The processes leading to the birth of high-mass stars are poorly understood. We characterise here a sample of 430 massive clumps from the ATLASGAL survey, which are representative of different evolutionary stages. To establish a census of molecular tracers of their evolution we performed an unbiased spectral line survey covering the 3-mm atmospheric window between 84-117 GHz with the IRAM 30m. A smaller sample of 128 clumps has been observed in the SiO (5-4) transition with the APEX telescope to complement the SiO (2-1) line and probe the excitation conditions of the emitting gas, which is the main focus of the current study. We report a high detection rate of >75% of the SiO (2-1) line and a >90% detection rate from the dedicated follow-ups in the (5-4) transition. The SiO (2-1) line with broad line profiles and high detection rates, is a powerful probe of star formation activity, while the ubiquitous detection of SiO in all evolutionary stages suggests a continuous star formation process in massive clumps. We find a large fraction of infrared-quiet clumps to exhibit SiO emission, the majority of them only showing a low-velocity component (FWHM~5-6 km/s) centred at the rest velocity of the clump. In the current picture, where this is attributed to low-velocity shocks from cloud-cloud collisions, this can be used to pinpoint the youngest, thus, likely prestellar massive structures. Based on the line ratio of the (5-4) to the (2-1) line, our study reveals a trend of changing excitation conditions that lead to brighter emission in the (5-4) line towards more evolved sources. Our analysis delivers a more robust estimate of SiO column density and abundance than previous studies and questions the decrease of jet activity in massive clumps as a function of age.