BigQuery: How to make FULL JOIN with multiple tables ON multiple keys without losing data?

Question

I have been struggling with this issue in BigQuery for a while now and can't find a solution for it. I have 3 tables with campaign costs that I need to join together in a single table, here are some examples below of what the tables look like:

Table 1

Date	Country	Costs Campaign A
2021-06-01	Argentina	10
2021-06-01	Brazil	30
2021-06-01	Colombia	10
2021-06-02	Argentina	50
2021-06-02	Brazil	65
2021-06-02	Colombia	40

Table 2

Date	Country	Costs Campaign B
2021-06-01	Argentina	54
2021-06-01	Brazil	38
2021-06-01	Germany	94
2021-06-02	Argentina	51
2021-06-02	Brazil	48
2021-06-02	Germany	88

Table 3

Date	Country	Costs Campaign C
2021-06-01	Argentina	27
2021-06-01	Brazil	55
2021-06-01	Poland	46
2021-06-02	Argentina	86
2021-06-02	Brazil	99
2021-06-02	Poland	47

My output should look like this in the end

Date	Country	Costs Campaign A	Costs Campaign B	Costs Campaign C
2021-06-01	Argentina	10	54	27
2021-06-01	Brazil	30	38	55
2021-06-01	Colombia	10	0	0
2021-06-01	Germany	0	94	0
2021-06-01	Poland	0	0	46
2021-06-02	and so on...	x	y	z

And here is a similar query to the one I've been trying:

SELECT
  t1.Date,
  t1.Country,
  SUM(t1.CostsCampaignA),
  SUM(t2.CostsCampaignB),
  SUM(t3.CostsCampaignC)
FROM `table1` AS t1
FULL JOIN `table2` AS t2 ON t1.Date = t2.Date AND t1.Country = t2.Country
FULL JOIN `table3` AS t3 ON t1.Date = t3.Date AND t1.Country = t3.Country
GROUP BY 1,2

Now my problem is that for the countries not present in table 1, they do not appear in my output at all (in the example, Germany and Poland). I would assume that the FULL JOIN would ensure that no data gets lost, but this is not the case.

I would appreciate if anyone has a solution for my issue or any suggestions for a workaround.

Thank you!

Answer 1

Consider below approach

select * from (
  select date, country, cost_campaign_a as cost, 'A' campaign from table_1 
  union all select date, country, cost_campaign_b, 'B' from table_2 
  union all select date, country, cost_campaign_c, 'C' from table_3 
)
pivot (any_value(cost) cost_campaign for campaign in ('A', 'B', 'C'))

if applied to sample data in your question - output is