As suggested by others, I would also recommend going one step further and turn this into a proper experiment, e.g.: - systematically record how often this happ

As suggested by others, I would also recommend going one step further and turn this into a proper experiment, e.g.: - systematically record how often this happens, in contrast to other ads, and for each of the reason topics - record for each random topic if you have also mentioned it anywhere else, e.g. in a Google search it some other digital media - make the choice of random topics more random, ie. not depending on current moods (which might be biased through subtle, external nudges -... These are of course just pointers, and by no means a proper experimental setup. I'm aware that this might take the fun out of your playful approach. However, you might be surprised by the results, in whatever direction. Also, it would give you a much more grounded fundament for further discussion. Of course you can just keep doing it the current, less tedious way. I'm only suggesting it because you seem to be interested in the topic and it might be more satisfying for yourselves to turn this into a little citizen science project.