I am currently trying to learn the two related concepts of the [Rao-Blackwell theorem][1] and the [Lehmann-Scheffé theorem][2]. My question relates to example 2.16 from [this document][3], and I also found much relevant information in [this document][4]. 

Assume we have the random sample $X_1, \dots, X_n$ with mean $\mu$ and variance $\sigma^2 < \infty$. We have that $E[S^2] = \sigma^2$, where $S^2 = \sum_{i = 1}^n \dfrac{(X_i - \bar{X})^2}{n - 1}$ and $\bar{X} = \sum_{i = 1}^n \dfrac{X_i}{n}$. Now assume the $X_i$ are Poisson random variables with parameter $\lambda$. [My understanding is that, using Lehmann-Scheffé][5], we get that $E[S^2 \mid \bar{X}] = \bar{X}$. Then, using the [law of total variance][6], we get that $\text{Var}(S^2) > \text{Var}(\bar{X})$. Based on what I've read, the two above theorems imply that, if we define, say, a sufficient statistic $T_1(\mathbf{X})$ and a complete sufficient statistic $T_2(\mathbf{X})$ for some parameter $\varphi$, then, under some circumstances, we can say that $\text{Var}(T_1(\mathbf{X})) > \text{Var}(T_2(\mathbf{X}))$. However, I'm having trouble understanding this last part. I've read over various notes on the subject, but I'm still not sure I understand what it's saying. Why can we say that $\text{Var}(T_1(\mathbf{X})) > \text{Var}(T_2(\mathbf{X}))$? And what are these 'circumstances' that make this inequality valid? 


  [1]: https://en.wikipedia.org/wiki/Rao%E2%80%93Blackwell_theorem
  [2]: https://en.wikipedia.org/wiki/Lehmann%E2%80%93Scheff%C3%A9_theorem
  [3]: http://www.maths.qmul.ac.uk/~bb/MS_NotesWeek10a.pdf
  [4]: http://www.baskent.edu.tr/~osezgin/LECTURE%2010.pdf
  [5]: https://math.stackexchange.com/q/2881768/356308
  [6]: https://en.wikipedia.org/wiki/Law_of_total_variance