When the media covers surveys, they are often focused on the so-called horse race. In a parliamentary system, they would for example ask what we call the "Sunday question": If the election were this Sunday, what party would win? The way we use surveys in election campaigns is not to predict the result, but with the following purpose:

• A survey should tell you where you really stand in terms of awareness and vote. There is a science to surveys and it is important to get the numbers straight. This might not always be pleasant information, but at least you can act accordingly. It might also save you a lot of money and sweat. It also helps to check on your own people and to track progress. If you don't know where you have started, how do you know you're making progress? And how do you know you're making enough progress?
• It should indicate the potential to grow your candidacy and to define low-hanging fruits in terms of target voters. This helps you spend a little as possible, and to spend wisely, instead of wasting precious resources. Once your vote share grows, it will help you get important endorsements and media coverage.
• A survey (and qualitative research) should serve as a basis to make an informed decision about your overall campaign message and strategy. There is this myth that candidates should raise awareness first, and then convert it to vote intention later on. I strongly disagree: awareness without conversion is an unguided missile. Noise that is empty of content might turn voters off.
• It should help test campaign ads and materials. In most countries, the expensive part of a campaign is the time buy, so it only makes sense to independently test ads before running them. Conceptually, I think that many ads are the same. They look and feel like, well, political ads. I'm also skeptical about ads that are limited to jingles and endorsements, but lack specific, credible content that is based on research and matches the political demand.