I first read that idea from the Science-based Medicine blog. Here's an example:

We did not find that placebo interventions have important clinical effects in general. However, in certain settings placebo interventions can influence patient-reported outcomes, especially pain and nausea, though it is difficult to distinguish patient-reported effects of placebo from biased reporting. The effect on pain varied, even among trials with low risk of bias, from negligible to clinically important. Variations in the effect of placebo were partly explained by variations in how trials were conducted and how patients were informed.

Let’s break this down a bit. First, they found that when you look at any objective or clinically important outcome – the kinds of things that would indicate a real biological effect – there is no discernible placebo effect. There is no mind-over-matter self healing that can be attributed to the placebo effect.

What the authors found is also most compatible with the hypothesis that placebo effects, as measured in clinical trials, are mostly due to bias. Specifically, significant placebo effects were found only for subjectively reported symptoms. Further, the size of this effect varied widely among trials.