Evidence comes in all shapes, sizes and strengths.
Marcello Truzzi is credited with saying "Extraordinary claims require extraordinary proof". Carl Sagan popularised this with a slight change to "Extraordinary claims require extraordinary evidence".
Quality of Evidence is a Spectrum
The whole idea of evidence presents itself as a spectrum. A single eyewitness account is evidence, but not very strong evidence given how much we now know about the fallibility of human memory. The very good podcast Skeptoid offers a very clear and easy to understand summary of this topic.
Multiple eyewitness accounts of similar events offers slightly stronger evidence, but these are still way down on the scale when it comes to actual proof. If one person can err, so can many. Some will claim that "so many people can't be wrong, some of them must be right." Well actually no, that's not true. Turn this around and say "if one person can be wrong, so can all of them."
A little more credibility can be afforded to multiple eyewitness accounts of a single event. However even this does not present as hard evidence. It's very possible that many people viewing a strange event could all be fooled by the same effect. If the viewers are near each other they can all be influenced by the comments and actions of those around them to form a cohesive sounding depiction of the event. This depiction could be more a result of influence than of what was actually seen. Further, whether the viewers were close together or even separated by great distance we come back to fallibility of memory problem. Each time one of the witnesses recalls the event, new information can be unintentionally injected. "Oh you thought the light was blue did you? I thought it was green. It could have been blue I suppose." The next time this same witness recalls the event, "It was definitely a blue light that I saw".
Photographic evidence and video evidence certainly help in any enquiry, but in the 21st century these types of evidence still cast a lot of doubt. Year by year photographic and video evidence become much more abundant, but so do our abilities to fake these media. Again you can counter the credulous commentator's "with so many good quality videos on Youtube, not all of them can be faked" with a simple "Yes they can. If one can be faked, they can all be faked". This is in no way to say that all videos on Youtube showing unusual phenomena are fakes, just that the abundance of these videos is not hard evidence in itself.
Physical evidence that is made available for all to see and available for testing is the strongest form of evidence.
The willingness of people to accept arguments or phenomena with varying degrees of evidence is also a broad spectrum. The truly credulous will believe something based solely on a story told to them by their neighbour. An overly sceptical person won't accept that man has landed on the moon despite the abundance of evidence that we have. Somewhere in between these two extremes is a healthy sceptical outlook where evidence is weighted, sorted and the most probable event or idea is taken as probably true. The strength of acceptance of this reality is maintained until more evidence is presented, at which time the probability of the truth is re-evaluated.