Virtually all web analytic tools on the market today suggest the use of persistent cookies to identify unique visitors instead of relying on the old and inaccurate measurements of IP addresses. The by tracking cookies, you can more accurately track user behaviour through the web site, track returning visitors, number of visits between purchases, average number of purchases by visitors, campaign effectiveness etc.
The real question we need to ask is just how effective are persistent cookies. Unfortunately, the answer may not be what you want to hear. First off, they are only as effective as the percentage of web site visitors who accept cookies. From my observations at various clients, I've seen an acceptance rate range between 60%-95%. The biggest factor in defining the acceptance rate is the target audience for the web site. When targeting tech savvy individuals especially those in the open source movement, I've seen the lowest cookie acceptance rates. When targeting general consumers I've seen the highest cookie acceptance rates.
Another factor, based purely on my observations is location. It appears that people in some countries are more likely then others to accept or reject all cookies. The following table contains some my observations. These figures were taken from a single client who operates multiple b to b web sites across the world all with the same target audience and promoting the same product over the last 90 days (bot and internal traffic excluded).
Of course, if you want to be 100% accurate, you'd need to have users register and log into your site and then track them by their user id. Of course, if you followed this approach, most web sites wouldn't get any visitors and then they wouldn't have anything to measure. Remember, web analytics is about measuring trends and measuring effectiveness and even with a cookie acceptance rate in the low 80's, you can still provide statistically relevant data