diff --git a/survey_record_generation/models/survey_user_input.py b/survey_record_generation/models/survey_user_input.py
index 8cccae0..68b2192 100644
--- a/survey_record_generation/models/survey_user_input.py
+++ b/survey_record_generation/models/survey_user_input.py
@@ -61,6 +61,11 @@ class SurveyUserInput(models.Model):
                     )
                     field_name: str = field_value.field_id.name
 
+                    if value is None:
+                        # No value (e.g. question not displayed or skipped):
+                        # keep the default/computed value of the field, and
+                        # let another question fill up the same field.
+                        continue
                     vals[field_name] = value
 
                 existing_record = self.find_existing_record(record_creation, vals)
@@ -195,7 +200,7 @@ class SurveyUserInput(models.Model):
         if unique_fields:
             uniq_domain = []
             for uniq_field in unique_fields:
-                uniq_domain.append((uniq_field, "=", vals[uniq_field]))
+                uniq_domain.append((uniq_field, "=", vals.get(uniq_field, False)))
             duplicate = self.env[model].search(uniq_domain, limit=1)
         return duplicate
 
diff --git a/survey_record_generation/tests/test_computed_fields.py b/survey_record_generation/tests/test_computed_fields.py
index 79f14ce..4ae40b2 100644
--- a/survey_record_generation/tests/test_computed_fields.py
+++ b/survey_record_generation/tests/test_computed_fields.py
@@ -110,3 +110,11 @@ class TestComputedFields(SurveyCase):
 
         self.assertEqual(len(line), 1)
         self.assertEqual(line.price_unit, 33.0)
+
+    def test_computed_field_applies_when_question_is_not_displayed(self):
+        # E.g. the "free amount" question is only displayed when the user
+        # chose "Other amount": otherwise the line keeps the computed price.
+        line = self._submit()
+
+        self.assertEqual(len(line), 1)
+        self.assertEqual(line.price_unit, 20.0)
diff --git a/survey_record_generation/tests/test_survey_record_creation.py b/survey_record_generation/tests/test_survey_record_creation.py
index 79e5e75..17f007e 100644
--- a/survey_record_generation/tests/test_survey_record_creation.py
+++ b/survey_record_generation/tests/test_survey_record_creation.py
@@ -1072,3 +1072,82 @@ class TestSurveyRecordCreation(SurveyCase):
         self.assertEqual(self.answer.partner_id, jean)
         self.assertNotEqual(self.answer.partner_id, jeanne)
         self.assertEqual(self.answer.email, jean.email)
+
+    def test_question_not_displayed_does_not_override_default_value(self):
+        # When a question is not displayed to the user (e.g. a conditional
+        # question), the field must keep its default (or computed) value
+        # instead of being emptied.
+        question_type = self._add_question(
+            page=None,
+            name="Type",
+            qtype="simple_choice",
+            labels=[{"value": "contact"}, {"value": "other"}],
+            survey_id=self.survey.id,
+            sequence=2,
+            answer_values_type="value",
+        )
+        for suggested_answer in question_type.suggested_answer_ids:
+            suggested_answer.value_char = suggested_answer.value
+        type_field = self.env["ir.model.fields"]._get("res.partner", "type")
+        self.env["survey.record.creation.field.values"].create(
+            {
+                "survey_record_creation_id": self.survey_record_creation.id,
+                "survey_id": self.survey.id,
+                "model_id": self.res_partner_model.id,
+                "field_id": type_field.id,
+                "value_origin": "question",
+                "question_id": question_type.id,
+            }
+        )
+
+        self.answer = self._add_answer(
+            survey=self.survey, partner=False, email="jean@test.fr"
+        )
+        self._add_answer_line(
+            question=self.question_name, answer=self.answer, answer_value="Jean"
+        )
+        self.answer._mark_done()
+
+        partner = self.env["res.partner"].search([("name", "=", "Jean")])
+        self.assertEqual(partner.type, "contact")
+
+    def test_several_conditional_questions_on_the_same_field(self):
+        # Several questions can fill up the same field (e.g. one list of
+        # amounts per periodicity, displayed conditionally): the answered
+        # question wins, the questions not displayed are ignored.
+        street_field = self.env["ir.model.fields"]._get("res.partner", "street")
+        questions = self.env["survey.question"]
+        for sequence, name in enumerate(["Street (France)", "Street (abroad)"], 2):
+            question = self._add_question(
+                page=None,
+                name=name,
+                qtype="char_box",
+                survey_id=self.survey.id,
+                sequence=sequence,
+            )
+            self.env["survey.record.creation.field.values"].create(
+                {
+                    "survey_record_creation_id": self.survey_record_creation.id,
+                    "survey_id": self.survey.id,
+                    "model_id": self.res_partner_model.id,
+                    "field_id": street_field.id,
+                    "value_origin": "question",
+                    "question_id": question.id,
+                }
+            )
+            questions |= question
+
+        self.answer = self._add_answer(
+            survey=self.survey, partner=False, email="jean@test.fr"
+        )
+        self._add_answer_line(
+            question=self.question_name, answer=self.answer, answer_value="Jean"
+        )
+        # Only the first question is displayed and answered
+        self._add_answer_line(
+            question=questions[0], answer=self.answer, answer_value="1 rue de Paris"
+        )
+        self.answer._mark_done()
+
+        partner = self.env["res.partner"].search([("name", "=", "Jean")])
+        self.assertEqual(partner.street, "1 rue de Paris")