What do you mean "a change in data"? Would it basically be a change in the catalog configuration?
What I would normally suggest is you figure out how you're deploying your pipeline. There a number of resources around deployment in the Kedro docs:
https://docs.kedro.org/en/stable/deployment/index.html Most production pipelines would be scheduled, depending on the deployment option.
Then, your CD pipeline would do those deployment steps, triggered on PR merge.
If you want something more simple--but most likely not recommended for production--you could consider not having a CD pipeline and running the pipeline in CI, using your CI machines. This will almost certainly not scale beyond sample data volumes, and you'd have to schedule CI runs in order to update results periodically. It's a pretty janky solution.