diff --git a/Cargo.lock b/Cargo.lock
index a005b4b3..95eabfab 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -2570,6 +2570,7 @@ dependencies = [
  "tracing-error",
  "tracing-futures",
  "zebra-chain",
+ "zebra-test",
 ]
 
 [[package]]
diff --git a/tower-fallback/src/lib.rs b/tower-fallback/src/lib.rs
index 533f4c8a..2ea7ea53 100644
--- a/tower-fallback/src/lib.rs
+++ b/tower-fallback/src/lib.rs
@@ -1,7 +1,15 @@
-/// A service combinator that sends requests to a first service, then retries
-/// processing on a second fallback service if the first service errors.
-///
-/// TODO: similar code exists in linkerd and could be upstreamed into tower
+//! A service combinator that sends requests to a first service, then retries
+//! processing on a second fallback service if the first service errors.
+//!
+//! Fallback designs have [a number of downsides][aws-fallback] but may be useful
+//! in some cases. For instance, when using batch verification, the `Fallback`
+//! wrapper can be used to fall back to individual verification of each item when
+//! a batch fails to verify.
+//!
+//! TODO: compare with similar code in linkerd.
+//!
+//! [aws-fallback]: https://aws.amazon.com/builders-library/avoiding-fallback-in-distributed-systems/
+
 pub mod future;
 mod service;