diff --git a/src/autora/serializer/__init__.py b/src/autora/serializer/__init__.py
index d43e9725..a36f7687 100644
--- a/src/autora/serializer/__init__.py
+++ b/src/autora/serializer/__init__.py
@@ -1,14 +1,30 @@
+"""A submodule which handles importing supported serializers."""
+
 import importlib
 import logging
 import pathlib
 from collections import namedtuple
 from enum import Enum
-from typing import Callable, Dict, Literal, Optional, Tuple, Union
+from typing import Dict, Optional, Union
 
 from autora.state import State
 
 _logger = logging.getLogger(__name__)
 
+# Developer notes:
+# Add a new serializer by:
+# - including its name in the SerializersSupported Enum. This {name} will be available on the
+#   command line: `python -m autora.workflow --serializer {name}`
+# - Adding the basic data about it to the _SERIALIZER_INFO dictionary. This should include the
+#   fully qualified name under which it can be imported, and which file mode it expects:
+#     "" for regular [i.e. files will be opened as `open(filename, "w")` for writing]
+#     "b" for binary [i.e. files will be opened as `open(filename, "wb")` for writing]
+
+_SERIALIZER_INFO_ENTRY = namedtuple(
+    "_SERIALIZER_INFO_ENTRY", ["fully_qualified_module_name", "file_mode"]
+)
+LOADED_SERIALIZER = namedtuple("LOADED_SERIALIZER", ["module", "file_mode"])
+
 
 class SerializersSupported(str, Enum):
     """Listing of allowed serializers."""
@@ -19,43 +35,32 @@ class SerializersSupported(str, Enum):
 
 
 # Dictionary of details about each serializer
-_SERIALIZER_INFO_ENTRY = namedtuple(
-    "_SERIALIZER_INFO_ENTRY", ["module_path", "file_mode"]
-)
 _SERIALIZER_INFO: Dict[SerializersSupported, _SERIALIZER_INFO_ENTRY] = {
     SerializersSupported.pickle: _SERIALIZER_INFO_ENTRY("pickle", "b"),
     SerializersSupported.dill: _SERIALIZER_INFO_ENTRY("dill", "b"),
     SerializersSupported.yaml: _SERIALIZER_INFO_ENTRY("autora.serializer.yaml_", ""),
 }
 
-# Import those serializers which are actually importable
-_AVAILABLE_SERIALIZER_INFO = dict()
-_LOADED_SERIALIZER_DEF = namedtuple(
-    "_LOADED_SERIALIZER_DEF", ["name", "module", "file_mode"]
-)
-for serializer_enum in SerializersSupported:
-    serializer_info = _SERIALIZER_INFO[serializer_enum]
-    try:
-        module = importlib.import_module(serializer_info.module_path)
-    except ImportError:
-        _logger.info(f"serializer {serializer_info.module_path} not available")
-        continue
-    _AVAILABLE_SERIALIZER_INFO[serializer_enum] = _LOADED_SERIALIZER_DEF(
-        serializer_info.module_path, module, serializer_info.file_mode
-    )
-
+# Set the default serializer for the package
 default_serializer = SerializersSupported.pickle
 
+# A dictionary to handle lazy loading of the serializers
+_LOADED_SERIALIZERS: Dict[SerializersSupported, LOADED_SERIALIZER] = dict()
+
+
+def load_serializer(serializer: SerializersSupported) -> LOADED_SERIALIZER:
+    """Load"""
+
+    try:
+        serializer_def = _LOADED_SERIALIZERS[serializer]
+
+    except KeyError:
+        serializer_info = _SERIALIZER_INFO[serializer]
+        module_ = importlib.import_module(serializer_info.fully_qualified_module_name)
+        serializer_def = LOADED_SERIALIZER(module_, serializer_info.file_mode)
+        _LOADED_SERIALIZERS[serializer] = serializer_def
 
-def _get_serializer_mode(
-    serializer: SerializersSupported,
-    interface: Literal["load", "dump", "loads", "dumps"],
-) -> Tuple[Callable, str]:
-    serializer_def = _AVAILABLE_SERIALIZER_INFO[serializer]
-    module = serializer_def.module
-    function = getattr(module, interface)
-    file_mode = serializer_def.file_mode
-    return function, file_mode
+    return serializer_def
 
 
 def load_state(
@@ -63,11 +68,11 @@ def load_state(
     loader: SerializersSupported = default_serializer,
 ) -> Union[State, None]:
     """Load a State object from a path."""
+    serializer = load_serializer(loader)
     if path is not None:
-        load, file_mode = _get_serializer_mode(loader, "load")
         _logger.debug(f"load_state: loading from {path=}")
-        with open(path, f"r{file_mode}") as f:
-            state_ = load(f)
+        with open(path, f"r{serializer.file_mode}") as f:
+            state_ = serializer.module.load(f)
     else:
         _logger.debug(f"load_state: {path=} -> returning None")
         state_ = None
@@ -80,14 +85,13 @@ def dump_state(
     dumper: SerializersSupported = default_serializer,
 ) -> None:
     """Write a State object to a path."""
+    serializer = load_serializer(dumper)
     if path is not None:
-        dump, file_mode = _get_serializer_mode(dumper, "dump")
         _logger.debug(f"dump_state: dumping to {path=}")
         path.parent.mkdir(parents=True, exist_ok=True)
-        with open(path, f"w{file_mode}") as f:
-            dump(state_, f)
+        with open(path, f"w{serializer.file_mode}") as f:
+            serializer.module.dump(state_, f)
     else:
-        dumps, _ = _get_serializer_mode(dumper, "dumps")
         _logger.debug(f"dump_state: {path=} so writing to stdout")
-        print(dumps(state_))
+        print(serializer.module.dumps(state_))
     return
diff --git a/tests/test_serializer.py b/tests/test_serializer.py
index 247f3f11..52b758f4 100644
--- a/tests/test_serializer.py
+++ b/tests/test_serializer.py
@@ -5,23 +5,17 @@
 
 import hypothesis.strategies as st
 
-from autora.serializer import _AVAILABLE_SERIALIZER_INFO, SerializersSupported
+from autora.serializer import SerializersSupported, load_serializer
 
 logger = logging.getLogger(__name__)
 
 
 # Define an ordered list of serializers we're going to test.
-# We use the same order as the SerializersSupported Enum.
 AVAILABLE_SERIALIZERS = st.sampled_from(
-    [
-        _AVAILABLE_SERIALIZER_INFO[k]
-        for k in SerializersSupported
-        if k in _AVAILABLE_SERIALIZER_INFO
-    ]
+    [load_serializer(s) for s in SerializersSupported]
 )
 
 
-
 @st.composite
 def serializer_dump_load_string_strategy(draw):
     """Strategy returns a function which dumps an object and reloads it via a bytestream."""