Collaborative tool launched for advertising and media decision-makers

The Publisher Research Council (PRC) and Nielsen have taken a giant leap forward in a collaboration with the Broadcast Research Council (BRC). They have integrated PAMS data with the BRC TAMS and Nielsen’s CPS and DCS (Digital Consumer Survey) datasets.

Each of these surveys is powerful in its own right

Josephine Buys, PRC CEO

Josephine Buys, PRC CEO, at the launch, reminded attendees of the robust PAMS sample (15,250 adults) drawn from all 233 municipalities across the country. Income was used in the sample design for the first time in SA media history: the Neighbourhood Lifestyle Index (NLI), developed by GTI, was used to disproportionately oversample upper income areas. Fieldwork was carried out from August to November 2019 prior to the pandemic.

Broadcast Research Council CEO, Gary Whitaker, stressed the solid nature of the TAMS panel comprising 3,134 households, representing the national TV viewing universe of individuals (4+). The data used in the fusion was from a 12-month pre-Covid period (to February 2020).

Terry Murphy, MD of Nielsen Media South Africa introduced the Nielsen databases. The CPS panel comprises 4,000 households, demographically and geographically representative of all South African households. It provides actual audited purchase data of FMCG goods across all retail outlets. Because the fieldwork ran from March 2019 to February 2020, the data was not affected by supply chain disruptions and atypical lockdown purchasing decisions.

The DCS uses the PAMS online universe as the establishment survey for its panel of 3,000 respondents nationally, representing adults (15+) who accessed the internet in the past week. The fieldwork took place from July to August 2020, once the initial pandemic media consumption surge had settled. The DCS provides a wide range of information about online behaviour across all platforms

Fusion provides the solution in a fragmented media world

Jonathon Wells, senior vice president of data science at Nielsen Media International, joined the webinar from Tampa, Florida, to explain the fundamentals of fusion. The number of paths consumers can take to consume media and make purchases have expanded exponentially. Accurately recording these multiple varying permutations and combinations places a greater burden on the respondent in terms of time and recall.

In this information age, there are a multitude of different data sources each capturing a different piece of the puzzle. Fusion builds a more complete view of the consumer’s habits by capturing all the data of interest in separate detailed surveys that are then combined to provide far greater insights than a classical single-source survey. This fused dataset is four times bigger than AMPS 2015 was.

The process of fusion

Wells gave a simple example of fusion: TAMS shows who watches a particular game, whilst CPS shows who buys a particular cold drink brand. Common demographic characteristics shared by respondents in both surveys can be used to match profiles across the surveys. However, fusions are not built by finding the same person across two data sets.

The underpinning assumption is that human behaviour is predictable: people with similar characteristics will behave in similar ways. Linking variables are the basis for deciding which respondents in the donor survey are the closest in profile to those in the recipient survey. The linking or matching variables (or hooks) must be good predictors of behaviour across both data sets and need to be well-matched. Variable selection begins with a simple analysis of sample sizes and profiles across both the databases to find commonalities. These are then assessed whether they are statistically significant.

PAMS is the hub

PAMS was selected as the recipient survey because of the robust sample and the comprehensive set of variables that can be used across the databases. TAMS was fused to PAMS using hooks such as the type of pay TV subscription and demographics.

Because they are currency measurements, it is essential that the data from each be preserved by matching every TAMS donor to a PAMS recipient. If there are not enough best matches amongst the PAMS recipients, fusion allows the fragmentation of some to allow for a best match for every donor. Importantly, whilst the number of respondents increases, the sum of weights is preserved to ensure that PAMS currency does not change.

Source: pixabay.com

Advantages of fusion

Wells summarised the advantages of fusion as being cost-effective and convenient, fitting easily into existing software programmes. It allows the harnessing of existing currency data with respondent level flexibility. It is a statistically precise solution, with robust quality checks and validations undertaken. By providing a single data set, fusion makes the planning process more efficient.

Fusion in action

Wells compared the build of incremental reach against purchasers of Lunch Bar, Kit Kat, Smarties and Lindt Lindor: not surprisingly social media delivered 2.5 times the incremental reach against the upmarket imported chocolate purchaser, than it did against the mid-market countline buyers.

He showed that DStv delivers superior reach against Lunch Bar purchasers between 15h00-18h00 and demonstrated a “360” view of a new car purchaser.

Peter Langschmidt, lead research consultant at PRC, demonstrated the level of granularity available in The Neighbourhood Lifestyle Index (NLI): adjacent Bloubosrand sub places within Randburg – rated NLI 1 and NLI 7.

Comparing Capitec and FNB customer profiles, he showed that the data had application for brand and media planning. Activity by platform by daypart in 2019 and 2020, revealed how the pandemic has impacted media habits from DCS data.

Making it easier for users

Every effort has been made to save users time and effort e.g. global cross-tabs and codes are being set up and will be available to all users in areas like target marketing, candidate media or a “day in the life”.

Ongoing developments

Langschmidt announced that the Mindsets & Attitudes (MAMS) study, covering values, motivations, aspirations values and much more was conducted in October and will be released to the industry in late November. The project to fuse Narratiive data by daypart should be completed by Q2 2021.