diff --git a/NEWS.md b/NEWS.md
index d33e8baef2..10e61338c4 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -226,6 +226,8 @@
 
 13. `test.data.table()` could fail if the `datatable.integer64` user option was set, [#3683](https://github.com/Rdatatable/data.table/issues/3683). Thanks @xiaguoxin for reporting.
 
+14. The warning message when using `keyby=` together with `:=` is clearer, [#2763](https://github.com/Rdatatable/data.table/issues/2763). Thanks to @eliocamp.
+
 
 ### Changes in [v1.12.2](https://github.com/Rdatatable/data.table/milestone/14?closed=1)  (07 Apr 2019)
 
diff --git a/R/data.table.R b/R/data.table.R
index 393ab81884..093cbb3047 100644
--- a/R/data.table.R
+++ b/R/data.table.R
@@ -1756,7 +1756,7 @@ replace_order = function(isub, verbose, env) {
         setkeyv(x,cnames)  # TO DO: setkey before grouping to get memcpy benefit.
         if (verbose) {cat(timetaken(last.started.at),"\n"); flush.console()}
       }
-      else warning(":= keyby not straightforward character column names or list() of column names, treating as a by:",paste(cnames,collapse=","),"\n")
+      else warning("The setkey() normally performed by keyby= has been skipped (as if by= was used) because := is being used together with keyby= but the keyby= contains some expressions. To avoid this warning, use by= instead, or provide existing column names to keyby=.\n")
     }
     return(suppPrint(x))
   }
diff --git a/inst/tests/tests.Rraw b/inst/tests/tests.Rraw
index 09fa9ff070..3e688942b2 100644
--- a/inst/tests/tests.Rraw
+++ b/inst/tests/tests.Rraw
@@ -1920,7 +1920,8 @@ test(670, DT[,z:=sum(y),keyby=x], ans)
 DT = data.table(x=1:2, y=1:6)
 test(671, DT[,z:=sum(y),keyby="x"], ans)
 DT = data.table(x=1:2, y=1:6)
-test(672, DT[,z:=sum(y),keyby=x%%2], data.table(x=1:2,y=1:6,z=c(9L,12L)), warning=":= keyby not straightforward character column names or list() of column names, treating as a by")
+test(672, DT[,z:=sum(y),keyby=x%%2], data.table(x=1:2,y=1:6,z=c(9L,12L)),
+          warning="The setkey() normally performed by keyby= has been skipped (as if by= was used) because := is being used together with keyby= but the keyby= contains some expressions. To avoid this warning, use by= instead, or provide existing column names to keyby=")
 DT = data.table(x=1:2, y=1:6)
 test(673, DT[,z:=sum(y),by=x%%2], data.table(x=1:2,y=1:6,z=c(9L,12L)))
 DT = data.table(x=1:2, y=1:6)