Set $x_1 = x$ and $x_{n+1} = B x_n$, where $x ∈ \{B, C, K, W\}$; and note that - up to η-equivalence, $x_n = x_{n+1} I$, since $B f I = f = B I f$ under η-equivalence. We can then also write $x_0 = x I$. The combinators $\{B, C, K, W\}$ are given by the rules
$$I x = x,\quad B x y z = x (y z),\quad C x y z = x z y,\quad W x y = x y y,\quad K x y = x.$$
Up to η-equivalence $B_0 = B I = I$.
Instead of explaining the general case, it should suffice to see an example:
$$\begin{align}
x v (z y) (y x)
&= I x v (z y) (y x)\\
&= B_4 I x v (z y) y x\\
&= B_3 (B_4 I) x v z y y x\\
&= C_3 X_b x v y z y x\\
&= C_4 (C_3 X_b) x v y y z x\\
&= C_5 (C_4 (C_3 X_b)) x v y y x z\\
&= C_4 (C_5 (C_4 (C_3 X_b))) x v y x y z\\
&= C_3 (C_4 (C_5 (C_4 (C_3 X_b)))) x v x y y z\\
&= C_1 (C_3 (C_4 (C_5 (C_4 (C_3 X_b))))) v x x y y z\\
&= W_4 X_c v x x y z\\
&= W_2 (W_4 X_c) v x y z\\
&= K_2 X_w v w x y z\\
&= X_k v w x y z,
\end{align}$$
where
$$
X_b = B_3 (B_4 I) = B_3 B_3,\quad X_c = C_1 (C_3 (C_4 (C_5 (C_4 (C_3 X_b))))),\\
X_w = W_2 (W_4 X_c),\quad X_k = K_2 X_w.
$$
The $X_b$ sub-term applies to your question. The remainder applies to the more general question that you didn't ask.
You could also write it as
$$K_2 (W_2 (C_1 (W_4 (C_3 (B_3 (C_4 (B_4 C_2))))))) v w x y z = x v (z y) (y x).$$
Edit: Before I forget, your example:
$$\begin{align}
a(bcd)(ef)
&= Ia(bcd)(ef)\\
&= B_3Ia(bcd)ef\\
&= B_2(B_3I)a(bc)def\\
&= B_2(B_2(B_3I))abcdef\\
&= B_2(B_2B_2)abcdef
\end{align}$$