Posts Tagged ‘Click Fraud’
How a patent came to be
Many, many years ago when I was a youngster growing up, I read lots of books about people like Thomas Edison and Alexander Graham Bell. I marveled at how many things they had invented, and the effects just the things they patented still had on our lives.
At the time, I never expected to be in a position where I would have my name associated with a published patent.
Yesterday, a patent I wrote was issued. It’s patent number 7,657,626, and refers to an original method of detecting click fraud. It’s a dry read, so I’m not sure if you want to rush out and read it, but if you’re having a hard time sleeping, you can find the patent here, at the USPTO office.
Now, in no way do I consider that this one issued patent in any way puts me, or any of my ideas in the league of these great inventors or innovators, but I am thrilled nonetheless. To have an idea recognized as truly original in this process is special, and I truly feel honored, and am indebted to everyone who helped guide me through the process.
Of all the things I’ve done, this one really was like a light bulb going on. The idea came to me over a holiday weekend a few years ago. My son and I were enjoying a boy’s weekend getaway at our cabin at a lake near Victoria. I’d been deep in data looking at random server activity data all week, so playing around in the water, cooking meals over the fire, and staring at the night sky was a welcome change. The second night I went to bed around 11, and was out solid until 2:30 a.m. I woke up with a start and an idea in my head. Normally, like most people I’d just roll over and go back to sleep, telling myself that I’d deal with it in the morning, by which time I’d forget about it. But this time, I grabbed a pen and paper and started writing. Two hours later I was stuck; I’d written many pages of notes, and long-form formulae / processes, but something was missing. I put the notes away and went back to sleep.
The next day I spent being very busy doing nothing! I didn’t even read my notes from the night before, figuring that I could get back to it after the weekend. That night, I woke up at 3:00 a.m., with an idea how to break through the last barrier in the process. I started writing, and that night completed what turned out to be the first draft of the provisional patent application.
The following week I sat down with our Mathematican, Rafael Parra Hernandez, and he helped me take my long-form formulae and convert them to more conventional mathematics. He also explained to me the various types of math I had incorporated in the processes. If my high-school math teachers could see it now, they’d never believe it!
And, that’s how a patent came to be.
Search Engine Strategies (SES) New York 2008
I’m off to SES New York this evening. No, I don’t particularly enjoy red-eye flights.
On Monday at SES, I’ll be speaking on the Click Fraud and Click Auditing panel. Jeff Rohrs, Shuman Ghosemajumber, Tom Cuthbert and myself are the only carry-overs from the Click Fraud panels at SES Chicago last December. As Tom didn’t have a powerpoint last time, I look forward to him bringing forward some new data. I’ve got an entirely new presentation, with perhaps only one holdover graphic. I hope those of you who will be there will enjoy it!
Two weeks ago, Shuman and I had lunch at the Googleplex. We discussed a lot of things, and I only realize now that one thing we didn’t discuss was this panel at SES NY, other than to say “see you there.”
On Wednesday, I’m also moderating a late session on Searcher Behavior. I’m looking forward to moderating this particular session as I’ve spoken on it a few times, and the change from speaker to moderator on this topic should be interesting.
If you’re at SES New York, please do say hello, come check out the sessions, and ask lots of questions.
Getting Refunds from Yahoo! and Google for PPC Campaign Errors
A lot of people focus in on how to get refunds from Google or Yahoo for Click Fraud issues. Google doesn’t always call it click fraud, they often call them invalid clicks, and when they catch “invalid clicks” they pro-actively discount your bill accordingly. They don’t catch everything, but they do try hard. Yahoo! does the same thing, but less obviously. They don’t actually show you how many invalid clicks you’ve received, they just don’t appear to bill you for every click.
In both cases there are defined processes for requesting refunds or more commonly, credits.
Completing the documentation to request a refund isn’t simple, trivial, or a speedy process, (unless you have PPC Assurance where it’s a one click process). In fact, it’s quite complicated. Rather than confusing matters by outlining processes for both Google and Yahoo!, I’ll focus on Google. They’re the 800lb gorilla which everyone cares about.
In Google’s case, to file a request for credits for clickthroughs you believe you were improperly billed for, you need to identify all the original referrals, which means figuring out which entry in your log file is the original referral, and isolating the unique Google Click ID (gclid). You then need to document everything possible about that click, as in the course of an investigation, Google’s team might ask you for a lot of data. Be prepared. They are just trying to be thorough.
One issue you’ll face is how to claim what. The obvious documentation on the web deals with “Invalid Clicks” Unfortunately, invalid clicks don’t always mean the same thing to you as they do to Google. Not all invalid clicks are Click Fraud. To you an invalid click might be a referral for an incorrect keyword match. These do happen, but you’re unlikely to notice them in a large campaign, as too many terms are flying across your screen. This type of mistake actually gets handled by a different department at Google. Challenging to navigate, that’s for sure.
It’s not that Google actually sets out to make it difficult to claim back a refund, or to get a credit for mistakes. Simply put, Google’ a big organization with responsibilities for different issues assigned to different groups. They are trying to be as efficient as possible, but these efficiencies don’t necessarily make processes simpler for you, or your clients. They simply need to be thorough.
Is it worth your while to manually track down all the errors? It depends on your cost per click, and your volume. Is it worth doing so automatically? Definitely. At a cost of 1% of campaign spend, knowing what’s going on, when things go wrong, and how to deal with them is invaluable. Knowing you can recoup more than that means the ROI is pretty simple to work out.