diff --git a/tests/testthat/test-emissions_profile.R b/tests/testthat/test-emissions_profile.R
index 5393d2303..f7ff00810 100644
--- a/tests/testthat/test-emissions_profile.R
+++ b/tests/testthat/test-emissions_profile.R
@@ -97,3 +97,80 @@ test_that("in each benchmark, `profile_ranking` increases with `*co2_footprint`"
 
   expect_true(in_all_benchmarks_profile_ranking_increases_with_co2_footprint)
 })
+
+
+test_that("at product level, `NA` in a benchmark column yields `NA` in the corresponding `risk_category` and `profile_ranking` (#638)", {
+  companies <- example_companies()
+
+  benchmark <- "isic_4digit"
+  co2 <- example_products("{ benchmark }" := NA)
+
+  out <- emissions_profile(companies, co2)
+  product <- unnest_product(out)
+
+  corresponding <- filter(product, grouped_by == benchmark)
+  expect_true(is.na(corresponding$risk_category))
+  expect_true(is.na(corresponding$profile_ranking))
+})
+
+test_that("at product level, with no match preserves unmatched products, filling with `NA`s (#657)", {
+  companies <- example_companies(!!aka("uid") := c("unmatched"))
+
+  products <- example_products()
+  out <- emissions_profile(companies, products) |> unnest_product()
+
+  expect_equal(nrow(out), 1)
+  expect_equal(out[[aka("uid")]], "unmatched")
+
+  na_cols <- setdiff(cols_na_at_product_level(), aka("uid"))
+  all_na_cols_are_na <- all(map_lgl(na_cols, ~ is.na(out[[.x]])))
+  expect_true(all_na_cols_are_na)
+})
+
+test_that("at product level, with some match preserves unmatched products, filling with `NA`s (#657)", {
+  companies <- example_companies(!!aka("uid") := c("a", "unmatched"))
+
+  products <- example_products()
+  out <- emissions_profile(companies, products) |> unnest_product()
+
+  expect_true("unmatched" %in% out[[aka("uid")]])
+
+  unmatched_row <- 1
+  expect_equal(nrow(out), length(flat_benchmarks(products)) + unmatched_row)
+
+  unmatched <- filter(out, out[[aka("uid")]] == "unmatched")
+  na_cols <- setdiff(cols_na_at_product_level(), aka("uid"))
+  all_na_cols_are_na <- all(map_lgl(na_cols, ~ is.na(unmatched[[.x]])))
+  expect_true(all_na_cols_are_na)
+})
+
+test_that("at product level, with no match preserves unmatched products, filling with `NA`s (#657)", {
+  companies <- example_companies(!!aka("uid") := c("unmatched"))
+
+  co2 <- example_products()
+  out <- emissions_profile(companies, co2) |> unnest_product()
+
+  expect_equal(nrow(out), 1)
+  expect_equal(out[[aka("uid")]], "unmatched")
+
+  na_cols <- setdiff(cols_na_at_product_level(), aka("uid"))
+  all_na_cols_are_na <- all(map_lgl(na_cols, ~ is.na(out[[.x]])))
+  expect_true(all_na_cols_are_na)
+})
+
+test_that("at product level, with some match preserves unmatched products, filling with `NA`s (#657)", {
+  companies <- example_companies(!!aka("uid") := c("a", "unmatched"))
+
+  co2 <- example_products()
+  out <- emissions_profile(companies, co2) |> unnest_product()
+
+  expect_true("unmatched" %in% out[[aka("uid")]])
+
+  unmatched_row <- 1
+  expect_equal(nrow(out), length(flat_benchmarks(co2)) + unmatched_row)
+
+  unmatched <- filter(out, out[[aka("uid")]] == "unmatched")
+  na_cols <- setdiff(cols_na_at_product_level(), aka("uid"))
+  all_na_cols_are_na <- all(map_lgl(na_cols, ~ is.na(unmatched[[.x]])))
+  expect_true(all_na_cols_are_na)
+})
diff --git a/tests/testthat/test-emissions_profile_any_at_product_level.R b/tests/testthat/test-emissions_profile_any_at_product_level.R
index c1ac1f9d5..e9f347041 100644
--- a/tests/testthat/test-emissions_profile_any_at_product_level.R
+++ b/tests/testthat/test-emissions_profile_any_at_product_level.R
@@ -20,51 +20,6 @@ test_that("outputs expected columns at product level", {
   expect_named(out, expected)
 })
 
-test_that("unmatched products don't introduce NA's (#266)", {
-  companies <- example_companies(!!aka("uid") := c("a", "unmatched"))
-
-  products <- example_products()
-  out <- emissions_profile_any_at_product_level(companies, products)
-  expect_false(anyNA(out$risk_category))
-
-  inputs <- example_inputs()
-  out <- emissions_profile_any_at_product_level(companies, inputs)
-  expect_false(anyNA(out$risk_category))
-})
-
-test_that("some match yields no NA and no match yields 1 row with `NA`s (#393)", {
-  companies <- example_companies(
-    !!aka("id") := c("a", "a", "b", "b"),
-    !!aka("uid") := c("a", paste0("unmatched", 1:3))
-  )
-
-  products <- example_products()
-  out <- emissions_profile_any_at_product_level(companies, products)
-
-  some_match <- filter(out, companies_id == "a")
-  expect_false(anyNA(some_match))
-
-  no_match <- filter(out, companies_id == "b")
-  expect_equal(nrow(no_match), 1)
-
-  na_cols <- cols_na_at_product_level()
-  all_na_cols_are_na <- all(map_lgl(na_cols, ~ is.na(no_match[[.x]])))
-  expect_true(all_na_cols_are_na)
-
-  inputs <- example_inputs()
-  out <- emissions_profile_any_at_product_level(companies, inputs)
-
-  some_match <- filter(out, companies_id == "a")
-  expect_false(anyNA(some_match))
-
-  no_match <- filter(out, companies_id == "b")
-  expect_equal(nrow(no_match), 1)
-
-  na_cols <- cols_na_at_product_level()
-  all_na_cols_are_na <- all(map_lgl(na_cols, ~ is.na(no_match[[.x]])))
-  expect_true(all_na_cols_are_na)
-})
-
 test_that("with duplicated co2 throws no error (#435)", {
   companies <- example_companies()
   duplicated <- c("a", "a")
diff --git a/tests/testthat/test-emissions_profile_upstream.R b/tests/testthat/test-emissions_profile_upstream.R
index 89623970a..685d7487e 100644
--- a/tests/testthat/test-emissions_profile_upstream.R
+++ b/tests/testthat/test-emissions_profile_upstream.R
@@ -140,3 +140,48 @@ test_that("in each benchmark, `profile_ranking` increases with `*co2_footprint`"
 
   expect_true(in_all_benchmarks_profile_ranking_increases_with_co2_footprint)
 })
+
+test_that("at product level, `NA` in a benchmark column yields `NA` in the corresponding `risk_category` and `profile_ranking` (#638)", {
+  companies <- example_companies()
+
+  benchmark <- "input_isic_4digit"
+  co2 <- example_inputs("{ benchmark }" := NA)
+
+  out <- emissions_profile_upstream(companies, co2)
+  product <- unnest_product(out)
+
+  corresponding <- filter(product, grouped_by == benchmark)
+  expect_true(is.na(corresponding$risk_category))
+  expect_true(is.na(corresponding$profile_ranking))
+})
+
+test_that("at product level, with no match preserves unmatched products, filling with `NA`s (#657)", {
+  companies <- example_companies(!!aka("uid") := c("unmatched"))
+
+  co2 <- example_inputs()
+  out <- emissions_profile_upstream(companies, co2) |> unnest_product()
+
+  expect_equal(nrow(out), 1)
+  expect_equal(out[[aka("uid")]], "unmatched")
+
+  na_cols <- setdiff(cols_na_at_product_level(), aka("uid"))
+  all_na_cols_are_na <- all(map_lgl(na_cols, ~ is.na(out[[.x]])))
+  expect_true(all_na_cols_are_na)
+})
+
+test_that("at product level, with some match preserves unmatched products, filling with `NA`s (#657)", {
+  companies <- example_companies(!!aka("uid") := c("a", "unmatched"))
+
+  co2 <- example_inputs()
+  out <- emissions_profile_upstream(companies, co2) |> unnest_product()
+
+  expect_true("unmatched" %in% out[[aka("uid")]])
+
+  unmatched_row <- 1
+  expect_equal(nrow(out), length(flat_benchmarks(co2)) + unmatched_row)
+
+  unmatched <- filter(out, out[[aka("uid")]] == "unmatched")
+  na_cols <- setdiff(cols_na_at_product_level(), aka("uid"))
+  all_na_cols_are_na <- all(map_lgl(na_cols, ~ is.na(unmatched[[.x]])))
+  expect_true(all_na_cols_are_na)
+})