Based on your reply to my comment, I'm going to say that this is what is happening:
What you're seeing is a characteristic of the "Unique Visitors" metric, which basically counts the number of people (ignoring cookie deletions, tracker blocking, etc).
Consider these 2 people: Harry and Sally.
- Harry visited your site on 23 Sep 2020 and saw zero products.
- Sally visited your site on 20 Sep 2020 and saw zero products. But she also visited on 23 Sep 2020 and saw 2 products.
Let's assume that both Harry and Sally had never visited your site before these visits.
When you run the report for 23 Sep 2020 only, AA gets the data for 23 Sep 2020 only. That means it includes:
- Harry's visit on 23 Sep 2020 with zero products.
- Sally's visit on 23 Sep 2002 with 2 products.
Only Harry had zero products. So he gets counted under the "21,115" number.
But when you run the report for 13 Sep - 23 Sep 2020, AA gets the data for the entire date range of 13 Sep - 23 Sep 2020. That means it includes:
- Harry's visit on 23 Sep 2020 with zero products.
- Sally's visit on 20 Sep 2020 with zero products.
- Sally's visit on 23 Sep 2002 with 2 products.
Harry had zero products. So he gets counted under the "23,734" number.
Although Sally had 2 products on 23 Sep 2020, she had zero products on 20 Sep 2020. She's the same person who had different experiences, one of which qualified for the "zero products" segment. So she too gets counted under the "23,734" number.
And that's why you get two different numbers for the seemingly same criteria. The interpretation lies in the combination of date range and metrics used in your report.