A dataset from an online studies on a simulated social trading platform using a chatbot to give participants advice on investment. 64 participants interacted with a chatbot across 4 conditions: human-like/not human-like, and with reply suggestion buttons/without reply suggestion buttons embedded in the user interface. They were shown 10 different portfolios to follow or unfollow at 5 separate month intervals, basing their decision on the advice of the chatbot or a separate news feed that would try to predict the next change in portfolio value. Participants were assigned an initial virtual balance of £1000. Image tagging was included as a distracting secondary task. All the messages exchanged to and from the chatbot are included, as well as the user actions and image tagging. Participant demographic data included in a separate file.