Summing Sales Based On Different Year Cases - sql

Question

| Product ID | YearBought | Sales | Min_Year | Max_Year |
| 1 | 2016 | $20 | 2011 | 2016 |
| 2 | 2016 | $10 | 2016 | 2018 |
| 2 | 2017 | $30 | 2016 | 2018 |
| 3 | 2017 | $5 | 2015 | 2019 |
| 3 | 2018 | $10 | 2015 | 2019 |
| 3 | 2018 | $20 | 2015 | 2019 |
| 3 | 2019 | $5 | 2015 | 2019 |
| 3 | 2019 | $30 | 2015 | 2019 |
| 4 | 2018 | $5 | 2018 | 2020 |
| 4 | 2019 | $10 | 2018 | 2020 |
| 4 | 2020 | $20 | 2018 | 2020 |
Min_Year = the year the product was first introduced
Max_Year + 1 = Product drop off year
Above is a sample of the table I'm working with. Trying to find:
the sum of sales new products brought in the year they were first introduced
the sum of "dropped sales" aka the sum of sales from products the year after they dropped off (had no sales). (Ex. Product brought in $15 in 2018 but had no sales in 2019, want to show $15 as dropped sales in 2019)
Expected Output:
| YearBought | New Product Sales | Dropped Product Sales |
| 2016 | $10 | |
| 2017 | | $20 |
| 2018 | $5 | |
| 2019 | | |
| 2020 | | $35 |
was thinking something like this but it's not working. any help would be appreciated!
select
YearBought,
sum(case when yearbought=min_year then sales else 0 end) as NewSales,
sum(case when yearbought=max_year+1 then sales else 0 end) as DropSales
from
#t
group by
yearbought

Answer 1

Aggregate the dropped product sales separately from the new product sales, then join the aggregations. You can do this with subqueries, or, as I have done below, with common table expressions.
with
droppedProds as (
select droppedYear = yearBought + 1,
foregoneSales = sum(sales)
from #t t
where YearBought = Max_Year
group by YearBought
),
newSales as (
select YearBought,
sales = sum(sales),
newSales = sum(case when yearBought = min_year then sales end)
from #t t
group by YearBought
)
select YearBought,
n.sales,
n.newSales,
d.foregoneSales
from newSales n
left join droppedProds d on n.yearBought = d.droppedYear
order by YearBought;
Results in:
+------------+-------+----------+---------------+
| YearBought | sales | newSales | foregoneSales |
|------------|-------|----------|---------------|
| 2016 | 30 | 10 | |
| 2017 | 35 | | 20 |
| 2018 | 35 | 5 | |
| 2019 | 45 | | |
| 2020 | 20 | | 35 |
+------------+-------+----------+---------------+

Answer 2

You could list the dates first, then join it with the original table and do conditional aggregation:
select
y.yearbought,
sum(case when t.yearbought = t.min_year then sales end) new_product_sales,
sum(case when t.yearbought = t.max_year then sales end) dropped_product_sales
from (select distinct yearbought from #t) y
inner join #t on y.yearbought in (t.min_year, t.max_year + 1)
group by y.yearbought

Answer 3

I think your problem is that the years are staggered.
with n as (
select year_bought as y, sum(sales) as sales,
sum(case when year_bought = min_year then sales end) as ns
from T group by year_bought
), d as (
select year_bought + 1 as y,
sum(case when year_bought = max_year then sales end) as ds
from T group by year_bought
)
select y, sales,
ns as newSales, coalesce(dropped, 0) as foregoneSales
from n left outer join p on p.y = n.y;
or use lead()
select year_bought, sum(Sales) as Sales,
sum(case when year_bought = min_year then sales end) as newSales,
coalesce(lead(sum(case when year_bought = max_year then sales end))
over (order by year_bought), 0) as foregoneSales
from T
group by year_bought;