From a Google spokesperson, S. Tran, in response to a letter from me:
I received your email re: Google Analytics and Google Trends for Websites.
Regarding the difference in data, the results from Trends for Websites are estimated and may differ from Analytics data. Trends for Websites is in Google Labs, so it's still in its early stages of development and may have inaccuracies. In the future, we may consider ways to improve the accuracy of Trends for Websites by allowing owners to contribute additional data from their sites.
As for the development of Trends for Websites to vanquish click fraud to expose sites with inaccurate claims, Trends for Websites is an extension to the Google Trends set of features and we think its the next step in helping users understand how the internet is used. The goal of Trends for Websites is to show interesting trend relationships between website audiences over time.
Thoughts:
- Google does not appear ready to say, for the record, what GoogleTrends will do, and how well it will do it. This is characteristic of items that are still in Google Labs: in fact, GoogleTrends for Websites may never emerge to join other Google products. I'm pleased someone took time to respond, even if the disclosures are limited;
- So far, research by myself and others shows GoogleTrends doing quite well, except among many large webcomics (not all). Since I started monitoring webcomics with it a while ago, I'll continue to check from time to time to see if anything changes. It remains odd that trends should seem much more out of whack with reports for webcomics, however, so more research is needed;
- Note the mention, in the letter, of opting in to include data sharing from Google Analytics with Google Trends. Should this come to pass, it might be a way for a website to demonstrate a commitment to honest traffic reporting;
- We must keep in mind that from the top down, addressing click fraud is by default central to Google's planning and actions, and that GoogleTrends might provide infrastructure by which they can address it;
- Or not, the last sentence suggests subdued plans, for now at least. Responding to inquires about projects under development is a tricky business, balancing privacy and honesty with great care;
- The last sentence also reminds us that you can do side-by-side comparisons of about five sites at once, which is something I've only tried a few times -- worth noting;
- If GoogleTrends for Websites does become yet another source of public analytic data for traffic, Google has choices to make: they can create a product that screens out false traffic click fraud, and/or attempt to become the source of record for how a site performs. Either has significant implications for anyone engaged in click fraud already. If Google does not address click fraud as part of its strategy, but creates a source of record, it is in effect addressing the click fraud issue;
- Any public analytics system striving for credibility must build in resistance to traffic click fraud, or it will be in the interest of the dishonest to increase click fraud. The threat of a system appearing that already has a track record of historical performance for a site, and which, at some future data, will screen out fake traffic, is a brilliant way to blunt fraud now, and expose fakery without going case-by-case and tossing people out of the Google system;
- If that is Google's goal, then people who run sites on steroids are going to be revealed publicly, as we see huge disparities between what they claim and what GoogleTrends claims. There's no need to become defensive and write denunciations with steam blowing out your ears, because you rise or fall based on your data legacy;
- If Google has a system for teasing out real clicks from false that builds upon what it already does, we are left to wonder whether it has been turned on. Recall ADSDAQ's sudden retreat from webcomics last year. Cartoonists may be fine people, but the historic webcomic business model that hosts ads from major brokers is right up there with spam sites in terms of potential for abuse. All vendor-consumer interactions have a subjective aspect, but webcomics particularly so: they are free, and yet they are not; you as reader are holding up your share and yet you are not; you are from a demographic that trends left and doesn't mind clicking some insurance company's ads or you are not; you are dealing with a website where the creator will talk to you on Twitter or you are dealing with a faceless company... these all contribute to abuse potential, in various ways;
- One would think that an aspiring webcomic professional who has co-written a book on the topic would be grateful when a blogger comes along and tries to reconcile conflicting public data about his performance, as part of a group. Anyone who dares discuss such data must be prepared for fits of rage, scapegoating and troll bombardments;
- When I talk to webcomics people as part of doing research, I notice that they are mostly friendly, sometimes cagey and the rest "other." The cagey types do not like to discuss their methods or data, and sometimes seem to feign ignorance of their own analytics. Some get hot, and write things questioning my motives. "It's amazing how much you scare them," said one veteran observer, amused. The fact that I would put people's advice to the test and report on my progress causes upset. Historic tendencies towards relaxed ethics now conflict with the message of professionalism, and the damage is self-induced. You can't argue for a viable, professional webcomic profession while you are behaving unethically, and the protests mostly serve to flag those who know, deep down, that they lack both the guts and the decency to apologize for transgressions, past and present.
- I'm convinced that under the right circumstances, webcomics can be monetized and can even make a viable career choice. But newly emerging careers often start with lax practices and end up self-regulating as people see that bad actors hold down progress. There is incredible peer pressure among some veteran cartoonists to hold the line against progressive thought, even as some candidly reveal the extent of deceptions in non-public conversations. Don't assume I take every such report seriously -- I am skeptical until convinced, regardless. An old Russian saying: доверие, но проверяй (Copy, then translate it)
Remember: I am not accusing anyone of traffic click fraud, and people who have data on Google Trends that makes them look sketchy might try being glad that someone is trying to figure out what is driving the odd results for webcomics.
