So as we think about providing corroborating evidence, what sources should we involve? Well, a few decades ago, a study looked at exactly this question. Students were recruited for studying how people respond to audio presentations. They're directed to show up at a laboratory, and when they arrive, they sat down at a desk and told they would listen to audio tapes and make judgments about them. The experimenters behind the study were interested in what makes people laugh, and in particular, how laughter is shaped by social influence. They had people listen what's called a laugh track, essentially a stand-up comedian with people laughing in the background, and it included canned laughter. You might think what people find funny is totally subjective, but hearing pre-recorded merriment often help primes the pump. In this case, as expected, the laugh track helped. Listeners are more likely to laugh and smile to comedians jokes when they heard other people laughing. But beyond the mere presence of laughter, the scientists also manipulated something else, who people thought was laughing. In both cases, they listen to recordings with people laughing in the background, but in one case the scientists told them the laughters where people like them, other students from the same university. The other group of students were told that those laughters were quite different, members of political party with the students had no interest in identifying. Even though the laughter sounded exactly the same, same people laughing on the same laugh track, whoever the listeners thought was laughing shaped their reactions. When listeners thought the laughters we're not like them, the fact that people were laughing didn't matter, didn't change their behavior, didn't make them think the video is any funnier than they would if there was no one laughing at all. When people thought the laughters were like them, they changed their behavior, they laughed nearly four times as long. A great deal of research finds similar things that similarity matters. Someone like me thinks a joke is funny, I'll probably find it funny as well, but someone who isn't like me finds it funny, well, that doesn't provide as much information, doesn't give me as much information about my likely reaction, it's not as diagnostic about how I'll feel about listening to it. Looking at hotels and trip advisor, you don't want to know whether hotel is highly rated, you want to know what other people like you think about it. If you're a family traveling with two kids, you probably want to place it other families recommend. The fact that hip 22-year-olds like the hotel, well, that's not that useful, and the reverse is also true. You're hip 22-year-old and you're looking at hotels, affected families like it isn't something you necessarily want. You probably feel the same way, you want other people like you to provide those reviews. In other words, the translation problem is less of a problem when there's less of a need for translation, and the absence of another use, similar sources and the next closest thing. Sources dealing with the same issue or challenges, other people with the same needs, other companies in the same vertical. The more similar they are, the more proof or corroborating evidence they provide, and the greater their impact. Rather than trying to push prospects, smart companies often left their existing clients do the talking. They host events such as dinners, where in addition to hearing from thought leaders are sitting through demos, potential clients can interact with current clients to get an outside perspective on an unbiased view of what working with the companies like. When thinking about where to seat a prospect at dinner or how best to change their mind, it's important to make sure they're sitting next to similar others. Sit them next to an existing client from the same industry, or another one who's from a different industry but of similar size. Encourage them to talk to someone who's like them, but has some differences will give them more information. Sources that are similar enough but different from one other offer the perfect combination. Similarity makes the feedbacks seem diagnostic and relevant, but Independence increases the chance that each adds additional value rather than being seen as redundant.