The Downside of Knowing (Almost) Everything

As an analytics group, a portion of our time goes toward tagging – deciding what should be tagged, how to tag it, and the metrics those tags will provide. 

As analysts are wont to do, we like the linear approach of determining business objectives, followed by developing key performance indicators (KPI), wrapping up with a little metric mapping where we determine which metrics are needed to report accurately on a KPI. (A metric map acts as a checklist to ensure that the appropriate site components are tagged, and provides guidance on the methodology of collecting data within those tags. For example, is the metric simply that a video was played, or that a video was played for 25% of its total length while being viewed on an iPad on a Saturday afternoon?) At this point, Have metrics list, Will tag.

Too often, though, a tagging request will come to us with a subline of, “We want to know everything we can.” The more bold clients include this phrase in their original request. Others wait, and casually mutter it under their breaths at a random moment during the analytics kick-off meeting, regardless of whether KPIs have been developed or not. If KPIs do exist at this point, the understanding is that KPIs will be included in a report, but that everything else is a “nice to have” and “could be useful.”

Really? An umbrella is nice to have and could be useful. A giraffe is nice to have and could be useful (no, really, I’m short so I’d get a lot of use out of this).

The real purpose of tagging is to populate an analytics platform with the necessary metrics to report on KPIs. With this data available, quantitative findings can be had, actionable insights provided, and business decisions made.

I once had a client who mandated the “tag everything just in case” approach. We were able to cut them off at a unique identifier on every tag (such as the same button on three different pages with each being given their own unique tags), but every button on that site had a button tag on it.

The first couple of days were like Christmas (or your holiday of choice). Data was flying in fast and furious, reports were being generated, congratulatory emails sent. Fast forward six months when we were wondering why the site continued to load slowly, why some reports required so many manual calculations, and a quiet confusion as to why some of the tags that had seemed so important before had now not been used since implementation.

Downfall of Too Many Tags 

 

Slowness of site – a single tag takes only a few milliseconds to load. However, these milliseconds quickly add up to many full seconds that add to the load time of a page/site. One NY Times article found that visitors expect a page to load within two seconds, and that a difference of 250 milliseconds of load time between competitor sites can determine which site a visitor will return to. Mobile web views get a little more leeway, but the acceptance of differences between desktop and mobile web experiences is quickly narrowing.

Introduction of risk – Any time code is manipulated, even by something as small as tags being added, an element of risk is introduced to a site’s functionality. This means greater responsibility for development teams, analysts and quality assurance and user acceptance testers not just at time of launch, but for future monitoring as well. What were once acceptable tag lengths and components, historic browser functionality, or new versions of analytics platforms, can all impact how data is captured and its accuracy in the analytics platforms. (Let’s be honest though, follow-up monitoring and maintenance on sites could find many of these issues before they drag one for too long, but unfortunately, once a site is launched it is typically considered “finished.”) This is not to say you should not tag. This is to say that a site and tagging are never ‘done’.

Loss of statistical validity/reliability – Depending on tag configuration and methodology used, it is possible for the data to lose its validity and/or reliability. Extreme caution needs to be used when making decisions against invalid or unreliable metrics as there is no guarantee the performance of a small group of defined actions this week will be mimicked again the following week.

Increase in resource needs – Even when using a tag management system, someone is going to need to put the tags in, validate the tagging, and monitor functionality. A simple point and click, repeated over dozens of tags, will add up in time. This continues to hold true with tag managers – even though there is a greater rate of flexibility, depending on the site build, a tag could have multiple customized conditions that each take time to build.

What to Consider When Developing a Tagging Strategy

What will help in making business decisions – Determine which metrics are going to be valuable and become part of a larger KPI against which decisions, recommendations, and actions can be made. Succinctly, what I need to know.

Contextually-relevant metrics – There is almost always some leeway when developing tagging in that there is no restriction saying that if a metric does not tie back to a KPI it can not be tagged. In addition, there are other metrics that are helpful and can add a contextual relevance to the performance of KPIs. For example, a KPI may be transactions from an online store. Contextually-relevant metrics could include page pathing, video segment timing and markers, or campaign tracking. These would provide additional insights into why a KPI may be performing well or poorly.

Upon Later Review of Tagging and Reporting

Once the tags are in place and some time has passed, a reassessment is always recommended.

What is tagged that is not being used – A review of the components that have been tagged, but that have been found to not be needed. Overall, a clean-up of the tags to reflect what is now of importance within the reports is a great idea.

A subcomponent of tags that is not being used is a change in methodology. Here, you need to consider things like: are actions still being counted the same way? Is a click on the Submit button considered a success, or has a Thank You page gone live that we can use to get more precise form completion values?

What am I missing – Gap analysis of metrics that were not included in the original tagging, but are now to be included in the reporting. When producing reports for some time, it is inevitable that someone will look at a metric and ask, “Why?” Those types of questions should be documented and incorporated into the next phase of taggi
ng review.

So What To Do

Analytics is always going to be a part of the marketing landscape. However, just because we can track something does not mean it is always the wise choice to do so. Tools such as Tag Manager have greatly increased the flexibility of tagging, but should not be used without conditions (and you get a tag, and you get a tag, a tag over here…).

Be mindful of what you need to know versus what you would like to know. As long as tracking is in place, you are ahead of the game. We just don’t want you to stumble before the finish line.

 

« Prev Article
Next Article »