Start by checking what your opponent threatens. A middlegame plan is only useful if it survives the next move. Once immediate tactics are covered, inspect the pawn structure, compare piece activity, and decide where a pawn break could improve your position. These observations give you a purpose for the next few moves without pretending that you can predict the whole game.
In a London setup, White may aim for e4, place a knight on e5, or build pressure towards the kingside. These plans require different preparation. The e4 advance can open the centre, but the pawn needs enough support and White's king must be safe. Ne5 can be an active outpost, but only if Black cannot exchange it advantageously or undermine its support.
Ask which of your pieces is doing the least. A knight behind a pawn chain might need a new route; a rook might belong on a file that a pawn exchange will open. Improving one piece is often a more realistic plan than announcing a kingside attack. Check the opponent's best reply after every proposed improvement.
At the end of an opening line, pause before restarting. Write one candidate move, the idea behind it, and the reply that worries you most. Compare your choice with an annotated game in the same structure. This is the bridge between knowing a sequence and playing chess: you are learning which features stay important after the memorised moves run out.
Try this at the board
In the diagram, list what White would need to check before playing e4. Include the uncastled king.