Is it possible to get a tight lower bound for $k$ (in terms of $q$), if the following inequality holds? $$\frac{q^2 - 1}{q^2} < \frac{q^{k+1}}{q^{k+1} + 1}$$ Here, $q$ is a prime number, and $k$ is a positive integer.
This question arises out of considerations surrounding this answer to a closely related MSE question.
Here are some data on the corresponding lower bounds for $k$ obtained from this inequality on various primes $q$:
$$ \begin{array}{|c|c|c|} \hline q& \text{Lower bound for } k & \text{Approximate value of lower bound} \\ \hline 2& \log_2(3/2)& 0.584963 \\ \hline 3& \log_3(8/3)& 0.892789 \\ \hline 5& \log_5(24/5)& 0.974636 \\ \hline 7& \log_7(48/7)& 0.989404 \\ \hline 11& \log_{11}(120/11)& 0.996539 \\ \hline 13& \log_{13}(168/13)& 0.997686 \\ \hline \end{array} $$
(I used WolframAlpha to manually pull these data one by one.)
From the first few values on the lower bounds for $k$ obtained from the inequality by plugging in the values of the first six primes, I hereby conjecture that:
CONJECTURE: If the following inequality holds $$\frac{q^2 - 1}{q^2} < \frac{q^{k+1}}{q^{k+1} + 1},$$ where $q$ is a prime number and $k$ is a positive integer, then $$k > \log_{q}(q - \frac{1}{q}).$$
Here are my:
QUESTIONS: Does this conjecture hold for all primes $q$? If so, can you prove it?