If I were to give an example, it would be that Bob says he's receiving calls in the middle of the night. They're harassing calls. They are from Alice. He is able to show me a phone number that he received the calls from. Now, I have no confirmation quite yet that a crime has been committed because I haven't heard the calls that have gone through, but I do have a phone number, so I suspect that an offence may have been committed.
I can reach out to a service provider and ask if they provide services to this particular phone number. If they respond yes, then I can move on, in this case, to a subscriber information production order because I'm still not quite sure that an offence has been committed, or it's not probable. It's possible, but it's not probable yet. If I receive a response that confirms that this account belongs to Alice, I start to increase my grounds, because Bob says that Alice was the one who did it, and this account happens to come back to Alice. I'd have to do some additional corroboration, but we're on our way to building grounds to believe. It's a step-by-step process.
