Commentary
Tag Ubiquity
- by Josh Chasin , Op-Ed Contributor, October 27, 2010
Unlike traditional media, however, the technology behind the Internet lends itself to such data collection -- hence the "most measurable" title. And of course, I'm talking about tagging.
As an old-guard media researcher here in the new-world digital space, I figure about half of you know way more about tagging than I do, and the other half, not so much. So for the older-school half... A tag is a piece of code on a web page that logs user requests for content elements (the page itself, or a video or ad or other object on the page), and instructs the site to drop a tracking cookie onto your machine. Audience measurement companies use these cookies for counting. Web Analytics providers use them to provide KPIs. Targeting companies use them to encode targeting attributes about you that help advertisers deliver better-targeted ads to your screen. Together, these requests for content elements ("tag requests") and cookies form the backbone of what we call census, or site-centric, measurement.
Let's take a step back and make this tangible. Please be our guest at the comScore home page by clicking here. Are you there yet? Good. Can we offer you a cold drink? Now, with your cursor on the page, right click (you Apple people are on your own) and select "view source code." That'll open up a new window on top of the home page. Now, see all that gobbledy-gook in the new window? That's the code that tells your computer how to assemble and display the page for you. Scroll down to lines 1068 through 1085. That particular chunk of gobbledy-gook is the comScore tag, which instructs the website to tell the comScore counting server that you've requested the page, and also tells the site to drop a comScore cookie-- which we'll recognize when you request another page, and also if you come back tomorrow-- unless it's gone, in which case you'll get another. (Unless you block or reject cookies. But you're reading the Metrics Insider, so you probably don't.)
In all, we've got four tracking tags on our site. Chances are, if you work at a publisher, you have way more tags than that on your site. You're probably working with ad servers, RTB platforms, networks, and exchanges. Over the past several years there has been an explosion in tagging. The data that is created, stored, mined, tracked, and deployed by these tag-based solutions is helping publishers monetize their audiences. It is helping advertisers reach their audiences more effectively and efficiently. And, it is even enhancing the consumer experience with the web. It's kind of a triple-win scenario.
But everything has a price. Publishers are concerned that tag ubiquity has the potential to affect site performance (the more things that have to happen when the user requests a page, the longer it takes that page to load, and each set of tags firing constitutes another thing happening). In addition, the time and effort required to manage these different tagging implementations can create a material demand on publisher resources. At the IAB Research Council meeting last month, when the topic of tagging came up, the top researcher at one leading web publisher commented that the analysts in his department were spending up to 30% of their time on tagging implementation issues. Another publisher researcher commented that she was spending 25% of her time on such implementation issues.
So over the past couple of years, the idea has been buzzing around that perhaps there should be one single, central tagging repository -- one single set of tags that every publisher could use, that would track all the data necessary for every third-party solution provider (audience measurement, web analytics, network, ad server, exchange.) In fact, at that IAB Research Council meeting, the notion of the centralized, universal tag was on the agenda.
It's a theoretically appealing idea, but I think there are some very good reasons to argue against it. One is the question of whether all the data that every third party requires could in fact be encoded into a single set of cookies (I'm assuming a central tagging warehouse would also involve a centralized cookie). Beyond feasibility questions, there are privacy questions (just how much data about a consumer should we be embedding into a single, essentially open-source cookie)? And too, I find myself concerned about the notion of a single point of failure; what happens if that tagging clearinghouse goes down for a couple of days?
But the publisher pain points about site latency and labor costs are legitimate. So the question becomes, given these real pain points, is a single universal tag the appropriate solution? Or are there other alternatives?
I think perhaps there are.
Tag ubiquity has spawned a new service category, called Tag Management Solutions (or TMS.) As Eric Peterson (who, full disclosure, works with one of the TMS providers, Ensighten) put it, TMS is "a ‘one stop shop' able to manage any number of tag-based technologies via a single user interface."
I'm aware of three players in this space (if your company is a fourth, say so in the comments to this piece): Ensighten; Tagman; and Tealium. To be clear, I don't have any sort of preference, recommendation, or endorsement for one or the other of these solutions. I'm not even saying you need one. And from a company perspective, since comScore is providing more and more applications that incorporate tagging, our philosophy is to be TMS-agnostic.
But I do think -- and the people who know way more about this than I do, and with whom I consulted after that IAB Research meeting, seem to agree (hey, Jodi!) -- that the notion of a single, centrally administered source for all tagging is probably the wrong solution to the pain points created by the explosion in tagging. The good news though, as is often the case in the digital world, is that where technology unfettered creates a problem, more technological innovation solves it.
Speaking of measurement: 1,031 words in this article and the word "advertiser" appears exactly twice, whereas "tag" (or a derivation thereof) appears 27 times.
You're so busy playing tag, you've forgotten what business you're in.
My understanding of central tagging solutions is that they enable you to put all of the tags into an all encompassing tag. However, this will not solve latency issues. The tags still belong to their owners.
GREAT (agnostic) insight into a hot topic Josh. Thanks.
Not all systems are the same, so do ask us all how we differ why each should be in the consideration set.
We recently did some (agnostic) research that confirms the need for Tag Management Systems with over 95% of respondents reporting issues trying to hardcode tags into pages. The summary is called "The Real Cost" and located here: http://www.tagman.com/index.php/the-business-case.html
Mike, the business I'm in is, measurement.
The column is called the Metrics Insider.
The column is about tagging.
But hey-- advertiser, advertiser, advertiser.
Sorry, Josh, but you're dead wrong. Your job may be measurement, but your business is advertising. To wit, who underwrites your efforts? Who stands to benefit from them? Every penny you make comes out of the hide of some poor advertiser who probably doesn't understand what in the hell you're talking about, let alone why he needs it.
Your cohorts at the IAB, their reductionist initiatives notwithstanding, at least have the good sense to include "advertising" in their name.
Good thoughts, Josh. Whether you're an advertiser, agency, demand side platform, analytics group, direct measurement platform, ad verification company, Comscore or my company, Safecount -we're all requesting and/or enabling both Ad AND Page level tagging - and the brunt of that falls on publishers. But - some of those companies are providing a publisher-centric remedy also (data exchanges, tag containers). I think the groups I mention above need to ask themselves - are they?
Hey Josh, good high-level summary of tag management systems, and thanks for linking to our write up and white paper on the subject. One thing that constantly surprises me --- and part of the reason I rail against the term "Universal tag" --- is how little folks seem to know about the idea and the, umm, strange assumptions they make.
For instance, why would a tag management system require a single cookie? Why would a tag management system be any more likely to fail than the hodge-podge of tags most companies have on their sites today? And why would a well-designed tag delivery system increase latency (as opposed to decreasing latency, which is what we are seeing in many instances?)
Maybe it's because people fear change and things they don't understand. As I recall we went through the exact same process (with many the same questions) back when tagging was new ... and look how that ended up!
Anyway, thanks for shining some light and as I mentioned in email I'd love to discuss how comScore can get on the tag management bandwagon if you're interested.
Sincerely,
Eric T. Peterson
Web Analytics Demystified, Inc.
http://www.webanalyticsdemystified.com
Agree with Eric's point on questions: specifically latency, a point I labored today with an agency who thought TagMan was a cure all. EG Just because you can now deploy hundreds of tags/pixels easily, segment on the fly etc through one interface, without IT - why would you? It's not guns that kill.... The idea is to use a platform to manage with more panache, reduce IT workload, but still keep them involved.
That said - TagMan do have 'ServerTag' that removes pixel calls completely with certain providers - so that helps reduce some areas of latency if someone did want to work with more providers.
Tom - the challenge with supplier owned containers is simply 'would you let the wolf guard the sheep' and also, these vendors are focusing on making their main offering first, the container second. Tag Management Systems are from the ground up solutions to cope only with tags at their heart and ensure they work seamlessly.
EG - do you want your banker to fix the brakes on the family sedan, or do you want the approved mechanic?
Hi Josh,
Mom taught me if I didn't have anything good to say, then I shouldn't say anything. Thus I am not saying too much right now about TMS' or anything about Einstein.
But if you ask I will say nice things about advertising, advertising, advertising and measurement, measurement, measurement, and how kind it is for people to count the words in your column.
Less serious, more later,
Judah
Judah,
No offense, but site analytics are the problem, not the solution. Human behavior is irrational, which makes any attempt to rationalize it, let alone predict it, a fool's errand.
Keep chasing your string in the rear view mirror but remember the sage advice once given to Charlie Tuna: Starkist doesn't want tunas with good taste, Starkist wants tunas that taste good.
Josh,
Great article. I agree with Eric and Chris's comments below. I do want to emphasize a few points.
1. Tag Management Systems should accelerate the delivery of tags by minimizing the number of connections made to get the tag and optimizing the delivery of the code needed to execute the tags.
2. Tag Management Systems do not require the use of a cookie and do not need to collect any personal data about the users of the site they are managing.
3. With the growing concerns about end user privacy (see resent WSJ articles) , Tag Management Systems provide organizations the ability to control at a very granular level what data is sent to what 3rd parties and under what circumstances.
Josh Manion
CEO Ensighten
www.ensighten.com
Thanks Mike Einstein. None taken and lol. No offense, but it's really obvious that you are too smart for me, Josh, the reductionists at the IAB, statisticians, and all of us other string chasers, but not for Charlie Tuna.
I have no idea what the weather will be tomorrow.
Sure, some guys have spent a lot of time chasing strings in the rear view mirror and tried to use that to predict tomorrow's weather, but they're on a fool's errand. Far easier to just give up and pray to the gods. That's what I hope my airline weather forecaster does, just leave it to chance and optimize nothing. Why try?
Useful post.I learn more from this article.Thanks for sharing.
http://shopatoutlet.com/
It sales <a href="http://shopatoutlet.com/" target="_blank">Christian louboutin</a> products at very low price but with good quality.