Best Practices for Topic Monitoring

Best practices for topic monitoring varies depending on your goals. However, general practices can be applied in all situations to help you find the information you are looking for.

1. Cluster responses. Cluster responses before you add stopwords or customize themes.
2. Review themes. Review the predefined themes and sub-themes to determine if they are satisfactory. Look for common words that you think might be breaking your themes into smaller, less definitive clusters. Also look for certain words that you think might be useful in creating distinctive, useful themes.
3. Add stopwords. If you determine that there are common words that do not contribute to the creation of distinctive clusters, add them to the stopword list.
   Note: Stopwords you add are applied to all the responses you cluster.
4. Add custom themes. If you determine that there are certain words that can help create distinctive, useful themes, add a custom theme containing these keywords.
   Tip: Custom themes are most helpful in creating distinctive clusters when they use synonyms. For example, a theme containing the word “software” should also contain the words “application” and “program.” If you find that a word is often misspelled, you can include the common misspelling as a keyword.
5. Cluster responses again. Cluster the responses again and, if necessary, repeat the previous steps of reviewing themes, adding stopwords, and adding custom themes until you are satisfied with the resulting themes.
   Tip: If you use these steps and get results that are less satisfactory than the predefined themes, reset clustering by clicking Reset Clustering in the Tasks section of the Results tab. Any custom themes you created are retained, but they are not used in the clustering session.