have a better reflection of real world journeys. In some instances it has led to longer exposure to OOH ads as better positioning data gives an improved understanding of visibility levels. Overall it produces a more realistic and reflective pattern of real-world journeys and encounters with OOH ads

Leave a Reply

Your email address will not be published. Required fields are marked *