# what can cause the best-bound to get tighter in the first MIP node?

I'm using gurobi MIP optimization engine for solving a mixed integer linear minimization problem.

I see that the engine didn't start the branch and bound stage yet, as the engine reports Explored 1 nodes (70674 simplex iterations) in 904.16 seconds.

However, I can see that the BestBd keeps getting tighter and tighter (higher in my case). What's causing this? I thought that as long as the engine stays in one the root node, it's only performing cutting planes, which improve the Incumbent. I thought that the only thing that can improve the best-bound is if we branch from the root node to 2 new nodes, where the best relaxed solution would be tighter than what it was at the root...

What other techniques can the solver use to tighten the best-bound???