The present experimental study examined whether a different focus during a pre-task planning observation task affects learners' subsequent oral task performance. Forty-eight ninth-grade students learning German as a foreign language were randomly assigned to two different planning conditions: video observations with a focus on language (FonL) and video observations with a focus on content (FonC). With a communicative oral task we measured the effects on oral task performance, in terms of attempted (accurate) use of the target structure and complexity, in terms of number of words, subordination and coordination. In addition we investigated whether there was a trade-off between attempted (accurate) use of the target structure and complexity. Results showed that the focus of the observations at the pre-task stage did indeed lead to different outcomes: students in the language condition used the grammatical target structure more often and more accurately, whereas students of the content condition generated more coordinate and subordinate clauses. Trade-offs were found between attempted (accurate) use of the target structure and the use of subordinate clauses. These findings imply that, depending on the purposes of the lesson, the observation of peer-model videos with different planning foci can be effectively used to promote (accurate) use of targeted grammatical structures and improve complexity during subsequent task performance.