In Google Sheets, how do I flatten a hierarchical table with dynamic rows?

I have a dataset like this:

The number of ads is dynamic. The number of attributes is also dynamic and different for each ad. The attributes may repeat for different ads. I want to transform it to something like this:

I used =UNIQUE(FILTER(A:A,A:A<>"")) for the ad_id column and =ARRAYFORMULA(TRANSPOSE(UNIQUE(FILTER(B:B,B:B<>"")))) for turning the attributes column into headers but am stumped for how to pull the correct values. If the rows between the ids weren't blank it would be easy.

I tried =ArrayFormula(INDEX(C:C, MATCH(1, (A:A=ad_id_to_search)*(B:B=attribute_to_search), 0))) but this only works for the first row of each id branch. ie, it will only find the cost value and only if cost is the first attribute in the branch.

If you don't mind the attributes being in alphabetical order in the destination table, you can try:

=map(
query(
{A1:C1;{scan(,A2:A,lambda(acc,arr,if(len(arr),arr,acc))),B2:C}},
"select Col1,max(Col3) where Col2 is not null group by Col1 pivot Col2",1),
lambda(x,if(len(x),x,"NA")))

This is assuming that your source table is in A1:Cx of the sheet. The SCAN assigns the appropriate ad_id for those rows where it is missing, the QUERY then pivots the data by the attribute rows, and finally the MAP assigns an 'NA' for those attributes which were not present for a particular ad_id.