Arbitrage: buying low and selling high. Impossible in a market with perfect information and perfectly efficient transactions. Possible where there are information gaps or barriers to transactions. Lucrative where those gaps and barriers are larger.
Keyword: the phrases people enter into search engines. The basis for Google's highly profitable AdSense and AdWords programs and the dozens of copycat keyword-based advertising programs.
Keyword Arbitrage: buying low-cost keyword ads (cost-per-click) in search engines and other places, then placing more valuable ads on the landing pages.
For instance: I agree to pay Google $.15 for every person who clicks on my ad which will show up when someone types "lung damage ailment." Not many people will type that phrase and it's fairly broad, so it's cheap. When they click, they are taken to my site, LungDamageAilments.com, where I have an article about mesothelioma, a lung disease. I've also signed up with Google AdSense, so my site shows ads. Since the article is about mesothelioma, Google shows ads related to that keyword. Fortunately, ads for that keyword are paying out $10 to $40. I get a cut of that, say $5. But only when people click on one of those ads! Most people who hit my site will bail immediately. But if 1 out of 10 does, I'll net $3.50 per 10 people who click on my "lung damage ailment" ad.
The Risks: Sounds great, huh? There are a number of serious drawbacks to this idea.
1. Low Traffic - to get the cheapest words, I'll be looking at words that may only be entered into a search engine 100 times per month. Maybe, maybe 30% of those people will click on my ad. That's 30 clicks -- from the example above, that would net me $10.50 revenue for the month.
2. Poor Relevance - again, for cheap words, I'll be targeting general or otherwise irrelevant words and trying to drive people into much more specific ads on my site. The 1 out 10 I conjectured above would typically be pretty high for such a campaign -- many campaigns run at closer to 1 out of 50 or 100. Or less.
3. Automated Traffic - this is functionally equivalent to poor relevance, but the approaches to correct it are entirely different. Robots that simulate browser sessions, poorly configured robots, popunders, and other automated approaches to inflate click traffic are, depending on who you believe, rampant in all of the popular pay-per-click networks. I've seen first-hand how easy it is to fake some of this stuff, so I tend to believe the worst.
4. Low Price Differences - the $.15 and $5 in the example above are still realistic, in some areas, but competition is increasingly pushing those numbers closer together. The marketplace fluctuates a great deal and within a single day it's possible to see those prices change so much the one is higher than the other -- you're losing money on each click, even if every single one clicks through to another ad. Reporting in all of these systems is delayed enough that these fluctuations can have a dramatic effect before you ever know it.
5. Low Value to End Customer - to my mind, the most significant risk of this all is that it will drive useless traffic to the poor saps paying the high click prices I'm using to make money off this system. Long-term, that will force those prices down or force people out of the market altogether.
The search engines lock up much of the information that would eliminate this arbitrage. Perfect information is hard to come by and transactions are difficult. (The search engines point to their APIs, but if you haven't used them, you have no idea how sloppy and inefficient an API can truly be.)
So if you're willing to wade through the difficulties that make this tactic possible at all, there's still a lot of 'lucrative' left in it.
And if you're a search engine, please release a standardized, stable API that people can comfortably build into applications. You don't make any money off arbitrage and it hurts the industry overall. Perfect information and efficient transactions will be better for everyone who ought to matter to you.