Actually, come to think of it, in a freemium subscription model a 5% increase in conversion doesn't necessarily do 5% to revenues instantly because a portion of your revenues are produced by existing subscribers. Sorry -- I do one-time purchases, sometimes I forget about the economics for other folks.
Still: what other businesses can assign one junior employee to work on button designs for a week and walk away with 5% added to the value of the business? And close to risk-free, because you can experiment on a fraction of your customers and terminate it almost instantly if it doesn't pan out?
That is leverage, broadly defined, that would make hedge fund managers green with envy. ("Hey, can you figure out a way to make us 5% extra on all our capital? Oh yeah, and do it risk free. Oh yeah, and I'd like whatever you do to be basically repeatable with a minor variation next week, so that we can institutionalize this and do it as a process rather than a one-off. Alright, call me when you've made me a billion, I'll be on the golf course.")
Basically, trial -> purchase conversion for freemium seems to max out at around 1%, so your efforts are usually better focused on improving retention rather than initial conversion.
If anyone remembers what presentation this was from please post the link, I can't seem so find it.
The presentation it references for support of the "1% is the limit" myth is here: http://particletree.com/features/web-app-autopsy/ "Myth" is strong language for me, but I believe it is justified in this case. If you play "Hey I got a BA degree for a reason, let's play secondary researcher", it is based off of a misstatement of a repetition of a claim that doesn't even appear in the primary research. All the primary research says is "Three particular firms we have convenient access to say they have about 1% visitor -> purchase conversions. That's probably typical, then." (Even the primary source skips over the fact that one of the three actually reported 1.14%, which is a 14% difference that falls straight to the bottom line, and the other two reported to exactly one significant digit.)
You'll note that visitor -> purchase is very different than trial -> purchase, but don't beat yourself up, different folks routinely quote both numbers at 1%. It should be self evidently obvious that both groups cannot possibly be equally right, but as it turns out they're probably both just equally wrong.
In addition to being incorrectly repeated from a misreported summary of anecdotal data, the claim is empirically false. I can't tell you other people's numbers, but I can tell you mine: 2.45% trial to purchase conversion. Feel free to do much better than that -- it is certainly possible. (Important possible apples-to-oranges alert: my freemium app is not sold on a subscription basis.)
This post is not meant as a personal attack: I have heard this statistic, and its older brother ("1% of shareware downloads convert") for years. Vaporizing it in the harsh light of actual evidence is one of my numerous personal crusades, because it leads to poor decisions that actually cost people money, like "Well, conversion is a fixed number so I totally can't add 5% to my bottom line by working on buttons or something absolutely stupid with a shopping cart redesign."
* As you "launch" there is like 1% of conversion
* As you keep optimizing, conversion rate increases (in your case by 1.45%)
* Something like +1% per major optimization (and targeting) (compound rate ?)
Can you confirm it?
And don't forget that you can always hack the buyer psychologically.
This is my single gripe about this article. This isn't true: although there's a higher profit margin because the cost of one customer on a general web application is close to nil, the cost of acquiring a customer is sometimes pretty high. Especially if you factor in that 1% of free users will convert to paid users (in a freemium model.)
Unless if you hit it out of the ballpark with some crazy lucky viral marketing, I'm pretty sure you're going to have to do some solid marketing to gain traction.