I would agree to @Mad on account of what he quoted and also reconfirming what @Rayer quoted from my facebook status. For over 5 years, we studied how Alexa algorithm works and how it ranks sites based on their site traffic data but we were badly surprised when we saw how badly could this algoirthm be fooled because it is simply the most misguiding Ranking metric which has diverted both publishers and sponsors for about two decades Alexa did little to optimize the way they collect traffic data from commercial websites and rarely provided convincing customer support, if there is one thing that recently changed on their site is their Website design!
During our study we brought down MBT’s ALEXA as low as 5K thus making it the first blogger blog to have achieved the Highest Alexa rank but today when our traffic is much better and smooth we are at 13K, Same goes for our other network websites. We touched the 5K benchmark through playing with our posting timings. For example publishing posts during midnights around the odd timings did massively dropped the ALEXA rank by 100 units if your ALEXA rank is around 10K and it brought it down by a further 1000 units if your site is under 100K. Surprisingly Asian midnight timings with odd numbers like 12:01 AM, 12:07 AM, 12:13 AM, 12:17 AM and so on dropped our Alexa about 100-200 units each day and we achieved that with daily posting. But the moment we stopped this trend or posted at a wrong even timing, our Alexa rose by 300-400 units in the reverse order which badly demotivate us. Because we were often busy with university exams and so were my authors, we could hardly post daily, hence we could not kept our Alexa rank stable at 5K.
After what we observed through trial and error, we came to a decision that ALEXA rarely monitored the rank based on visits to a site. There are sites with much better ALEXA compared to ones which have far greater traffic and organic flow of visitors, ALEXA surely miscalculates sites and provides false analysis data which can rarely be trusted as further confirmed by many webmasters if you Google.
We kept a close eye on ranks of different sites and compared them with us, their posting frequency, publishing time, inbound links and organic traffic but rarely could we build a trend which ALEXA followed. As far as those old tips of ALEXA widget, web development tutorials, toolbar, reviews of ALEXA, backlinks etc. are concerned, we applied every tip but rarely any worked out. These are just suggestions found widely on internet but they never work accurately when applied practically.
PageRank too has not been updated recently for quite a long time because many webmasters found miscalculations there too, but as far as the data for backlinks is concerned PR can surely be trusted but when it comes to judging a website traffic, ALEXA is never a wise choice
Focus only on your backlinks and orgnaic traffic because sponsors do rarely care about ALEXA anymore and this is one reason why many ad networks have stopped showing ALEXA rank stats which includes buysellads on the front.