Aims. We study the chemical evolution of galaxy clusters by measuring the iron mass in the ICM after dissecting the abundance profiles into different components. Methods. We use Chandra archival observations of 186 morphologically regular clusters in the redshift range [0.04, 1.07]. For each cluster we compute the azimuthally-averaged iron abundance and gas density profiles. In particular, we aim at identifying a central peak in the iron distribution, associated with the central galaxy, and an approximately constant plateau reaching the largest observed radii, possibly associated with early enrichment occurred before and/or shortly after the virialization of the cluster. We are able to firmly identify the two components in the iron distribution in a significant fraction of the sample, simply relying on the fit of the iron abundance profile. From the abundance and ICM density profiles we compute the iron mass included in the iron peak and iron plateau, and the gas mass-weighted iron abundance of the ICM, out to an extraction radius of 0.4 r 500 and, extending the abundance profile as a constant, to r 500 . Results. We find that the iron plateau shows no evolution with redshift. On the other hand, we find marginal (< 2σ c.l.) decrease with redshift in the iron mass included in the iron peak rescaled by the gas mass. We measure that the fraction of iron peak mass is typically a few percent (∼ 1%) of the total iron mass within r 500 . Therefore, since the total iron mass budget is dominated by the plateau, we find consistently that the global gas mass-weighted iron abundance does not evolve significantly across our sample. We are also able to reproduce past claims of evolution in the global iron abundance, which turn out to be due to the use of cluster samples with different selection methods combined to the use of emission-weighted instead of gas mass-weighted abundance values. Finally, while the intrinsic scatter in the iron plateau mass is consistent with zero, the iron peak mass exhibits a large scatter, in line with the fact that the peak is produced after the virialization of the halo and depends on the formation history of the hosting cool core and the strength of the associated feedback processes. Conclusions. We conclude that only a spatially-resolved approach can resolve the issue of the iron abundance evolution in the ICM, reconciling the contradictory results obtained in the last ten years. Evolutionary effects below z ∼ 1 are marginally measurable with present-day data, while at z > 1 the constraints are severely limited by the poor knowledge of the high-z cluster population. The path towards a full and comprehensive chemical history of the ICM necessarily requires the use of high-angular resolution X-ray bolometers and a dramatic increase in the statistics of faint, extended X-ray sources.