If I was actually running this experiment and had ideal resources, I'd

• Set up the students' default SRS server to be hosted by the school, and give them extra credit if they log in every day and do their daily reviews.
• Add some automated checks to detect usage patterns that probably represent just clicking through the thing without thinking.
• Provide a small common set of new cards to everyone based on each lecture, after the lecture. Everyone sat through the same lecture, so they can handle some of the same questions.
• Have instruction about making good cards, and have the students make their own cards as homework at the beginning of the course. The homework exercise cards would be reviewed like any homework.
• Would give extra credits to students for up to tripling the default deck with their own questions, which would pass some sort of spam filter check for random gibberish and a human inspection in questionable cases.

Yes, I know, overjustification effect and everything for using credit as the carrot to make the students use the system. Still, should be worth a try.

Also could just make the finished deck the "course project" which you'd submit along with doing the final exam for review, and which would contribute to your grade. So you'd want to have covered the material with good questions and have a good review profile.

Could also give students the choice between a SRS course and a regular one, where in the regular one their grade would be entirely defined by the exam, while in the SRS course points from the review of the deck they made would be added to their test score for determining the grade. "You do some random busywork and you'll get a guaranteed grade lift on that nasty calculus course."