We present a new, simple method to predict activity-induced radial velocity variations using high-precision time-series photometry. It is based on insights from a simple spot model, has only two free parameters (one of which can be estimated from the light curve) and does not require knowledge of the stellar rotation period. We test the method on simulated data and illustrate its performance by applying it to MOST/SOPHIE observations of the planet host-star HD189733, where it gives almost identical results to much more sophisticated, but highly degenerate models, and synthetic data for the Sun, where we demonstrate that it can reproduce variations well below the m/s level. We also apply it to Quarter 1 data for Kepler transit candidate host stars, where it can be used to estimate RV variations down to the 2-3m/s level, and show that RV amplitudes above that level may be expected for approximately two thirds of the candidates we examined.