3.9. How to build strong queries

In the following topics we will describe a four-step technique to build good queries/searches.

1. State your search question
2. Define objects of your search
3. Hunt for keywords
4. Good & bad keywords
5. Query types & calibration

What is a good query ?

A query can be evaluated upon three criteria that are interlinked:

• Relevancy rate VS Noise
• Completeness
• Number of results

Relevancy rates: 100% interesting articles in the results aren’t good news actually. In fact, in means you are surely losing interesting publications. Instead, go for a 30% relevancy (one interesting results for 3 not useful) or even 20% (1 vs 4). It is quite comfortable to monitor the news with such a rate.

Completeness: are you losing information that is not present in your results? Of course, the more you want to be sure of not losing anything, the more articles you will need to check. In the end, it means to dramatically lower you relevancy to increase the completeness.

Number of results: How many results you have ? You can find that value after launching a search in a theme, just below the search bar. This amount is a yearly one. For example, 10,000 results a year represents +- 27 items per day average. With a 20% relevancy rate, it could mean 5 interesting articles vs 22 that you won’t even open.

Goal

First state the number of results you are ready to process. Then try to reach that amount with the best relevancy possible. That would be a good mix.

In video