If there is a huge difference between the estimated number of rows and the actual number of rows.
IF the estimation has a high percentage of qualifying rows when compared to the Actual percentage will represent the query needs to be fine tuned.
The cost calculation is always based on estimates and not actual. We need to identify the Thickness of the arrow which says the more data flow rate between two operations but this is based on actual number of rows rather than the estimated value.
Hence we need to pay special attention to areas in the plan where thick arrows appear before expensive iterators such as Sort, Spool, etc.