Those can be based on their previous orders and interests.
NN/g recommends prioritizing them over more generic recommendations. Put that ‘Recommended For You’ section high up on the homepage. (source)
Specify how you came up with those personalized recommendations.
A good example is Netflix:
Amazon suggests products based on products you viewed:
At a minimum, mention that these recommendations are ‘Based on Your Past History’ or ‘Related to Items You’ve Purchased’
Break down the personalized recommendations in categories
… especially if you have a large inventory. For example, Spotify groups their recommendations by music genre.
Provide users with a way to fine-tune the suggestions
This can be for example a Like or Dislike action. A long while back, I was using Pandora, when it was available in Canada. From what I remember, I could give feedback as to which songs I liked, so that it could suggest more of that style.
On Amazon, you can open your browsing history and give feedback on the recommendations, by clicking ‘More Like This’ or ‘Remove’ (image 3)
As soon as the feedback is provided, the recommendation list should update immediately.