Context. In solar-like oscillators, acoustic waves are excited by turbulent motion in the convective envelope and propagate inward, generating a variety of standing pressure modes (p modes). When combining the power of several solar acoustic modes, some studies have reported an excess that is not compatible with pure stochastic excitation. This excess could be a signature of a second mode excitation source. Aims. With over 27 years of helioseismic data from the Sun-as-a-star observations by the Solar and Heliospheric Observatory (SoHO), we aim to study the variation in mode energy over this period, covering solar cycles 23 and 24, as well as the beginning of cycle 25. In particular, we focus on the possible sources of high peaks in the mode-energy time series, namely, instrumental problems, or other exciting mechanisms, such as flares or coronal mass ejections (CMEs). Methods. We reconstructed the energy time series for each mode, resulting in 36 time series with a sampling time of 1.45 days. By combining the small timescale variations in energy for several low-degree modes (ℓ ≤ 2) in the 2090–3710 μHz range, we were able to study the correlation between the modes and their compatibility with the hypothesis that modes are only stochastically excited by convection. Results. The observed excitation rate significantly deviates from what would be expected in the case of a purely stochastic excitation. Our results indicate that this energy excess cannot solely be attributed to instrumental effects and it does not exhibit a cyclic variation. Although high-energy excesses are occasionally associated with observations of flares and/or CMEs, no consistent pattern could be identified. The excitation is slightly more frequent for modes probing the upper layer of the convective zone. Furthermore, the energy supply rate seems to vary over time with the mean value following a modulation that can match the quasi-biennial oscillation (QBO) observed in other solar indicators, with the variance shown to be anti-correlated with the cycle.