The structure functions for deep-inelastic electron-proton scattering at small x are calculated using the QCD k(T)-factorization theorem together with the gluon distribution obtained from solving the Lipatov (or BFKL) equation. We study the theoretical uncertainties in the predictions and show, in the HERA regime, that the effective slope lambda, defined by F-2(x,Q(2))-F-2(bg)(x,Q(2))=Cx(-lambda), is a remarkably stable prediction of perturbative QCD, where F-2(bg) is suitably defined background. We propose a procedure for identifying the Lipatov effect from small x measurements of F-2. We apply it to the recent HERA data, and hence determine lambda.