It depends on your specific case. The key factors are how strongly qualified your incoming traffic is, what the risk to value proposition is, and what obstacles occur in your conversion funnel.
Attrition rates upwards of 70% are probably normal - sometimes per step in your funnel. You're going to lose a bunch to gotchas which the user didn't see before clicking through, or to people who got diverted due to the duration, or to people who are intending to come back later, and so forth. If your overall visit to sub rate is stable at 25%, that may be good - depending on the traffic source, on the type of product or service, on whether that follows through to later conversion stages which generate lifetime value, and so forth.
Abandon rates go way up if the process is unnecessarily lengthy, or with longer forms, or with any of numerous factors which decrease trust or confidence, or if you need to ask for financial or other personal information. Sometimes it's critical to capture this information up front - e.g. "first paid month" results may be higher if you capture payment info up front, in spite of it causing a reduced conversion rate at the sign-up stage of the funnel.
Practically, not only are rates very case-specific, but you should be doing A/B testing to improve your rates combined with user testing to understand what factors are affecting user behaviors. You should also be looking at information at a more granular level than successful signups out of total visits, e.g. at what point in the process do they leave vs. what was the apparent traffic source.
Any time you make a change, you should be able to measure what effects there were at what stage of the process, and then estimate what the effect on your revenue will be. And ideally, you will run those changes against test groups before changing your live site for all incoming traffic. Done well, revenue goes up while keeping risks limited.
