diff --git a/README.md b/README.md
index ec4d04c..676c135 100644
--- a/README.md
+++ b/README.md
@@ -21,7 +21,7 @@ In order to recreate the server setup, you have to make the following steps:
     - `abaevdict_en` and `abaevdict_ru` contain the preprocessed, standardized English and Russian XML editions of HEDO. This is done because maintaining a single bilingual database is impractical: users will want to interact with the English and Russian versions only. At some point, they may even diverge. For basic linking, entry IDs should suffice.
     - `abaevdict_index` includes various helper documents that are not directly displayed, but used for faster lookup, and for REST API queries.
     
-4. Import the contents of the `entries` folder of [abaev-xml](https://code.cucurri.ru/abaevdict/abaev-xml) into the db `abaevdict`, and the file `biblio/abaev_biblio.xml` into the db `abaevdict_index`.
+4. Import the contents of the `entries` folder of [abaev-xml](https://code.cucurri.ru/abaevdict/abaev-xml) into the db `abaevdict`, and the file `biblio/abaev_biblio.xml` into the db `abaevdict_index`. Do not import them as subdirectories/collections; just put them under the database root.
 
 5. Execute the script `scripts/update-indices.bxs` from the GUI or CLI. This does the following things:
     - optimizes `abaevdict`;