At page 303 in the book Quantum Field Theory for the Gifted Amateur by Blundell and Lancaster, they argue that the renormalization group equation for the coupling $\lambda$ in $\phi^4$ theory can be derived as follows. Firstly, we introduce the renormalized coupling $$\lambda_R(s_0) = \lambda +\lambda^2 C \ln \left( \frac{s_0}{\Lambda^2}\right) + \ldots , $$ and rewrite the matrix element in terms of $\lambda_R$: $$ iM = - \lambda_R(s_0) -C \ln \left( \frac{s }{s_0}\right) \lambda_R^2(s_0) + \ldots \tag{1} $$ For a different scale $s_1$, we find $$\lambda_R(s_1) = \lambda +\lambda^2 C \ln \left( \frac{s_1}{\Lambda^2}\right) + \ldots , $$ $$ iM = - \lambda_R(s_1) -C \ln \left( \frac{s }{s_1}\right) \lambda_R^2(s_1) + \ldots \tag{2} $$ They then argue that by "subtracting" it follows that $$\lambda_R(s_1) = \lambda_R(s_0) +C \ln \left( \frac{s_1 }{s_0}\right) \lambda_R^2(s_0) + \ldots \tag{3}$$ I'm failing to see why this should be true. If we subtract Eq. 2 from Eq. 1 we find a term of the form $$ -C \ln \left( \frac{s }{s_1}\right) \lambda_R^2(s_1)+C \ln \left( \frac{s_0}{\Lambda^2}\right)\lambda_R^2(s_0) $$ which is not equal to the one in Eq. 3 unless $\lambda_R^2(s_0) = \lambda_R^2(s_1)$.
Am I missing something or is the derivation shown here wrong?