An Achilles heel of pay-per-click internet advertising is the risk that someone will click on an ad they’re not the least bit interested in.
To the extent that unqualified clicks are limited to a few over-caffeinated users who click wildly, this isn’t really a big concern — especially to advertisers accustomed to the targeting constraints of traditional media.
Yet click fraud, usually called invalid clicks by advertising platforms, has been a real concern over the years.
Click fraud comes in many forms. Paid to read schemes target ads which compensate the hosting site with a percentage of every click by paying people to click on ads appearing on selected sites.
Automatic software, known as bots (from robots), can scan (crawl) web pages on Facebook, Google and other sites hosting advertising in order to collect information contained in the web pages. In the process ad links may inadvertently be followed, or “clicked on”. Facebook has estimated 1.5% of its accounts are being used for purposes that violate their terms of service, terms which would include a prohibition on data harvesting.
Most internet advertising platforms employ sophisticated anti-fraud measures to protect the integrity of their systems: advertiser satisfaction is key to obtaining repeat business. Besides, disgruntled advertisers can and do sue. Google seems to have perfected their click fraud detection mechanisms to the extent that the topic doesn’t appear to be the significant advertiser concern it once was.
80% Of Facebook Clicks Are Fraudulent?
I might have dismissed a Techcrunch report that an advertiser claimed 80% of billed Facebook ad clicks to be fraudulent, thinking of it as a successful attempt to garner earned media visibility. But a friend had reached out a month earlier with the same concern (the article is in Italian, use Bing or Google translate for a readable English version).
I don’t really know if Facebook’s ad system has an integrity problem. They shouldn’t, as the lessons that Google and others have learned about click fraud are well known within the industry.
Perhaps the run up to the IPO distracted management attention from the importance of click fraud detection? One thing appears certain: Facebook has a PR problem to manage.
Perhaps Facebook should better explain its proactive and reactive handling of click fraud — providing data to support claims of its efficiency. In the meantime, Facebook advertisers may want to verify the integrity of their Facebook advertising campaigns. I thought I’d share some of the same audit advice with you that provided to my friend. Ready?
Ad Clicks Will Almost Never Equal Browser Based Web Analytics Systems Page Views
A user clicks on an ad. The advertiser is billed. Yet this doesn’t mean the user ever went to the target site: they may have changed their mind and closed the browser before the landing page loaded. Is this common? Maybe not. But we need to keep it in mind.
It is also important to understand that browser-based web analytics systems, like Google Analytics, won’t track a web site page view unless a series of conditions are met. In the case of Google Analytics, the browser must load images, accept first party cookies and execute JavaScript. Did you catch all that? Hang on, we’re not done yet.
Tracking code at the bottom of a page may not execute if a JavaScript error exists earlier in the page (all sites should be using the latest version of the Google Analytics tracking code, the asynchronous version, for more accurate tracking).
Users can also disable Google Analytics tracking, although I suspect very few know about this feature designed, I imagine, to head off German regulation. In the real world, so many websites require JavaScript to work that the number of non-JavaScript page views should be around 1-2% – and many of those are probably robots.
Web Server Logs Can Provide Great Click Fraud Audit Data
Web severs were designed to log all requests for pages and other files required to compose a page in a browser window. The data collected usually includes the requesting internet address, the time and date, the file requested, the provenance of the request (referrer) and the software used, such as browser and operating system.
Real users should request not just the web page, but the associated image, css styling and/or javascript files as well. Human visitors will be using a known browser, like Chrome, Firefox, Opera, Safari or Internet Explorer.
Bots can fake some of this data, but at least these considerations make for a good starting point. Log analysis does have some limitations. Server cache directives need to be properly configured to insure web server logging occurs even if a page is already in the browser cache.
This isn’t hard, but configuration can lead to under-counting. Theoretically a web sever may serve pages but stop logging, such as if the log is saved to a full disc partition, or if the server suffers from certain overload conditions. As they say, the only true certainties in life are death and taxes.
What Says Facebook About Invalid Clicks/Click Fraud?
Facebook devotes an area of its help center to what it calls Click and Impression Quality. As an aside, searching Google for Facebook and click fraud won’t get you here as we have a classic search engine optimization keyword exercise: advertisers will probably search for click fraud while advertising platforms use invalid clicks. Facebook says:
Although we cannot guarantee that our systems will be absolutely precise, we try, for example, to detect and invalidate the following:
Human clicks that don’t indicate a genuine interest in the ad or may be associated with ad testing, like repetitive or accidental clicks, or visits from the Facebook corporate network.
Clicks generated through prohibited means, such as bots, scrapers, browser add-ons or other non-human methods.
That may well be the case. The question remains, does their trying to filter invalid clicks match high advertiser expectations, often set by their experience with Google?
Do Facebook Page Name Changes Require A Payment? Essentially Yes.
The start-up cited by Techcrunch, Limited Run, also complained that they cannot change their Facebook Page name without paying Facebook. In my experience, they are essentially right.
Facebook wants to protect its users from a bait-and-switch type of abuse where a user likes a page for Politician X supporters and subsequently finds that a page admin changes the page name to Those to unseat politician Y.
In the real world companies and products change their names. The Facebook help article on the topic is unhelpful. If your page has accumulated more than 200 fans… you’re out of luck. If you look really hard, you might find a page about contacting Facebook support. The message is clear: Facebook support is self-service.
If you’re still not happy, you can provide a suggestion – but don’t expect a response. To a degree, this is understandable. Support is expensive. Yet what is one to do when Facebook’s self-service system is broken?
I contacted Facebook’s country manager who nicely put me in touch with a Facebook Account Rep. The answer: they only help those who spend significant sums over time.
A year later, I’m still waiting for a page’s name to be fixed. Despite Facebook’s apparent denial, Limited Run is right on this account, at least in my experience.
What Facebook’s help article really should say is that to change a page name the only option is to create a new page then try to convince the old page’s fans to migrate to the new page. You’ll naturally lose a few fans along the way, as will Facebook.
Commentaires