> If you spend $600k on four days of advertising anything using any medium (and then quit), the odds are that you will not meet your goals. It doesn't matter if

> If you spend $600k on four days of advertising anything using any medium (and then quit), the odds are that you will not meet your goals. It doesn't matter if you spend $600k on one thirty second television slot, a full page ad in the newspaper, or 10 million Facebook ads. I disagree with you based on personal experience. I know several companies with an online marketing cost per order of say (number fuzzed just in case someone links me, but order of magnitude), <$5 and average order size of $100 and gross margin of 40%, and average time between entry and purchase of an hour or two. I've heard several instances at these companies of someone saying "let's make our end of month numbers look better" and subsequently blowing extra cash (sometimes millions) on online advertising with immediate results, and little change in company CPO. The budget is literally a function of what capital is at hand to be allocated to marketing. E.g. you don't want to grow too fast or you get operational issues, so you cut marketing spend a bit and sales slow down a bit so the warehouse can catch up. At a (very high) CPC of $1 those $600,000 are up to 600,000 visitors... that's a lot of data to think about. Spending it in chunks, optimizing for ROI instead of fluffy "KPIs", and doing a bit of stats, is usually enough. As the original article effectively says. What is true is that how "X" is doing (where X can be any marketing platform, channel, magic solution sold by slick suited, Rolex wearing sales guys) is impossible to generalize. I was fortunate enough to see the same methodology used in around 40 countries around the world and yield vastly different results: sometimes Facebook display ads are your best channel, sometimes they cost you 100x what you make back; sometimes Google is amazing, other times display retargeting is it. You should always try everything and make decisions once you have enough data. And sometimes it even changes after a few years: e.g. a country which goes from blackberries to large screen touch phones with massive CPUs is going to see an improvement in display ads. So you keep dipping just in case. It's a high dimensional problem, further complicated by the quality of your own team (and evolution of the product) - as Ogilvy wrote at length, the intrinsic quality of the ad goes a very long way towards its success.