diff --git a/best-practices.md b/best-practices.md
new file mode 100644
index 0000000..5297947
--- /dev/null
+++ b/best-practices.md
@@ -0,0 +1,145 @@
+### let expressions
+
+You _should_ put `let` before each variable inside a `do` block.
+But beware of name shadowing (though compiler can help with it).
+
+```haskell
+foo = do
+  let x   = 10
+  let f 1 = 5
+      f _ = 0  -- possible shadowing here with let
+  return $ x + f 2
+```
+
+_WARNING_: try to avoid aggressive autoderiving because
+[it can slow down compilation](http://www.stephendiehl.com/posts/production.html).
+
+> Deriving instances of Read/Show/Data/Generic for largely recursive ADTs can
+> sometimes lead to quadratic memory behavior when the nesting gets deep.
+
+You _should not_ use `() <$` to ignore the result of function, because it less legible
+than `_ <-` or `void $`.
+
+```haskell
+foo = do
+  _ <- sendCert ourVssCertificate  -- cannot be used as last statement
+  void $ sendTransactionAndReport 1 "tx42"
+```
+
+### ApplicativeDo
+
+Use `-XApplicativeDo` only when necessary.
+
+You _should_ use `-XApplicativeDo` when you are writing a parser of CLI arguments with
+`optparse-applicative` because it is very easy to mess up the order of the arguments.
+However, be aware of
+[some non-obvious behavior](https://www.reddit.com/r/haskell/comments/7679g8/ghc_821_applicativedo_possible_bug/)
+when you use pattern-matching inside `do`-blocks.
+
+## Dealing with laziness
+
+By default, use strict data types and lazy functions.
+
+### Data types
+
+You _should_ enable [`StrictData`](https://downloads.haskell.org/~ghc/latest/docs/html/users_guide/glasgow_exts.html#extension-StrictData)
+globally for your project, marking fields explicitly with `~` when you need laziness.
+
+```haskell
+{-# LANGUAGE StrictData #-}
+
+-- Good
+data Point = Point
+  { pointX :: Double  -- ^ X coordinate
+  , pointY :: Double  -- ^ Y coordinate
+  }
+
+-- OK, but needs justification
+data Point = Point
+  { pointX :: ~Double  -- ^ X coordinate
+  , pointY :: ~Double  -- ^ Y coordinate
+  }
+```
+
+Additionally, unpacking simple fields often improves performance and
+reduces memory usage:
+
+```haskell
+data Point = Point
+  { pointX :: {-# UNPACK #-} Double  -- ^ X coordinate
+  , pointY :: {-# UNPACK #-} Double  -- ^ Y coordinate
+  }
+```
+
+As an alternative to the `UNPACK` pragma, you can put
+
+```haskell
+{-# OPTIONS_GHC -funbox-strict-fields #-}
+```
+
+at the top of the file.  Including this flag in the file itself instead
+of e.g. in the Cabal file is preferable as the optimization will be
+applied even if someone compiles the file using other means (i.e. the
+optimization is attached to the source code it belongs to).
+
+Note that `-funbox-strict-fields` applies to all strict fields, not
+just small fields (e.g. `Double` or `Int`). If you are using GHC 7.4 or
+later you can use `NOUNPACK` to selectively opt-out for the unpacking
+enabled by `-funbox-strict-fields`.
+
+For backwards compatibility, you _may_ choose to keep `StrictData` off globally, but you are
+strongly advised not to. If you choose not to enable it, you _should_ mark all record
+fields as strict, unless there is an explicit reason to make them lazy.
+
+```haskell
+-- Good
+data Point = Point
+  { pointX :: !Double  -- ^ X coordinate
+  , pointY :: !Double  -- ^ Y coordinate
+  }
+
+-- Needs justification for implicit laziness
+data Point = Point
+  { pointX :: Double  -- ^ X coordinate
+  , pointY :: Double  -- ^ Y coordinate
+  }
+```
+
+### Functions
+
+Have function arguments be lazy unless you explicitly need them to be
+strict.
+
+The most common case when you need strict function arguments is in
+recursion with an accumulator:
+
+```haskell
+mysum :: [Int] -> Int
+mysum = go 0
+  where
+    go !acc []    = acc
+    go acc (x:xs) = go (acc + x) xs
+```
+
+### Warnings
+
+ The following GHC options _should_ be enabled globally for your project    
+([source](https://medium.com/mercury-bank/enable-all-the-warnings-a0517bc081c3)):   
+
+ ```    
+-Weverything    
+-Wno-missing-exported-signatures    
+-Wno-missing-import-lists   
+-Wno-missed-specialisations 
+-Wno-all-missed-specialisations 
+-Wno-unsafe 
+-Wno-safe   
+-Wno-missing-local-signatures   
+-Wno-monomorphism-restriction   
+-Wno-implicit-prelude   
+-Werror 
+``` 
+
+ A successful build _must_ not trigger any of the enabled warnings. 
+
+ You _may_ use `{-# OPTIONS_GHC -fno-warn-orphans #-}` on a per-module basis.
diff --git a/haskell.md b/haskell.md
index 51b0f5c..44c2b53 100644
--- a/haskell.md
+++ b/haskell.md
@@ -568,6 +568,12 @@ group independently.
 You *may* use [`weeder`][weeder] to detect
 unused dependencies and exported modules.
 
+## Best practices
+
+Some of the things that were previously located in this guide were deemed to be
+best practices rather than style guidelines and moved
+[elsewhere](best-practices.md).
+
 [stylish-haskell]: https://github.com/serokell/metatemplates/blob/master/templates/haskell/.stylish-haskell.yaml
 [hlint]: https://github.com/serokell/metatemplates/blob/master/templates/haskell/.hlint.yaml
 [editorconfig]: https://github.com/serokell/metatemplates/blob/master/templates/haskell/.editorconfig