Mention that batches are using transactions more

docs/pages/docs/Getting started/writing_queries.md

@@ -194,6 +194,7 @@ otherwise.
 
 Multiple insert statements can be run efficiently by using a batch. To do that, you can
 use the `insertAll` method inside a `batch`:
+
 ```dart
 Future<void> insertMultipleEntries() async{
   await batch((batch) {
@@ -216,6 +217,10 @@ Future<void> insertMultipleEntries() async{
 }
 ```
 
+Batches are similar to transactions in the sense that all updates are happening atomically,
+but they enable further optimizations to avoid preparing the same SQL statement twice.
+This makes them suitable for bulk insert or update operations.
+
 ### Upserts
 
 Upserts are a feature from newer sqlite3 versions that allows an insert to 

drift/lib/src/runtime/api/batch.dart

@@ -1,8 +1,16 @@
 part of 'runtime_api.dart';
 
-/// Contains operations to run queries in a batched mode. This can be much more
+/// Contains operations to run queries in a batched mode.
-/// efficient when running a lot of similar queries at the same time, making
+///
-/// this api suitable for bulk updates.
+/// Inside a batch, a set of SQL statements is collected and then run at once.
+/// Conceptually, batches are similar to a transaction (and they will use
+/// transactions internally).
+/// Additionally, batches are very efficient when the same SQL statement is
+/// executed with different parameters. Outside of a batch, a new statement
+/// would be parsed and prepared for each execution. With batches, statements
+/// are only prepared once and then run with the parameters needed.
+///
+/// This makes batches particularly suitable for bulk updates.
 class Batch {
   final List<String> _createdSql = [];
   final Map<String, int> _sqlToIndex = {};