Updated: Apr 28
When people see breaking news and want to drill deeper, they talk with trusted friends. They turn to trusted media. They search online. The quest for information is a precursor to forming an opinion, making a decision, and potentially acting on that decision.
For companies covered in the news, that information-gathering process is a form of customer engagement. From a communication research perspective, that makes Google Trends a go-to data source tracking online search volumes. At its core, Google Trends is a barometer of online customer engagement.
Aside from the price point -- free -- the beauty of Google Trends lies in its simplicity. Instead of reporting absolute search volume, the data is indexed. The search volume on the most active day is represented as 100, and search volume on any other given day is represented basically as a fraction of that most active news cycle. Reliable. Validated. Replicated.
OK, so what does this have to do with news volume and tonality? Stick with me here as we begin to forge the news stream.
As we've discussed in earlier blogs, news volume for companies covered in the national press can equate to billions of "impressions," which tend to spike on the heaviest news days. At the same time, tonality of that coverage could be positive, negative or neutral, and tends to spike on the lightest news days.
Media monitoring generates two data points representing published news -- both moving in different directions at the same time, and neither fully representing qualities of news coverage on a given day. Plus the "impressions" numbers can be very large -- to the point of becoming abstract.
By combining those two data points -- daily volume and average tonality -- we can effectively create a single metric representing two key qualities of published news. Using a media monitoring tool like Meltwater that efficiently aggregates news data into spreadsheets, we are only a pivot table away from generating a news stream,
First let's talk about how we could do this. Then let's talk about why we should do this.
Here's how we do it in class. Use a Boolean search string that will capture coverage mentioning the company in the first few paragraphs to reduce tertiary references. Search a directed sample of news outlets to reduce reliability error caused by boiling the ocean. Download the date, headline and tonality of each story into a spreadsheet.
Multiplying the volume and tonality scores together yields a single number. Dividing each daily scorer by the absolute value of the largest score indexes the data on a 100-point scale in both directions -- positive and negative. I can send you a one-pager with step-by-step directions if you are interested in trying this at home.
Ultimately, your news stream will look a lot like the example on the Whiteboard. Reliable. Validated. Replicated. Good stuff.
The next question, then, is why go to the trouble?
For one, indexing is statistical alchemy. It takes an abstract concept -- say, for example, the news volume and tonality of year-end news coverage -- and makes it more discrete. Here's a simple example. For most public companies, the news stream will confirm that year-end earnings tend to generate a spike in coverage. That spike becomes a handy benchmark for assessing news about a legal settlement, product launch, or customer mishap.
By indexing on a 100-point scale, we also have simplified things immensely. Indexing the news stream is like shifting to the communication research equivalent of the metric system. The math is easier -- and more intuitive -- when we are working with 100-point scales.
And most importantly, by tracking media content using an indexed 100-point scale, we can directly compare news data to survey data, which also uses a 100-point scale. Qualities of what was published align to perceptions of what people saw or heard about our company or products. Content -> Consumer.
At this point, you my be asking yourself "why bother?" Eventually, we'll talk about Separating the Signals from the Noise. But first, let's look at Navigating the Storm.
Back to the Whiteboard.
If you would like to receive a one-pager
outlining how to build an indexed news stream,
feel free to email me at