How to advanced query based on join - mysql

Question

My schema looks like this:
SET GLOBAL sql_mode=(SELECT REPLACE(##sql_mode,'ONLY_FULL_GROUP_BY',''));
create table ads(
ad_id int,
ad_name varchar(10)
);
create table ads_insight(
id int,
ad_id int,
date date,
ad_clicks int
);
create table products(
id int,
name varchar(10)
);
create table products_insight(
id int,
product_id int,
sale int,
date date
);
create table ads_products(
ad_id int,
product_id int
);
insert into ads(ad_id, ad_name) values
(1,'ad1'),
(2,'ad2'),
(3,'ad3');
insert into ads_insight(id, ad_id, date, ad_clicks) values
(1, 1, '2021-04-25', 1),
(3, 1, '2021-04-23', 2),
(4, 1, '2021-04-22', 8),
(5, 2, '2021-04-25', 6),
(6, 2, '2021-03-03', 7);
insert into products(id, name) values
(1,'prod1'),
(2,'prod2'),
(3,'prod3'),
(4,'prod4'),
(5,'prod5');
insert into products_insight(id, product_id, sale, date) values
(1, 1, 10, '2021-04-25'),
(2, 1, 13, '2021-04-24'),
(3, 1, 15, '2021-04-23'),
(4, 1, 14, '2021-04-22'),
(5, 1, 17, '2021-04-21'),
(6, 1, 15, '2021-04-20'),
(7, 1, 13, '2021-04-19'),
(8, 2, 15, '2021-04-25');
insert into ads_products (ad_id, product_id) values
(1, 1),
(1, 2),
(2, 3),
(2, 4),
(2, 2),
(3, 1);
Here you have fiddle
A quick explanation of schema:
I have ads:
each ad has insights, which tell us when a certain ad was active(=> ad_clicks has to be > 0).
each ad has products(many2many - ads_products table). Each product has products_insight which tells us how many sales that product generated on a certain day.
Now I want to get all ads from the time range 2021-04-20 - 2021-04-25 which had ad_clicks > 0 (which I have done) AND count how many sales each ad has generated when it was active. So count sale only if the ad has ad_insight and ad_clicks > 0.
My query looks like this:
SET #from_date = '2021-04-20';
SET #to_date = '2021-04-25';
SELECT
ads.ad_name,
IFNULL(ad_clicks, 0) AS clicks,
IFNULL(product_sale, 0) AS product_sale,
IFNULL(products, '') AS products
FROM ads
LEFT JOIN (
SELECT ad_id, SUM(ad_clicks) AS ad_clicks
FROM ads_insight
WHERE date BETWEEN #from_date AND #to_date
GROUP BY ad_id
) AS ai ON ai.ad_id = ads.ad_id
LEFT JOIN (
SELECT ad_id, SUM(sale) AS product_sale
FROM ads_products AS ap
LEFT JOIN products_insight AS pi ON pi.product_id = ap.product_id
WHERE date BETWEEN #from_date AND #to_date
GROUP BY ad_id
) AS pi ON pi.ad_id = ads.ad_id
LEFT JOIN (
SELECT ap.ad_id, GROUP_CONCAT(DISTINCT p.name) AS products
FROM ads_products AS ap
JOIN products AS p ON ap.product_id = p.id
GROUP BY ap.ad_id
) AS p ON ads.ad_id = p.ad_id
WHERE ad_clicks>0;
And it generates the following result:
| ad_name | clicks | product_sale | products |
| ------- | ------ | ------------ | ----------------- |
| ad1 | 11 | 99 | prod1,prod2 |
| ad2 | 6 | 15 | prod2,prod3,prod4 |
But I want this(there is a difference in the product_sale column)
| ad_name | clicks | product_sale | products |
| ------- | ------ | ------------ | ----------------- |
| ad1 | 11 | 55 | prod1,prod2 |
| ad2 | 6 | 15 | prod2,prod3,prod4 |
54 because it counts only rows with id: 1,3,4 from products_insight because in these days ad with id 1 was active. (active means that there is a row in ads_insight table.