If you do it in the most sensitive direction before a near gravitational slingshot past a larger body, you change its course through the deeper gravity well. Effectively, the course correction to put in before, is amplified by the next interaction. The best time to get one is several years before a close pass, letting the ship sent (which can be robotic of course) hover over it for several years. Even a small mass can do the job with enough lead time.
So no, don't wait to see if it is too close after the pass, calculate well and deflect years before the 2029 pass.
"So no, don't wait to see if it is too close after the pass, calculate well and deflect years before the 2029 pass."
Burp.