What recommendation tool handles sparse click data well?

A good default choice for sparse click data is a matrix factorization–based recommender, especially:

• Implicit Alternating Least Squares (Implicit ALS)
  Best when you have many users/items and very sparse interactions. It works well on implicit feedback like clicks, views, purchases.

Other strong options depending on your setup:

• BPR (Bayesian Personalized Ranking)
  Good for ranking from implicit data when you care about ordering items.
• LightFM
  Useful if you also have side features for users/items, and it handles sparse implicit data well.
• Factorization Machines / Field-aware FM
  Better if you want to combine sparse interaction data with metadata.
• Item-item collaborative filtering
  Simple and often surprisingly strong for extremely sparse click logs, especially at small scale.

If you mean a tool/library, good choices are:

• implicit (Python) for ALS/BPR on sparse matrices
• LightFM
• Cornac
• RecBole for more model variety

If you want, I can recommend the best one based on your data size, sparsity level, and whether you have user/item features.