We explore the behavior in time of the energy exchange between a
system of interest and its environment, together with its relationship
to the non-Markovianity of the system dynamics. In order to evaluate
the energy exchange we rely on the full counting statistics formalism,
which we use to evaluate the first moment of its probability
distribution. We focus in particular on the energy backflow from
environment to system, to which we associate a suitable condition and
quantifier, which enables us to draw a connection with a recently
introduced notion of non-Markovianity based on information backflow.
This quantifier is then studied in detail in the case of the spin-boson
model, described within a second-order time-convolutionless
approximation, observing that non-Markovianity allows for the
observation of energy backflow. This analysis allows us to identify the
parameters region in which energy backflow is higher.