# Sources, Not Keywords: Source-First vs. Media Monitoring

> Keyword alerts show where a word appears. Source-First shows what a market player says itself. How it compares with media monitoring and social listening.

- Author: Dr. Karsten Richter
- Publisher: Picasi GmbH
- Topic: Approaches compared
- Last updated: 2026-09-24
- Canonical URL: https://source-first-intelligence.com/en/sources-vs-keywords/
- Language: en


Keyword monitoring answers the question of where a word appears. Source-First Intelligence answers the question of what a specific market player says itself. The two questions sound alike, but they call for different tools. People who want to understand a market and its competitors often need the answer to the second question. The common tools are built for the first and do that job well. The second needs a different starting point.

This page compares [three common practices of the Topic-First logic](https://source-first-intelligence.com/en/topic-first/) (keyword alerts such as Google Alerts, media monitoring and social listening) with the Source-First approach. It sets out what each one delivers and where each one reaches its limits. It also takes the objections seriously, including the strongest one: a list of sources only contains the sources you already know.

## Two questions that sound the same

A hypothetical example: a maker of industrial sensors wants to know what its largest competitor is planning. If it sets up a keyword alert on the competitor’s name, it gets places where that name appears. Those might be press articles, job boards, reseller listings, a forum thread, or a different company with a similar name. It may also get the LinkedIn post in which the competitor’s CEO announces a new product line. But only if the post contains the keyword and the search engine lists it as a result.

“What is this competitor saying?” starts from a sender. “Where does this name show up?” starts from a word. That is not a matter of settings but of order. In the first case you decide whom to listen to, then read everything that source publishes. In the second you decide which words to search for, and the sender is something you learn along the way.

## How three common Topic-First practices work

### Keyword alerts: the word decides

Google describes its alert service in one line: you can get emails [“when new results for a topic show up in Google Search”](https://support.google.com/websearch/answer/4815696?hl=en), for example news, products or mentions of your name. The topic is a search term. The Google News Initiative’s training for journalists says the same thing: you start by [“entering the search terms you want to get email notifications about”](https://newsinitiative.withgoogle.com/resources/trainings/google-alerts-stay-in-the-know/), and creating alerts on relevant keywords keeps a reporter up to date on a beat.

The options are revealing. The [Google Alerts interface](https://www.google.com/alerts?hl=en) asks how often (“As-it-happens”, “At most once a day”, “At most once a week”), which language, which region and how many results (“Only the best results” or “All results”). Under “Sources” it lists Automatic, News, Blogs, Web, Video, Books, Discussions and Finance. Those are types of channel, not senders. The word “source” means something different there than on this site, where a source is always the sender: a company, an association, a person. The list has no setting for the sender. Google’s own training recommends the Automatic option, [“which provides you with the best results from multiple sources”](https://newsinitiative.withgoogle.com/resources/trainings/google-alerts-stay-in-the-know/).

That is what the service is designed to do. Keyword alerts exist to report new hits for a term. Who used the term plays no part in whether you are notified.

### Media monitoring: coverage about someone

Media monitoring tracks what the press, broadcasters and online media report about a company or a topic. In practice this typically means collecting individual items of coverage and then counting and analysing them. The measurement association AMEC calls this kind of coverage [“earned media”](https://amecorg.com/glossary/): “third-party media coverage secured through a relationship or news worthy event, rather than paid-for advertising”.

The focus is clear. Media monitoring looks at what editors and journalists write about someone. What that someone publishes on their own channels appears only once a newsroom picks it up.

### Social listening: other people’s conversations about a brand

Social listening analyses what people say on social networks and across the web about a brand, a product or a topic, often with sentiment analysis. AMEC defines the related field of “social analytics” as identifying, tracking and listening to [“the distributed conversations about a particular brand, product or issue”](https://amecorg.com/glossary/), with emphasis on sentiment and influence. In communication research, Stewart and Arnold describe social listening as an emerging type of listening and [“a means of attaining interpersonal information and social intelligence”](https://scholars.unf.edu/en/publications/defining-social-listening-recognizing-an-emerging-dimension-of-li/).

Social listening listens to the conversation about something. For understanding how customers see a brand, that is exactly the right design.

## Not what they are built for: other organisations’ own channels

AMEC also defines [“owned media”](https://amecorg.com/glossary/): channels owned by or in the control of an organisation, typically websites, company blogs, newsletters and brand accounts on social media. The association has your own communication in mind, the part you want to measure.

Turn that around and you get the argument of this page. The three common tools watch either words on the open web or what third parties say about someone. They pick up other organisations’ owned media (a competitor’s newsletter, the openings on its careers page, its CEO’s LinkedIn posts, a trade association’s webinar) only when it contains a search term, when an editor reports on it, or when a tool is set up to follow specific accounts. Source-First Intelligence makes exactly those channels the object of attention. You decide which sources matter and follow what they publish on their own channels. That is what the [Source-First Manifesto](https://source-first-intelligence.com/en/manifest/) means by “Track who, not what.”

## Why keywords are noisy

Keyword searches return irrelevant hits whatever the product, because of how language works. Information retrieval research, the study of searching large collections of text, describes the mechanics precisely. Manning, Raghavan and Schütze’s textbook (Cambridge University Press, 2008) sets them out.

**One word, several meanings.** Polysemy is the case where a term such as “charge” [has multiple meanings](https://nlp.stanford.edu/IR-book/html/htmledition/latent-semantic-indexing-1.html), so that a word-based system overestimates how similar a document is to what the user wanted.

**One thing, several words.** Synonymy means [the same concept may be referred to using different words](https://nlp.stanford.edu/IR-book/html/htmledition/relevance-feedback-and-query-expansion-1.html), and according to the textbook it affects the recall of most retrieval systems, that is, the share of relevant documents that are found at all. If a competitor announces a product without using your keyword, the alert stays silent.

**Precision or recall.** Precision measures how many hits are relevant; recall measures how many relevant documents are found. The two trade off: [“you can always get a recall of 1 (but very low precision) by retrieving all documents for all queries”](https://nlp.stanford.edu/IR-book/html/htmledition/evaluation-of-unranked-retrieval-sets-1.html). For Boolean search the authors note that AND tends to give high precision and low recall, OR the reverse, and [“it is difficult or impossible to find a satisfactory middle ground”](https://nlp.stanford.edu/IR-book/html/htmledition/the-extended-boolean-model-versus-ranked-retrieval-1.html).

**A query is not an information need.** The textbook separates the two explicitly. A user interested in a topic wants relevant documents [“regardless of whether they precisely use those words”](https://nlp.stanford.edu/IR-book/html/htmledition/an-example-information-retrieval-problem-1.html). The information need “what is competitor X planning?” is hard to express as keywords. As a list of senders, it is easy.

Source-First does not fix these problems with a better search. It moves the filter. Instead of picking the right documents out of millions after publication, you decide beforehand whose publications you read. The sender becomes the first precision step, and synonymy does not matter for that step: the product announcement appears in the competitor’s newsletter whatever words it uses.

## More hits do not mean more attention

The obvious reply is to read more. Research on information overload, a subject studied for decades, suggests that does not work. In 2004 Eppler and Mengis reviewed the literature on the concept from organisation science, marketing, accounting and information systems, covering [definitions, causes, effects and countermeasures](https://www.alexandria.unisg.ch/54792) from the previous 30 years.

A finding from an unrelated field shows how it plays out day to day. In a study published in 2017, based on 112 primary care clinicians, on average [“clinicians became less likely to accept alerts as they received more of them, particularly more repeated alerts”](https://pmc.ncbi.nlm.nih.gov/articles/PMC5387195/). A quarter of drug alerts and a third of clinical reminders were repeats. This is an analogy, not a measurement of market monitoring. But it describes a mechanism that any team with a folder of unread alerts will recognise.

## Comparison at a glance

The table sums up the differences. The cells on noise and blind spots are assessments based on the evidence above, not measured values.

| | Keyword alerts (e.g. Google Alerts) | Media monitoring | Social listening | Source-First Intelligence |
|---|---|---|---|---|
| **Starting point** | A search term | Topics, names or search terms in editorial media | A brand, product or topic that people talk about | A list of sources: whose publications you follow |
| **What you get** | New results containing the word, whoever wrote them | Clippings and analysis of coverage (earned media) | Sentiment, influence and the course of third-party conversations | The content specific market players publish on their own channels (their owned media) |
| **Noise** | High for ambiguous terms and broad OR queries | Medium, pre-sorted by media selection and coding | High; on top of that, sheer data volume creates work of its own | Low, because unknown senders drop out; sources still publish things that do not matter |
| **Blind spots** | Anything said without the keyword; the sender is incidental | What organisations publish themselves unless the media report it | Organisations’ professional communication that is not a conversation about a brand | Sources you do not know yet |
| **Typical use** | Keeping an eye on a topic or term; finding something new | PR evaluation, press clippings, issues tracking | Brand perception, campaign impact, crisis communication | Competitor and market monitoring: what specific players say and announce |

Research backs up the point about data volume. A literature review by Stieglitz and colleagues found that [“the volume of data was most often cited as a challenge by researchers”](https://www.research.ed.ac.uk/en/publications/social-media-analytics-challenges-in-topic-discovery-data-collect/). The same paper splits the process into four steps: discovery, collection, preparation and analysis.

## When the other tools are the right choice

The point of this site is not that keyword alerts, media monitoring and social listening are useless. They are built for other questions, and for those questions they are the better tool.

In **crisis communication**, what third parties say is exactly what matters. A systematic review by Mats Eriksson (2018) lists [“use social media monitoring”](https://instituteforpr.org/lessons-for-crisis-communication-on-social-media-a-systematic-review-of-what-research-tells-the-practice/) among the five lessons research offers for effective crisis communication on social media. If you need to see how outrage spreads, you need the conversation, not the sender.

For **PR evaluation**, media monitoring and coverage analysis are the established method. If you want to know whether your own press work lands, you want the coverage about you.

For **topics where the sender does not matter**, a keyword alert is the right entry point: a new technical term, a standard, a piece of draft legislation whose spread you want to follow. Here the question really is “where does this show up?”.

So the line does not run between good and bad tools. It runs between the question about the conversation around someone and the question about what someone says. For market analysis the second question usually matters more. How Source-First relates to competitive intelligence, market intelligence and OSINT is covered in [Source-First in context](https://source-first-intelligence.com/en/source-first-in-context/).

## Common objections

### “Google Alerts is free and good enough.”

For the question “what new results mention this word?”, yes; that is what the service is built for. For the question “what is competitor X saying?”, a search term is the wrong way in. The options ask about channel type, language, region and volume, not about the sender. Whether the CEO’s post turns up depends on whether it contains the keyword and appears as a search result. Use keyword alerts for what they do well: tracking terms and discovering the new.

### “Good search operators will take care of the noise.”

Partly. Precise queries reduce irrelevant hits. But retrieval research describes the cost: every AND condition that removes noise also removes relevant results, and every OR that finds more also brings more noise. The textbook calls a satisfactory middle ground difficult or impossible.

The obvious candidate is the `site:` operator, which Google’s help describes as the way [to search for results from a specific site](https://support.google.com/websearch/answer/2466433?hl=en). That gets closer to the sender, but in our observation its limits are clear. It only sees web pages the search engine has indexed. Newsletters delivered by email do not appear, and posts on social media accounts only patchily. A source usually has several channels, and a domain is just one of them. And the entry point is still a search query: whatever the search engine does not find under that domain, it does not report.

Source-First changes the order instead: fix the sender first, then read, on every channel the sender publishes on.

### “We need completeness. Nothing can slip through.”

High recall is a legitimate goal. The textbook notes that [paralegals and intelligence analysts](https://nlp.stanford.edu/IR-book/html/htmledition/evaluation-of-unranked-retrieval-sets-1.html) aim for as high recall as possible and tolerate fairly low precision to get it. The real question is whether your team then reads the volume. And completeness across every place a word appears is not the same as completeness across what your twenty most important market players publish. The second is achievable. The first hardly is.

### “Our social listening tool already covers this.”

Social listening is designed around conversations about a brand, a product or a topic, with a focus on sentiment. That is a different question from what a competitor announces on its own channels. If you configure a social listening tool to follow specific accounts, you are already working source-first. The remaining question is whether the tool covers the channels your sources actually publish on.

### “What about the sources I don’t know?”

This is the strongest objection, and it lands. A list of sources contains only the ones you know. A new, small competitor is missing until someone adds it. George Day and Paul Schoemaker open their Harvard Business Review article with Mattel, which between 2001 and 2004 [lost 20% of its share of the worldwide fashion-doll segment to smaller rivals such as MGA Entertainment](https://hbr.org/2005/11/scanning-the-periphery), and “didn’t see it coming”. According to the authors, MGA had spotted a customer trend that Mattel missed: preteen girls were maturing faster and preferred dolls that looked like their teenage siblings. New rivals, technologies and regulations, the authors write, often seem to come out of left field.

The answer is a division of labour: **keywords to find, sources to follow.** Keyword searches and mentions are a good tool for the edge of your field of view. Use them to discover new players, and ask of every hit: who said this, and does this sender belong on the list? Ongoing monitoring belongs to the sources. That gives the high-recall approach the job where completeness matters, and the high-precision approach the job where reading time matters. Social media research also treats [discovery as a step of its own before data collection](https://www.research.ed.ac.uk/en/publications/social-media-analytics-challenges-in-topic-discovery-data-collect/).

Part of this is reviewing the source list regularly: who has appeared at trade fairs, in associations or in public tenders? How to build and maintain such a list is the subject of [Source-First in practice](https://source-first-intelligence.com/en/source-first-in-practice/).

### “Sources speak in their own interest.”

True. A press release is a primary source, but it is not neutral. Source-First tells you what a player says, not whether it is true. For market monitoring, that is precisely the value: what a competitor promises its customers, which roles it hires for and which markets it names are themselves the information. Judging what holds up remains work that no tool does for you. What that looks like in concrete situations is shown in the [Source-First cases](https://source-first-intelligence.com/en/source-first-cases/).

## The question picks the tool

Before choosing any tool, one test is worth running: do I want to know where something is said, or what someone says? Keyword alerts, media monitoring and social listening are built for the first, and they do it well. For the second, you start with the sources. Keywords keep their place as a tool for discovery. But following what specific market players say and plan starts with “who?”.

Some questions fit neither pattern. “Who is hiring right now?” or “Who is putting work like ours out to tender?” ask neither about a word nor about a known sender, but about every entry of one kind in one place. For those there is a third logic alongside [Topic-First](https://source-first-intelligence.com/en/topic-first/) and Source-First: [Platform-First](https://source-first-intelligence.com/en/platform-first/), which follows a platform such as a job board or a tender portal as a whole. How the three approaches differ and complement each other is covered in [Market Monitoring: Topic-First, Source-First, Platform-First](https://source-first-intelligence.com/en/three-approaches-to-market-monitoring/). Source-First and Platform-First are compared directly in [Source or Platform?](https://source-first-intelligence.com/en/source-or-platform/).

## References

1. Google: Create an alert. Google Search Help, undated. [https://support.google.com/websearch/answer/4815696?hl=en](https://support.google.com/websearch/answer/4815696?hl=en) (accessed 23 September 2026)
2. Google: Refine Google searches. Google Search Help, undated. [https://support.google.com/websearch/answer/2466433?hl=en](https://support.google.com/websearch/answer/2466433?hl=en) (accessed 23 September 2026)
3. Google: Google Alerts, interface with options (how often, sources, language, region, how many). Undated. [https://www.google.com/alerts?hl=en](https://www.google.com/alerts?hl=en) (accessed 23 September 2026)
4. Google News Initiative: Google Alerts: Stay in the know. Training, undated. [https://newsinitiative.withgoogle.com/resources/trainings/google-alerts-stay-in-the-know/](https://newsinitiative.withgoogle.com/resources/trainings/google-alerts-stay-in-the-know/) (accessed 23 September 2026)
5. AMEC (International Association for Measurement and Evaluation of Communication): Glossary. Undated. [https://amecorg.com/glossary/](https://amecorg.com/glossary/) (accessed 23 September 2026)
6. Stewart, Margaret C.; Arnold, Christa L.: Defining Social Listening: Recognizing an Emerging Dimension of Listening. International Journal of Listening 32 (2), pp. 85–100, 4 May 2018, DOI 10.1080/10904018.2017.1330656. [https://scholars.unf.edu/en/publications/defining-social-listening-recognizing-an-emerging-dimension-of-li/](https://scholars.unf.edu/en/publications/defining-social-listening-recognizing-an-emerging-dimension-of-li/) (accessed 23 September 2026)
7. Manning, Christopher D.; Raghavan, Prabhakar; Schütze, Hinrich: Introduction to Information Retrieval, section 18.4 “Latent semantic indexing”. Cambridge University Press, 2008. [https://nlp.stanford.edu/IR-book/html/htmledition/latent-semantic-indexing-1.html](https://nlp.stanford.edu/IR-book/html/htmledition/latent-semantic-indexing-1.html) (accessed 23 September 2026)
8. Manning; Raghavan; Schütze: Introduction to Information Retrieval, chapter 9 “Relevance feedback and query expansion”. Cambridge University Press, 2008. [https://nlp.stanford.edu/IR-book/html/htmledition/relevance-feedback-and-query-expansion-1.html](https://nlp.stanford.edu/IR-book/html/htmledition/relevance-feedback-and-query-expansion-1.html) (accessed 23 September 2026)
9. Manning; Raghavan; Schütze: Introduction to Information Retrieval, section 8.3 “Evaluation of unranked retrieval sets”. Cambridge University Press, 2008. [https://nlp.stanford.edu/IR-book/html/htmledition/evaluation-of-unranked-retrieval-sets-1.html](https://nlp.stanford.edu/IR-book/html/htmledition/evaluation-of-unranked-retrieval-sets-1.html) (accessed 23 September 2026)
10. Manning; Raghavan; Schütze: Introduction to Information Retrieval, section 1.4 “The extended Boolean model versus ranked retrieval”. Cambridge University Press, 2008. [https://nlp.stanford.edu/IR-book/html/htmledition/the-extended-boolean-model-versus-ranked-retrieval-1.html](https://nlp.stanford.edu/IR-book/html/htmledition/the-extended-boolean-model-versus-ranked-retrieval-1.html) (accessed 23 September 2026)
11. Manning; Raghavan; Schütze: Introduction to Information Retrieval, section 1.1 “An example information retrieval problem”. Cambridge University Press, 2008. [https://nlp.stanford.edu/IR-book/html/htmledition/an-example-information-retrieval-problem-1.html](https://nlp.stanford.edu/IR-book/html/htmledition/an-example-information-retrieval-problem-1.html) (accessed 23 September 2026)
12. Eppler, Martin J.; Mengis, Jeanne: The Concept of Information Overload: A Review of Literature from Organization Science, Accounting, Marketing, MIS, and Related Disciplines. The Information Society 20 (5), pp. 325–344, 1 May 2004, DOI 10.1080/01972240490507974. [https://www.alexandria.unisg.ch/54792](https://www.alexandria.unisg.ch/54792) (accessed 23 September 2026)
13. Ancker, Jessica S. et al.: Effects of workload, work complexity, and repeated alerts on alert fatigue in a clinical decision support system. BMC Medical Informatics and Decision Making 17, 36, 2017, DOI 10.1186/s12911-017-0430-8. [https://pmc.ncbi.nlm.nih.gov/articles/PMC5387195/](https://pmc.ncbi.nlm.nih.gov/articles/PMC5387195/) (accessed 23 September 2026)
14. Stieglitz, Stefan; Mirbabaie, Milad; Ross, Björn; Neuberger, Christoph: Social media analytics – Challenges in topic discovery, data collection, and data preparation. International Journal of Information Management 39, pp. 156–168, 30 April 2018, DOI 10.1016/j.ijinfomgt.2017.12.002. [https://www.research.ed.ac.uk/en/publications/social-media-analytics-challenges-in-topic-discovery-data-collect/](https://www.research.ed.ac.uk/en/publications/social-media-analytics-challenges-in-topic-discovery-data-collect/) (accessed 23 September 2026)
15. Eriksson, Mats: Lessons for Crisis Communication on Social Media: A Systematic Review of What Research Tells the Practice. Author’s summary at the Institute for Public Relations, 23 October 2018 (original article: International Journal of Strategic Communication 12 (5), pp. 526–551, 2018). [https://instituteforpr.org/lessons-for-crisis-communication-on-social-media-a-systematic-review-of-what-research-tells-the-practice/](https://instituteforpr.org/lessons-for-crisis-communication-on-social-media-a-systematic-review-of-what-research-tells-the-practice/) (accessed 23 September 2026)
16. Day, George S.; Schoemaker, Paul J. H.: Scanning the Periphery. Harvard Business Review, November 2005. [https://hbr.org/2005/11/scanning-the-periphery](https://hbr.org/2005/11/scanning-the-periphery) (accessed 23 September 2026)