A more precise determination of the effective fine structure constant α eff (E) is mandatory for confronting data from future precision experiments with precise SM predictions. Higher precision would help a lot in monitoring new physics by increasing the significance of any deviation from theory. At a future e + e − -collider like the ILC, as at LEP before, α eff (E) plays the role the static zero momentum α = α eff (0) plays in low energy physics. However, by going to the effective version of α one loses about a factor 2 × 10 2 at E = mµ to 10 5 at E = MZ in precision, such that for physics at the gauge boson mass scale and beyond α eff (E) is the least known basic parameter, about a factor 20 less precise than the neutral gauge boson mass MZ and by about a factor 60 less precise than the Fermi constant GF . Examples of precision limitations are α eff (mµ) which limits the theoretical precision of the muon anomalous magnetic moment aµ and α eff (MZ) which limits the accuracy of the prediction of the weak mixing parameter sin 2 Θ f and indirectly the upper bound on the Higgs mass mH. An optimal exploitation of a future linear collider for precision physics requires an improvement of the precision of α eff (E) by something like a factor ten. We discuss a strategy which should be able to reach this goal by appropriate efforts in performing dedicated measurements of σ hadronic in a wide energy range as well as efforts in theory and in particular improving the precision of the QCD parameters αs, mc and m b by lattice QCD and/or more precise determinations of them by experiments and perturbative QCD efforts. Projects at VEPP-2000 (Novosibirsk) and DANAE/KLOE-2 (Frascati) are particularly important for improving on α eff (MZ) as well as α eff (mµ). Using the Adler function as a monitor, one observes that we may obtain the hadronic shift ∆αpQCD where the first term includes the full non-perturbative part with the choice s0 = (2.5 GeV) 2 or larger. In such a determination low-energy machines play a particularly important role in the improvement program. We present an up-to-date analysis including the recent data from KLOE, SND, CMD-2 and BABAR. The analysis based on e + e − -data yields ∆α