Topics group insights together on a common theme, and show how that theme relates to records in your CRM. When you open the topic we show all the CRM records relating to the insights in the topic, but you can further filter the insights on this screen to show only ones relating to specific CRM scenarios.

Creating topic sets

Topic models group insights according to how semantically similar they are. When you create a model it will automatically discover and generate topics for your insights. You can then edit these topics, or tune the model to recreate the topics.

Topic model settings

Minimum Topic Size: Minimum Topic Size controls the minimum number of insights that can form a topic and thereby the number of topics that will be generated. Increasing this value results in fewer topics but of larger size whereas decreasing this value results in more micro topics being generated. We advise increasing this value rather than decreasing it.

Maximum number of topics: Maximum number of topics specifies, after training the topic model, the maximum number of topics that will be created. Topics will be combined to reach this target number. Making this number too small may force topics together that are not closely related.

How do I reduce outliers?

The number of insights classified as outliers is handled by the topic density parameter. Lowering the density reduces the number of outliers by making more topics.

Although lowering the density will decrease the number of outliers found in the data, this might force outliers to be put into topics where they do not belong. So make sure to strike a balance between keeping noise and reducing outliers.

I have only a few topics, how do I increase them?

If you have a small number of insights then it can be difficult to properly extract topics due to the amount of data available. Filtering to increase the number of insights will give better results.

Second, your minimum topic size might be too large for your number of insights. If you decrease the minimum size of topics, then you are much more likely to increase the number of topics generated. You could also lower the topic density to achieve a similar effect.

Third, although this does not happen very often, there simply aren't that many different topics to be found in your insights. You can often see this when you have many outliers.

I have too many topics, how do I reduce them?

If you have a large number of insights, then it is possible to generate a very large number of topics. But there are a few days to decrease the number:

First, we can set the minimum topic size much higher, relative to the number of insights to make sure that those small topics will not be generated. This is a modelling parameter that specifies the minimum number of documents needed in a topic. More insights in each topic mean fewer topics will be generated.

Second, you set the topic density to a higher level. This prevents those micro-topics being generated as the model needs many neighboring insights to create a topic.

Third, we can set number of topics to a value that seems logical to you, relative to the number of insights being considered. Note that topics are forced to merge which might result in a lower quality of topics. In practice, we would advise leaving "Number of Topics" as "Auto" as that will merge topics that are very similar. Dissimilar topics will therefore remain separated.