diff --git a/docs/decisions/0026-csv-loader-multi-runs-update.rst b/docs/decisions/0026-csv-loader-multi-runs-update.rst
new file mode 100644
index 0000000000..57dd92c914
--- /dev/null
+++ b/docs/decisions/0026-csv-loader-multi-runs-update.rst
@@ -0,0 +1,46 @@
+26. CSV Loader Update to Support Multiple Course Runs
+=====================================================
+
+Status
+------
+Accepted (November 2023).
+
+Context
+-------
+Historically, the CSV loader was limited in its ability to handle multiple course runs for a single course.
+The existing functionality only allowed for the creation of a course along with its corresponding course run or the
+update of course and course run metadata for an already existing course. As a result, decisions regarding the placement
+of external LOBs courses' course run/variant information were recorded in the ``AdditionalMetadata`` fields, and these courses were
+designated for marketing purposes only and couldn't be purchased through the platform.
+
+Objectives
+----------
+This enhancement aims to address two primary objectives:
+
+1. Backpopulate dates and variants from additional metadata to CourseRun by either creating new runs or updating existing ones. This is a one-time process and applies only to external courses.
+2. Allow active and future products to follow a consistent, multiple run format.
+
+Decision
+--------
+
+To achieve the goals outlined above, the following criteria will determine when to create a new run:
+
+- For courses with a provided "variant Id" (External LOBs courses such as ExecEd, Bootcamps, etc.), the ``CSVLoader`` will check if the incoming
+    ``variant_id`` matches the ``variant_id`` of the active course run. If they are different, the system will then compare the schedule dates of the
+    incoming run with those of the active run in Discovery. If the schedule dates differ, a new run will be created. In all cases,
+    the dates and variants data will be either backpopulated to the existing run or used to create a new run.
+
+- For cases where the variant ID is not provided, the ``CSVLoader`` will compare the schedule dates of the incoming run with the active run in Discovery.
+    If there is a difference in schedule dates, a new run will be created. Otherwise, the existing run will be updated with the incoming dates and variants data from the CSV.
+
+As we have decided to support multiple runs, the ingestion summary email will be updated to include details on whether
+a new product or a new variant of an existing product is created during the ingestion process. This provides clearer insights into the changes
+made during each run of the CSV loader.
+
+Consequences
+------------
+Implementing this enhancement in the CSV loader will result in the following consequences:
+
+1. Backpopulating dates and variants from additional metadata ensures accurate representation in ``CourseRun``.
+2. The introduction of a multiple run format ensures uniformity for active and future products.
+3. Variant ↔ CourseRun mapping will become unique, aligning with the distinct key for each new run.