SCMP Signal, the First Publisher-Built Brand Suitability Tool
2021 Finalist

SCMP Signal, the First Publisher-Built Brand Suitability Tool

South China Morning Post

Hong Kong, Hong Kong

Category Data and Research

Media associated with this campaign

Overview of this campaign

Objective 1: Keep brands safe whilst on SCMP across all platforms from the first impression. 

  • Current technology only starts to analyse content after a critical mass is reached, leaving brands exposed to unsafe content and most only work on Desktop and mobile content. 

  • These tools only use negative keyword blocking (exclusion lists). These become generic lists and are unable to determine the difference between a ‘photoshoot’ and a shooting. 

  • Signal was designed to work the moment the article is published and works across all platforms by using NLP to read Keywords, context and sentiment to determine the true nature of the article. 

Objective 2: Drive client’s performance & reach during a campaign.

  • Exclusion lists limit a client’s access to key audiences by up to 40% as reported by Trinity mirror using IBM watson - this affects campaign optimisation.

  • We aim to extend the awareness  of the average campaign by at least 70%  and increase the average CTR by 40% 

Objective 3: Understand more about our audiences.

  • To collect data and add it into our central customer data platform (CDP) 1PlusX and append it onto their unique ID. 

  • This data could include: 

  • Behavioral content consumption and intent

  • Observed sentiment preference 

  • Reading preference by article reading complexity

Objective 4: Make the data easily available for usage.

  • The tool is built into our CMS so editors can use it to understand if the sentiment, reading level and more are what they expected. 

  • So that the article is scored for brand safety/suitability as soon as it is published and that value is immediately passed to the Ad server.  

Objective 5: Drive additional revenue for SCMP.

  • We want to see clients who are sensitive to brand safety issues start to spend with SCMP again and in a broader range of content verticals. 

Results for this campaign

The build began in February 2020 and took four months to build. We launched to market in Signal in May 2020.

This build was a cross departmental effort including Data, Product, Technology and the Commercial team. 

Three components were necessary 1) Sentiment analysis 2) Sensitive keywords tagging and 3) article readability analysis. Working together to provide a complex  profile of the written articles. We employed all of these innovations, plus the use of a lexicon-analysis tool, called Valence Aware Dictionary for Sentiment Reasoning (VADER). 

The chart below represents the reading level difficulty of the articles SCMP publishes every day:

  • Green - 5th -7th grade - Romance novels

  • Blue - 8th - 12th grade (Secondary School) - New York Times

  • Orange - College - Harvard Law Review

  • Red - College Graduate - Scientific Journals

As you can see below, over 50% of our content is written at a college level." alt="" width="600" height="248" />

The y-axis represents the % of our articles each day published at various readability levels.

SCMP Signal is now used to ensure the safety of 100% of all direct booked deals.  

Since launching we have successfully run SCMP Signal on more than 450 campaigns and been able to build the trust of several new sensitive brands including the likes of Cartier, HSBC etc.  to spend with us with data led campaigns.  

Providing best in class brand safety tools to our clients:" alt="" width="249" height="172" />


Example Campaign: 

Campaign: Content Banners_ 2020

Client: SCMP Group Marketing & Events

Dates: 10th July 2020 - December 2020

The Results: 

55% increase in CTR & a 54% overall campaign uplift. 


To contact a company representative about this campaign, click here for the INMA Member Directory

By continuing to browse or by clicking “ACCEPT,” you agree to the storing of cookies on your device to enhance your site experience. To learn more about how we use cookies, please see our privacy policy.