How Tag Management Benefits Web Analytics: An Interview with Gary Angel
Editor’s Note: The following is an interview with Gary Angel, President and CTO of Semphonic, a leading digital measurement and analytics consulting firm. Gary was recently named Most Influential Industry Contributor in the Digital Analytics Assocation’s 2012 Awards for Excellence.
1. Tell us a little bit about Semphonic?
Semphonic is a digital measurement and analytics consultancy. I often stress the word analytics, because we focus on delivering actual site and digital channel analysis to our clients, most of whom are very large, multi-channel enterprise clients. Most agency and internal teams that describe themselves as web/digital analytics seem to be pretty much limited to tagging and reporting. Getting clients to the point where we can do real analysis has also pushed us to deliver both infrastructure (tagging) and reporting solutions and, increasingly, it’s meant that we’ve been helping our clients drive analytics data out of traditional web analytics solutions into data warehousing. But the end goal is to drive change – and analysis is what does that.
2. When did tag management first come on your radar?
A couple years back. Seems like ages ago. I think the first real application we saw was actually with you guys. We had a client with a critical site section (The Account Application Process) that was built in a manner that changes could only be done with a full development deployment cycle. So it took about six months to drop an image request on the page. And since it was the single most critical conversion point, they kept wanting to add tags.
3. What are some ways that tag management benefits web analytics?
Good analytics is dependent on good infrastructure. Costly, time-consuming infrastructure deployments greatly extend the time-to-benefit from analytics and can even chew up all the available measurement dollars.
Our team greatly prefers to do deployments these days with a Tag Management System (TMS). A TMS moves the bulk of the tagging work from IT to the measurement team. It’s actually much easier for us, in most cases, to customize the tag than to explain to an IT team how to do it. It also puts us in control of the measurement. I think it makes for better and cheaper implementations.
Tag management also gives you much greater ability to tune and customize the tag without impacting site and IT cycles. That’s critical because measurement is the tail, not the dog. Lots of measurement changes get de-prioritized or delayed when you have to run them through traditional IT cycles.
4. From an analytics standpoint, what are some of the signs that you may need a tag management system?
Honestly, I’m not sure you need signs. In my view, every tag deployment at the enterprise level would benefit from a TMS. But if I have to pick some signs, here they are:
- Governance issues with missing tags
- Measurement queue of tagging changes
- Reduced measurement designs to fit implementation cycles
- IT concerns with tag proliferation
- Constant changes to existing tags or foregone reporting because changes can’t be made
- Page load performance issues from tags
5. What are some of the things you should look for in choosing a good TMS?
We’re finding that the decision about a TMS blends two separate elements: IT and marketing. For years now marketing has been adding tags to the web site without a lot of feedback, governance or control. Interestingly, when it comes time to think about tag management, IT concerns tend suddenly to surface. That’s not a bad thing. I think you want to look at a TMS that supports core IT concerns around page weight, performance, control, and reliability.
On the other hand, measurement is a marketing function. The single biggest reason (by far) for having a TMS is to create a system in which the actual measurement design can be quickly and seamlessly controlled without tying up IT cycles. The current system of tagging puts tremendous pressure on the design-cycle to create a comprehensive infrastructure. It makes for bulky implementation process – and makes it very challenging to work on any kind of an agile cycle.
So I think the features that matter most in a TMS are the ones designed to facilitate that flow: configurability of the measurement system from a GUI, ability to control and deploy page-based customizations from a GUI, workflow management that provides sound governance, and a content structure or UI layer that makes it possible to track and manage large numbers of site customizations.
6. Is web analytics vendor lock-in a thing of the past with tag management?
I’m going to have to say no. Tag management DOES significantly reduce web analytics vendor lock-in. Tag implementation is probably the biggest single barrier to changing solutions – so the impact is real. But there are a number of other friction points that don’t go away. Collection design is actually somewhat different on different systems – so there is design and implementation work to be done even with a TMS in place. And, of course, things like training, reporting, data feeds, APIs, etc., all introduce significant friction. I think it’s fair to say that tagging is the one major difference between web analytics and other types of enterprise software when it comes to friction and vendor lock-in. But vendor lock-in exists in pretty much every single type of enterprise software.
7. Semphonic has grown tremendously. What are your plans for the future?
I wish I knew! Seriously, one of the hardest things in our industry is the pace of technology change. It puts a lot of pressure on consulting companies to constantly adapt. Going back to my first answer, I’m pleased with the increasing level of maturity in our industry. It’s creating real demand for customer analytics that actually drive the business. It’s funny to say, but we’re really starting to do the kind of analysis we hoped to do when we began the business 14 years ago. Digital is just catching up with the types of statistical, modeling and customer analytics techniques we were using 10-20 years ago in database marketing. That’s not because we somehow forgot the techniques – it’s because the infrastructure, platforms, richness of data, and customer maturity just weren’t there. Tag Management Systems are a big part of reaching that level of maturity – a level where implementation is a small part of your analytics program – not the whole enchilada.