[step:Identify the covers in the product poset]Write $a\lessdot_P b$ to mean that $b$ covers $a$ in $P$, and write $u\lessdot_Q v$ to mean that $v$ covers $u$ in $Q$. We claim that, for $(p,r),(p',r')\in P\times Q$,
\begin{align*}
(p,r)\lessdot_{P\times Q}(p',r')
\end{align*}
if and only if exactly one of the following two alternatives holds:
\begin{align*}
p\lessdot_P p' \text{ and } r=r',
\end{align*}
or
\begin{align*}
p=p' \text{ and } r\lessdot_Q r'.
\end{align*}
Suppose first that $(p,r)\lessdot_{P\times Q}(p',r')$. Since $(p,r)<(p',r')$, we have $p\le_P p'$ and $r\le_Q r'$, with at least one strict inequality. If both inequalities were strict, then
\begin{align*}
(p,r)<(p',r)<(p',r')
\end{align*}
would be a strict intermediate element of $P\times Q$, contradicting that $(p',r')$ covers $(p,r)$. Hence exactly one coordinate changes. If $r=r'$, then $p<_P p'$. If there were $s\in P$ with $p<_P s<_P p'$, then
\begin{align*}
(p,r)<(s,r)<(p',r)
\end{align*}
would be an intermediate element in $P\times Q$, again impossible. Thus $p\lessdot_P p'$. If instead $p=p'$, then $r<_Q r'$. If there were $s\in Q$ with $r<_Q s<_Q r'$, then
\begin{align*}
(p,r)<(p,s)<(p,r')
\end{align*}
would be an intermediate element in $P\times Q$, again impossible. Thus $r\lessdot_Q r'$.
Conversely, suppose $p\lessdot_P p'$ and $r=r'$. If
\begin{align*}
(p,r)\le (a,b)\le (p',r)
\end{align*}
for some $(a,b)\in P\times Q$, then $p\le_P a\le_P p'$ and $r\le_Q b\le_Q r$. Hence $b=r$, and since $p\lessdot_P p'$, either $a=p$ or $a=p'$. Therefore $(a,b)$ is either $(p,r)$ or $(p',r)$, so $(p,r)\lessdot_{P\times Q}(p',r)$.
Now suppose $p=p'$ and $r\lessdot_Q r'$. If
\begin{align*}
(p,r)\le (a,b)\le (p,r')
\end{align*}
for some $(a,b)\in P\times Q$, then $p\le_P a\le_P p$ and $r\le_Q b\le_Q r'$. Hence $a=p$, and since $r\lessdot_Q r'$, either $b=r$ or $b=r'$. Therefore $(a,b)$ is either $(p,r)$ or $(p,r')$, so $(p,r)\lessdot_{P\times Q}(p,r')$.[/step]