diff --git a/README.md b/README.md
index 50634f6..1d127a5 100644
--- a/README.md
+++ b/README.md
@@ -371,6 +371,13 @@ Section creation order:
 1. ✅ `idx_*`. requires offsets to `groups` and `users`.
 1. ✅ Header.
 
+For v2
+------
+
+These are desired for the next DB format:
+- Compress strings with fsst.
+- Trim first 4 bytes from the cmph headers.
+
 [git-subtrac]: https://apenwarr.ca/log/20191109
 [cmph]: http://cmph.sourceforge.net/
 [id]: https://linux.die.net/man/1/id
diff --git a/src/bdz.zig b/src/bdz.zig
index 2701b88..40e8d17 100644
--- a/src/bdz.zig
+++ b/src/bdz.zig
@@ -4,13 +4,13 @@ extern fn bdz_search_packed(packed_mphf: [*]const u8, key: [*]const u8, len: c_u
 
 pub fn search(packed_mphf: []const u8, key: []const u8) u32 {
     const len = std.math.cast(c_uint, key.len).?;
-    return @as(u32, bdz_search_packed(packed_mphf.ptr, key.ptr, len));
+    return @as(u32, bdz_search_packed(packed_mphf[4..].ptr, key.ptr, len));
 }
 
 const u32len = 5;
 
 pub fn search_u32(packed_mphf: []const u8, key: u32) u32 {
-    return @as(u32, bdz_search_packed(packed_mphf.ptr, &unzero(key), u32len));
+    return @as(u32, bdz_search_packed(packed_mphf[4..].ptr, &unzero(key), u32len));
 }
 
 // encode a u32 to 5 bytes so no bytes is a '\0'.
diff --git a/src/cmph.zig b/src/cmph.zig
index b6a5f8b..965370a 100644
--- a/src/cmph.zig
+++ b/src/cmph.zig
@@ -39,7 +39,7 @@ pub fn pack(allocator: Allocator, input: [][*:0]const u8) error{OutOfMemory}![]c
     errdefer allocator.free(buf);
     cmph_pack(mph, buf.ptr);
     cmph_destroy(mph);
-    return buf[4..];
+    return buf;
 }
 
 // perfect-hash a list of numbers and return the packed mphf