Firstly, did you change the long url of your snipping at some time? Snip count is not automatically set to 0 if you edit the long url--there is a manual option to do this in Manage Snips--so what you think is the inflated count may be the counts accumulated between more than one urls.

Secondly, Snipurl is not intended to be a precision tool. We track all clicks to a snip from a registered user and update the counts every 10 minutes. It's a free service so it is prohibitively resource-intensive (and costly) to provide truly live services. Not even Google does.

That said, I have found snip counts to be fairly representative of actual clicks. 

Please note that your Google data is certainly not as accurate as you seem to think. Google analytics, like the free tool Urchin before it (which Google acquired), relies on Javascript. 

Many savvy users have JS-inhibiting tools in their browser (such as Adblock or NoScript in Firefox) which render the Google Analytics tracker utterly useless. You are not getting any data from these users. 

For example, if you use Snipurl in a Twitter context, your audience is by definition savvier than the regular Internet folk, which means it is fair to assume that these users are familiar with ad and JS blocking tools.

To be honest, the only truly accurate way of doing stats is not with JS-reliant technologies such as Google Analytics, but with your Apache or webserver log, which is the only true test of which page or object was served. 

Snipurl tracker also works server side -- if a snip is clicked, it is logged. No JS dependance. As such, it is likely to be more accurate.

Please note that "Unique Clicks" in our case were meant to be a flood control mechanism. If an IP address clicks on a snipurl after more than 10 minutes, it will be counted as a unique click. 

Anyway, if the aberration is huge (for example, if you know that 2 people clicked a snipurl but its snipcount shows 200) then please let us know and we'll check into it.