One survey is useless for prediction. Surveys are an instant photo of a particular and defined population. You need several surveys(same population, same sampli

One survey is useless for prediction. Surveys are an instant photo of a particular and defined population. You need several surveys(same population, same sampling method) to find a trend. And even if you find a trend in the data, if the situation change(google start to charge by search, bing gives out kittens and rainbows for each search, etc.) your model/trend isn't any good anymore. Surveys are a tool, they don't confirm anything and they don't give answers. If you base your decisions only on surveys you're a moron.