Getting Started without a Data Layer

When people first begin thinking seriously about implementing a tag management system (TMS), they often ask me if they should spend time creating a data layer beforehand, or if diving right in is the way to go.

My response is always the same: you’re asking the wrong question.

The most important thing to consider when beginning the conversation about TMS is how long you can continue to wait. Are you collecting any data at all? If so, can you live with your current data collection method for another six to 12 months without losing momentum and/or customers?

Of course, it is my firm belief that any time spent creating a data layer, one that helps you organize and structure the metadata of your website or application, will never be wasted time. It’s never a bad idea, and it will certainly serve you well in the future.

That said, a data layer is not the single foundation upon which all measurement must be based on, nor is it critical that the data layer is constructed before you do anything else.

Of course, there are people who disagree with that assertion.

Talking about the need for (and the timing of) data layer construction is not unlike debating which kind of barbeque sauce is the absolute best in the southeast. You’ll get a lot of discussion, sure, but there’s almost no chance that you’ll get a consensus by asking a dozen random people.

Put more simply: it’s a never-ending debate with deep-seated beliefs and historical superstitions on all sides. Even though we know that Carolina mustard style is the absolute best, we also know that there’s zero chance that everyone will eventually agree.

The problem is that data layers are being touted as the number one panacea to cure all of your digital measurement woes when in truth that is a very, very narrow view of the digital measurement ecosystem. To put it another way: the tastiest sauce is not necessarily always the right thing to put on your food.

Just like any IT project that you undertake, it comes down to timing and tradeoffs. If it’s going to take you nine months to implement a data layer, should you wait on any other measurement implementation until that’s done?

Of course not. You wouldn’t wait nine hours to eat just because the sauce isn’t quite finished, would you? You’re hungry now.

Sometimes, you’ve got to look for ways to build a solution that can grow and mature with your existing infrastructure over time.