diff --git a/src/main/resources/Documentation/config.md b/src/main/resources/Documentation/config.md
index d401030..db93df6 100644
--- a/src/main/resources/Documentation/config.md
+++ b/src/main/resources/Documentation/config.md
@@ -566,7 +566,7 @@
 	By default, replicates without matching, i.e. replicates
 	everything to all remotes.
 
-remote.NAME.slowLatencyThreshold
+<a name="remote.NAME.slowLatencyThreshold">remote.NAME.slowLatencyThreshold</a>
 :	the time duration after which the replication of a project to this
 	destination will be considered "slow". A slow project replication
 	will cause additional metrics to be exposed for further investigation.
diff --git a/src/main/resources/Documentation/metrics.md b/src/main/resources/Documentation/metrics.md
index aeb71fa..f2ac06d 100644
--- a/src/main/resources/Documentation/metrics.md
+++ b/src/main/resources/Documentation/metrics.md
@@ -3,7 +3,7 @@
 
 Some metrics are emitted when replication occurs to a remote destination.
 The granularity of the metrics recorded is at destination level, however when a particular project replication is flagged
-as slow. This happens when the replication took longer than allowed threshold (see _remote.NAME.slowLatencyThreshold_ in [config.md](config.md))
+as slow. This happens when the replication took longer than allowed threshold (see [`remote.NAME.slowLatencyThreshold`](config.md#remote.NAME.slowLatencyThreshold)).
 
 The reason only slow metrics are published, rather than all, is to contain their number, which, on a big Gerrit installation
 could potentially be considerably big.