# Media content analysis: definition, criteria and example

What media content analysis is, which quantitative and qualitative criteria it uses, how a codebook works, how it fits AMEC and where its limits lie.

**Media content analysis is the systematic, usually quantitative and qualitative evaluation of coverage about an organisation, brand or topic. Using fixed rules, it shows how much was published, where, in what tone and with which messages.**

A folder of clippings does not answer a question on its own. Media content analysis turns it into statements such as "we were quoted more often than our competitors in top-tier outlets, but our core message rarely came through." In German-speaking markets the same work is called _Medienresonanzanalyse_, which is often rendered in English as media response analysis. The shorter term media analysis is also common. It always builds on [media monitoring](/en/glossary/media-monitoring), which supplies the coverage.

## Quantitative and qualitative criteria

Media content analysis combines two perspectives. Quantitative criteria count and measure:

- **Volume:** the number of relevant articles or mentions in the period.
- **Reach:** estimated opportunities to see, usually derived from circulation, visits, page views or broadcast audience figures. These are proxies with different units: circulation counts copies, visits and page views do not count unique people, and traffic for a whole outlet is not the readership of one article. State the unit, source and period. Adding up reach across articles may count the same people several times.
- **Share of voice:** the brand's share of coverage within a defined competitive set. See [share of voice in PR](/en/blog/share-of-voice-pr).

Qualitative criteria assess the content:

- **Tone:** positive, neutral or negative towards the brand. See [PR sentiment analysis](/en/blog/pr-sentiment-analysis).
- **Key messages:** whether and how fully your messages appear, often called [key message pull-through](/en/blog/key-message-pull-through).
- **Prominence:** whether the brand appears in the headline, the opening paragraph or only in passing.
- **Outlet relevance:** whether the article ran in a defined target outlet.

The Institute for Public Relations lists similar quality measures in its [proposed standards for traditional media analysis](https://instituteforpr.org/wp-content/uploads/Traditional-media-standards-6-12-12-v-2.pdf), including placement, prominence, message penetration, spokesperson quotes and whether an organisation is the dominant subject of an item.

## The codebook

To keep results comparable, a codebook fixes before the analysis starts how each item is coded. It defines, among other things:

- which items count as relevant and which are excluded,
- what makes tone positive, neutral or negative,
- how a key message is recognised and whether paraphrases count,
- which outlets are target outlets,
- how duplicates and syndicated copies count: if the same agency story runs in 30 outlets, count each publication, but flag it as syndicated and report the number of unique stories separately,
- which counting unit applies, the article or the brand mention, when one article names several brands. For share of voice, fix the denominator. In the example below it is the sum of articles for all brands in the competitive set, so an article naming two brands counts for both.

The IPR recommends documenting search terms, reach calculations, sentiment rules and target media in writing, and using the same coders and method throughout an ongoing project.

To make different coders reach the same judgements, train them on the codebook and calibrate their coding on sample items. Then check how often they agree on a sample, and tighten the rules where they diverge. If an AI model does some of the coding, record the model and version and check a sample by hand. When the method changes, for example through new rules, different coders or another model, recalibrate before you compare periods.

## How it fits evaluation frameworks

Media content analysis is one part of PR evaluation, not the whole of it. The [AMEC Integrated Evaluation Framework](https://amecorg.com/amecframework/) separates inputs, activities, outputs, out-takes, outcomes and impact. The [AMEC taxonomy](https://amecorg.com/amecframework/home/supporting-material/taxonomy/) places quantitative and qualitative media content analysis, along with metrics such as share of voice, tone and messages placed, at the output stage.

The German [DPRG/ICV framework for communication controlling](http://www.communicationcontrolling.de/indexc777.html?id=280&L=3), agreed in 2009 by the German Public Relations Association (DPRG) and the International Controller Association (ICV), follows the same logic with its levels of impact from input and output to outcome and outflow. Its [detailed description of the levels](http://www.communicationcontrolling.de/indexef73.html?id=33) (in German) lists media clippings and share of voice as external output: necessary conditions for communication, but not yet evidence that it worked.

The [Barcelona Principles 4.0](https://amecorg.com/wp-content/uploads/2025/06/AMEC-Barcelona-Principles-4.0-Launch-Presentation-FinalV.pdf) call for combining quantitative and qualitative methods, reporting on outputs, outcomes and impact, and not using advertising value equivalents (AVEs).

## Example: a quarterly analysis

The figures below are a fictional illustration. A company analyses one quarter of coverage against two competitors.

| Metric                          | Own brand    | Competitor A | Competitor B |
| ------------------------------- | ------------ | ------------ | ------------ |
| Relevant articles               | 120          | 90           | 60           |
| Share of voice (by articles)    | 44%          | 33%          | 22%          |
| In target outlets               | 36 (30%)     | 45 (50%)     | 12 (20%)     |
| Brand in headline or first para | 48 (40%)     | 27 (30%)     | 18 (30%)     |
| Positive / neutral / negative   | 30 / 78 / 12 | 18 / 63 / 9  | 9 / 45 / 6   |
| At least one key message        | 42 (35%)     | not coded    | not coded    |

44% + 33% + 22% adds up to 99% because the shares are rounded.

How to read it: the brand leads on volume, but competitor A is far stronger in target outlets. The core message appears in just over a third of articles. That points to a concrete objective, such as more presence in target outlets rather than more articles overall. For more metrics, see [PR KPIs 2026](/en/blog/pr-kpis-2026).

## Limits of the method

Media content analysis describes what was published. It does not measure whether anyone read the article, understood it or changed their mind. Reach is an estimate of opportunities to see, not proven attention. To show effect, you need additional data, such as surveys for attitudes and website analytics or enquiries for behaviour, and you need to connect the results to your communication objectives.

## How it differs from a press review

A [press review](/en/glossary/press-review) mainly documents coverage. It can include editorial summaries, but it does not code articles systematically. Media content analysis codes the same articles against fixed criteria and condenses them into metrics and conclusions. The press review answers "what was reported?"; the analysis answers "what does this mean for our objectives?"

## How aclipp supports media content analysis

In aclipp, clippings from [media monitoring](/en/media-monitoring) land directly in the workspace. AI suggests tone and relevance, which you can review and correct. [Reporting](/en/reporting) calculates metrics such as volume, reach and share of voice from the same data, so you can build dashboards and reports without exporting spreadsheets. How AI assistants describe a brand can be analysed alongside through [AI visibility](/en/ai-visibility).