This could turn into quite a long post so bear with me. I work as a PPC manager for an online marketing company and we use a third party bid management tool for tracking conversions, profit etc. Unfortunately none of the built in bid management strategies fit our need so we are trying to figure out a new one. Now we buy in on a CPC basis obviously but we also get out revenue on a CPC basis from our client deals, all on a single website. There is virtually zero time lag between the inbound click and the conversion. But the number of clients we have and the CPCs they pay us vary on a regular basis so any strategy has to be reactive to these changes. I am working on an idea that basically tracks every single click in and any subsequent conversions then works on the basis that once you have sufficient data you can work out a mean revenue per click. This then allows a suitable bid level to be calculated. My question then is: what sample size do you think would be required for a statistically significant conclusion to be acquired? The strategy would then work off of this as a minimum with any more clicks simply adding to the accuracy. Thanks in advance for your views.