Media monitoring: how to track what is said about you

Media monitoring is the routine tracking of what newspapers, broadcasters, online news sites and social media say about your organisation, your leaders and the issues you work on. Press monitoring is the older, narrower form: collecting and reading print and broadcast coverage. Social listening goes further. In the context of public health, the World Health Organization describes it as the process of listening to and analysing conversations and narratives.¹

For NGOs, UN agencies, ministries and companies, the purpose is practical. You want to know whether your messages are reaching people, how they are being interpreted, which questions and concerns are circulating, and whether something needs a response before it grows.

The most common mistake is to start collecting clippings without knowing why. The Barcelona Principles 3.0, the measurement standard published by the International Association for Measurement and Evaluation of Communication (AMEC), open with the statement that setting goals is an absolute prerequisite to communications planning, measurement and evaluation.² The UK Government Communication Service (GCS) recommends planning how to evaluate a campaign at the start, and establishing a baseline measure of what would happen if no communication activity took place.³

In practice, write down two or three questions your monitoring must answer. For example: is our new programme being covered accurately? Are our spokespeople quoted? Which concerns are people raising about the service we provide? Record where things stand today, so that next quarter's figures mean something.

The Barcelona Principles state that holistic measurement and evaluation includes all relevant online and offline channels.² For many audiences in East Africa, that means radio and television as well as newspapers, news websites and social platforms. WHO and UNICEF guidance on infodemic insights also draws on both online and offline data sources, and stresses that because not everyone is online, community-level data matters.⁴

Then build a clear list of search terms. A monitoring service or tool is only as good as the words it is asked to find.

• Your organisation's full name, short name, acronym and common misspellings
• Names of your leaders and official spokespeople
• Programme, campaign and product names, and their hashtags
• The issues you work on, in every language your audiences use
• Partners, peer organisations and the main policy debates in your sector

A count of articles tells you little on its own. The GCS Evaluation Framework 2.0 suggests breaking coverage down by national, regional, trade and broadcast media, measuring key message penetration and the percentage of coverage that includes a ministerial or spokesperson quote.⁵ These measures show whether coverage is carrying your message, not just mentioning your name.

The Barcelona Principles also say that measurement should include both qualitative and quantitative analysis, and state plainly that AVEs, or advertising value equivalents, are not the value of communication.² If a report converts column centimetres into an advertising price, it is measuring space, not effect.

Treat automated sentiment scores with care. The GCS framework considers sentiment valid for manual assessment of press coverage or a defined number of stakeholders, but not for a mass automated approach.⁵ A trained reader checking a sample will usually give you a more reliable picture.

Social listening is most valuable when it reveals what people are asking, misunderstanding or worried about. WHO and UNICEF set out six steps to build an infodemic insights report: choosing the question, selecting data sources and an analysis plan for each, conducting an integrated analysis, developing strategies and recommendations, writing the report, and disseminating it while tracking the actions taken.⁴

When listening picks up rumours or false information, UNICEF guidance for Eastern and Southern Africa describes three types of strategic action: preventive strategies, early detection and associated indirect actions, and direct engagement.⁶ Its stated aim is to build trust with the communities served.⁶ Listening to the public also carries responsibilities. WHO's guidance on social listening frames the work around the protection of and respect for human rights.¹ Analyse conversations and trends, not individuals, and do not collect more personal information than the purpose requires.

Most teams need three rhythms: an immediate alert when something significant appears, a short weekly summary and a monthly or quarterly analysis linked to objectives. The GCS distinguishes outputs, such as reach and press coverage, from outtakes, which capture what the audience thinks, feels or intends to do, and outcomes, which are changes in behaviour or active engagement.³ A good monitoring report says which level each finding belongs to.

WHO and UNICEF insights reports describe the narratives in circulation, how risky each one is and the recommended actions.⁴ That structure works well beyond health. End every report with what you suggest doing, who should do it and by when, and check at the next report whether it happened.

Software can collect online news and social media posts quickly, but radio, local-language programmes and community feedback often still need people to listen, note and summarise. Whatever mix you choose, document your search terms, sources and definitions so that figures can be compared over time.

In our own practice, we find that the most useful monitoring is modest in scope and read by decision-makers. A one-page weekly summary with three clear observations and one recommendation is more likely to change what an organisation does than a long folder of clippings that nobody opens.

• Two or three questions the monitoring must answer
• A baseline recorded before activity starts
• Online and offline sources, including radio and local languages
• A documented list of names, terms and hashtags
• Coverage tracked for key messages and spokesperson quotes
• No advertising value equivalents in reports
• Sentiment checked by a person on a sample
• Every report ends with recommendations and follow-up