Wikidata statistics make it obvious; the numbers of items, statements and labels are huge. To have an impact it is important to use the tools that are available.
The mathematics involved explain how you can have the most impact. Set theory helps you understand how to approach this effectively. When you add one label and it affects 10,000 items it is powerful. When you add every known lawyer of a country in one go and it affects 10,000 items it is powerful.
Choosing the tools that help you be effective is therefore what you teach the participants of a datathon. They go home and they are likely to continue using the tools.
The most obvious tool to introduce is Reasonator. For it to function well, it needs to be configured. This is done partly from the personal settings. This ensures that you Wikidata data will show as information in Reasonator in *your* language by preference.
The other part to configure Reasonator is Widar. Widar is a tool that authenticates the edits you do in Reasonator to Wikidata. In the screenshot above there is a picture that was added to Wikidata in this way. The most relevant use case is adding labels in your language. The missing ones are underlined in red.
AutoList is probably the second most relevant tool to demonstrate. It provides an interface to several tools. One of them is the WDQ or Wikidata Query another is the CatScan. Combining these tools allows you to find every human in a category that is not known in Wikidata for what the category indicates is true about this human... and then add the statement that does just that.
There are other tools that are extremely useful. In a datathon it is important to have a mix of people. People who know the tools well, people who are interested to learn and people who have a "mission". When the activities concentrate on the "missions", it means that you zoom in on a subject and apply the skills and labour on doing good. This makes it clear how relatively little effort can have a huge impact. Not only for a language but also for a subject domain.