Referral data is 100% required if you’re going to attempt to attribute advertising spend to revenue. I track user activity and session starts in a proprietary database (including referral data). I use persistent cookies (including the Flash type) and “stitch” cookies together using GeoIP, email address, IP, browser type, referring URL…everything including the kitchen sink. Why? Because it provides competitive advantage.
My dataset goes all the way back to 2003 for one of my sites. In a recent analysis of sales attribution I found that GA was incorrect more than 63% of the time. GA was correct to the point of its limitations. When examining the click behavior of a typical purchaser I often see 5+ organic searches, referrals from non-search engines, 3-7 PPC terms, a Facebook referral, and interactions with dozens of newsletters. I had a sale today where the user first visited our site in 2004. Long ago we thought that our sales cycle happened inside of 30 days. The average is 182 days. Knowing the click path and the length of the cycle is so important because it impacts strategy.
Referring URL is a key element in my analytics system. Without it, I have a colossal loss of insight. More sophisticated analytics provides better controls on spending. I find that I’m able to keep campaigns going that I would have otherwise turned-down for high CPO. And not just PPC campaigns. My model gives me a much broader advertising budget. It supports SEO, SEM, off-line, content generation...
The loss of referring URL is going to hurt my organic, e-mail, and “other-e” advertising buckets. Other-e in my book includes spend on blog seeding, link development, social, link bait, SEO, and anything where I don’t control the physical link. My ability to calculate and ROI on these investments will be lost.
The loss of referring URL will likely further my reliance on external tools like Hitwise, ComScore and the like (where the data is gathered from ISP click-stream data) – tools only available to those who have big budgets. It will also spawn a new breed of click stream providers that step in to provide actual referring URL data (the new gap). This will disadvantage smaller sites. In my book this is not such a bad thing because it is a barrier to entry and will help me defend my market space from pesky “one-man” knock off sites.
The loss of referring data will likely concentrate reliance on GA when and if they make changes. In my opinion, GA is not very good at attributing sales to the origin of advertising/spend over long periods of time – for another thread. Right now, GA is as affected by this change as any other web analytics platform. Amusingly, Google could make the click-through information available in GA, since it captures this information via the redirect it uses on the outbound links from the SERPs. I predict that they’ll do this at some point in some limited aggregated way. We’ll still know that the referral came from Google because they’re using a non-secure redirect that has the query parameter value removed from the redirect URL. It will be interesting to see how Webtrends, Omniture, Lyris, etc. respond over the next year or so to fill the gap.
I really don’t get the privacy concern. I get that users inadvertently pass semi-private information in the referring URL (e.g. if a user browses from a #*$! site directly to my site I know about the #*$!). Do we marketers use this knowledge in a bad/evil way? Not that I’m aware. We don't have time to spy on users. We use data to help us be better marketers. That's it! Why take away referring URL now? Answer: it hurts SEOs who manipulate the SERPs.
There is huge overhead in the SSL handshake. Is anyone smart enough to calculate the added cost of performance needed to handle SSL vs non-SSL? My guess is 5-10x. This means at least 5x the cost in server/SSL accelerators. We’re talking big money. How big? The larger the cost to Google the bigger the opportunity for Google. No business spends without thinking about the return. This change has nothing to do with end user privacy. I’m offended that Google is so dishonest. If this is about thwarting SEOs who manipulate results – just say so. No doubt, this is about Google making more money aka providing more relevant and less manipulated SERPs.
The roll-out globally was expected.