Long tail search terms can be the most cost-effective when it comes to ad-cost vs conversation revenue, but they are harder to analyze and optimize because they are inherently more varied and less frequent. To help clients understand and improve these metrics, I used word vectors with PCA dimension-reduction and density-based clustering in order to group the terms into topics. Then I used PowerBI to help the client aggregate the cost and conversion numbers of the topics while being able to click into any of them to get the details of the exact search terms that comprised the group.