High Revenue in Cheating Business: Top 100 Ranking Worths Thousands
Wang Yi: I heard you can manipulate Alexa ranking? Can you achieve a ranking around 100?
A: Of course, even higher if needed. For a ranking around 100, it would cost about 20,000 yuan per month or more.
Wang Yi: Do you know how many domestic competitors like you there are?
A: As far as I know, only two, both from Chongqing.
Wang Yi: Is manipulating Alexa ranking your main job or side business?
A: It's definitely a side business. But now this business earns me more than my salary.
Wang Yi: How many people are involved in this?
A: I am the core person, with a few friends helping. We've been doing this since 2002.
Wang Yi: What is the best result you have achieved?
A: Now I can maintain a no-traffic website at around rank 50, which is the limit. However, we usually don't go overboard. And no one is willing to pay such a high price for us to do that. We charge based on the ranking level.
Wang Yi: Will maintaining these websites' rankings become easier? Because after boosting them, the traffic really increases?
A: Clients generally advertise alongside our efforts.
Wang Yi: Your point is that while they raise their rankings, they also promote their websites? Or do they use their rankings to attract advertisements?
A: The former. People who run websites know that having a fake ranking is meaningless. The mindset of clients is simple: they place ads, and their real popularity rises, but their ranking may not necessarily follow.
Wang Yi: So, Alexa is just another way to enhance brand awareness?
A: They spend big money on advertising and smaller amounts on us. By advancing on both fronts, when the actual traffic increases somewhat, the ranking improves too. Then, when they seek advertisements, it becomes easier. Advertisers see good rankings and try placing ads, ensuring their own traffic doesn’t look too bad, making advertisers more likely to approve.
Programming to Crack Toolbars, Fully Controllable Rankings
Wang Yi: You should be very clear about how Alexa toolbars work, right? Yesterday, I saw some cracked code at a friend's place. It returns a string of code each time a user browses, with about 9-10 parameters including URLs, browsing methods, etc., even monitoring internet speed. Is your understanding similar?
A: More or less, but that's not internet speed; it's page loading time. For example, accessing the same page locally results in "very fast," while accessing it from abroad might be "slow." Therefore, I think Alexa's Speed metric is unfair.
Wang Yi: Could you elaborate on your methods? As far as I know, Alexa ranking has two basic parameters, Reach and PV, both of which can be manipulated, but through different means. Additionally, Alexa has some anti-cheating mechanisms. How do you handle them?
A: Write programs.
Wang Yi: Cracking Alexa toolbars?
A: Yes, we don't install Alexa toolbars on our machines because they're slow and cause memory leaks. We wrote simulators in C under FreeBSD that can affect both Reach and PV values.
Wang Yi: Does this simulator simulate multiple users accessing a site simultaneously on one machine?
A: You could say that. After cracking its algorithm, we can simulate many users.
Wang Yi: How many users can one machine simulate? If we want to bring a new website up to rank 100, how much resource and time would be required?
A: Sorry, can't disclose that, but it basically simulates multiple users.
Wang Yi: Have you experimented with any websites?
A: Definitely, dsdiy.com is one I’ve worked on. You can check its Rank graph on Alexa. The webmaster agreed to let me use it for client demonstrations; otherwise, I wouldn't tell you. We value credibility in this line of work and don't reveal client information easily. This is the only one I can tell you about.
Wang Yi: Is this virtual multi-user access method very memory-intensive?
A: Not particularly, mainly CPU-intensive, but FreeBSD is powerful enough. It would be tiring under Windows.
Wang Yi: Is it mainly about using the CPU to calculate large amounts of disguised Alexa return codes and then sending them back as long as there's enough bandwidth?
A: Yes, but bandwidth isn't an issue, at least not for me. To put it bluntly, the most important thing is the realization of the simulator's algorithm.
Wang Yi: The code strings I saw from a friend were only a hundred bytes or so?
A: Right, so bandwidth isn't a problem. A decent CPU will suffice, not too bad, running in thread mode. A couple of clients directly approached me about buying the technology, but the price was too low, so I didn't agree. They underestimated both the technology and its value.
Wang Yi: Since you've been tracking Alexa for two years, has the algorithm rule for the code returned by the toolbar changed?
A: Only minor changes.
Wang Yi: Was there a change in October this year? It seems domestic websites were generally affected.
A: They adjusted the weight of two parameters, increasing the weight of PageView. Thus, websites that solely rely on advertisements to boost traffic but fail to retain users suffered. Such websites might have increased traffic, but low PageViews indicate that users attracted by ads find the content uninteresting and quickly move to other sites.
Wang Yi: Do you know the complete algorithm formula for Alexa rankings?
A: Only Alexa knows the exact calculation, but they certainly hope the algorithm is reasonable. So they adjust it according to common cheating methods. However, our approach directly cracks the toolbar, and unless they abandon current toolbar technology, they can't stop us.
Wang Yi: Besides the toolbar, does Alexa ranking have any other basis?
A: Currently, people only know about the toolbar. Alexa hasn't mentioned any other methods, so it can only be sample data. Many often ask, why does our website's independent IP increase so much yet the ranking falls? I tell them there are two reasons: 1) More independent IPs don't necessarily mean more visits from users with Alexa toolbars installed; 2) Even if yours increase, others might increase more. Rankings are comparative.
Cheating Has Principles Too: Don't Forget the Well-Digger
Wang Yi: How many clients do you currently serve?
A: We generally don't do random jobs; now we're handling no more than five, very cautiously. But there are always many people looking for us.
Wang Yi: All require rankings around 100?
A: Not all, some only need to reach around 2,000, these clients are rational.
Wang Yi: What are the principles for choosing these clients? Is having money enough?
A: No, some small websites come wanting to be ranked 100, we don't take those.
Wang Yi: You don't accept overly demanding requests?
A: It's not about being too demanding; if I wanted to, I could achieve it. Frankly speaking, we are parasites of Alexa, and we don't want to kill our "host."
Wang Yi: Strategy matters too?
A: Yes, last time a client got upset with me. Hehe, a girl, I spoke too frankly, saying what good is a high ranking for a junk website, she got mad.
Wang Yi: Can we say that there are five websites online whose Alexa rankings you maintain, all below 100?
A: Yes, there are very few websites that can afford the price for a top 100 ranking. Basically, I focus on rankings between 300 and 1,000.
Wang Yi: How much is the price for 300th place?
A: It depends on the website, generally costing over ten thousand. If marked as 300, it actually fluctuates between 300 and 400.
Wang Yi: This price is not low for small websites.
A: Why should a small website be among the global top 300?
Wang Yi: Are all your clients big websites?
A: At least they are qualified. Some small websites only require to reach around 10,000, and we don't bother with such cases; they're not worth it.
Wang Yi: What do you require from them?
A: At least the website should look decent, and they should be able to afford the cost. One client only did it for a month because their server was hacked, causing a significant drop in ranking. They asked me to help recover for a month.
Wang Yi: Won't it fall back significantly once you stop after a month?
A: No, that website still has a solid user base, and the traffic is gradually recovering. They just needed me to recover their lost quarterly average ranking. So every day I had to adjust the intensity gradually, making it hard for others to detect someone was helping them cheat.
Wang Yi: Can you control the pace as you wish, going up and down at will?
A: Yes, it wasn't possible before, but later improvements were made because clients' demands kept increasing.
Wang Yi: Is it like this: at the start of the month, your role is more significant, then gradually diminishes as the website’s natural traffic recovery takes effect, by the end of the month, even without your help, their traffic is mostly restored?
A: Yes, in this way, the Alexa Rank curve of the website remains relatively stable throughout the month, and uninformed people won't notice anything. They only see the sudden drop during the attack days and the quick recovery afterward, which actually lasted a month. My role was to ensure the quarterly total average ranking of the website wasn't affected. Such capable websites would recover even without my help after a month, but due to the low rankings for that month, it would affect the quarterly total average rankings for three consecutive months.
Wang Yi: Do such capable websites care about Alexa rankings?
A: Very much so.
Wang Yi: Why?
A: For instance, if you run a website and tell advertisers how great it is, but they check Alexa and find your ranking is very low, who would believe you?
Wang Yi: But there aren't many Alexa toolbars installed domestically, which seems disadvantageous for domestic websites, right?
A: It has increased significantly now, whether you believe it or not. I've known this technology since 2002, but I only started this business this year because Alexa has only recently gained attention domestically.
Wang Yi: How widespread is cheating among domestic websites as far as you know?
A: Many people approach me. Many domestic websites engage in this practice, but their cheating methods are not sophisticated. Those who understand can analyze it from Alexa's historical curve chart.