Thread by @victorpan

Happy Friday friends. Today we're going to rant about the trials and tribulations of data accuracy for digital marketers, why data-based goal setting is hard, and how you can't fix both of these things forever, using the fabled "Bounce Rate" as an example.

1/n

A common mistake in digital marketing is not understanding, fundamentally, how website data or 3rd party competitor website data is being collected. How is SimilarWeb, SEMRush, and Ahrefs collecting data? How does GA, GSC, Bing, Pikwik, Adobe, HubSpot collect data?

2/n

I can't count the amount of times different implementations of Google Analytics goes awry - double tags, scraper site uses your GA ID, missing tags, non-standard implementations that weren't working, deprecated implementation, code in the wrong place...

3/n

The reason why a lot of SEO's get really good at troubleshooting these things is that our goals are often directly tied to traffic or "user engagement" signals.

Work hack? Increase traffic by properly attributing them to organic. There are worse things you could do.

4/n

Sometimes the mess is just enormous and you're working with a black box. Groupon pulled the switch in 2014, and found that up to 60% of their direct traffic was actually organic by de-indexing their whole website for 6 hours searchengineland.com/60-direct-traf…

5/n

And sometimes you're working on Page Speed, you thought the traffic increase was from that... but really....

"Speed = Better Analytics"
"More Traffic Recorded = "Traffic Increased"

slideshare.net/patrickstox/se… h/t @patrickstox

6/n

Good marketers want the data to be as descriptive to actual website visitor behavior as possible. Bad implementation creates the opposite effect. Before goals are set, you need to do a "good faith" clean-up. Sometimes you let certain metrics be useless. Me? GA Bounce Rate.

7/n

How did GA Bounce Rate become useless? Well, we started tracking scroll depth. That's an "action" that's triggered the moment a visitor scrolls. With that action the visitor is no longer a bounce. There are other similar examples out there in the wild.

8/n

As you get into the nuances of web analytics, you start to learn how to break or bias them - this includes 3rd party tools. Blocking AhrefsBot or SEMrushBot user-agents when they visit your servers? Yeah it's a thing.

Fun, but keep the knowledge creating value.

9/n

This is why it's very easy to feel uncomfortable with setting goals - especially when you're likely improving the website whilst playing the analytics consultant. Don't let that stop you.

Fixing bad data usually means you can see a % change. Adjust for that and move on.

10/n

This is true if there's a new adblocker/browser security/cookie consent law and banner that eats away on your website analytics. You simply have to accept it and move on (or use an web analytics platform that uses web logs for the most part).

11/n

We care about traffic and behaviors that influence $$$. It's easy to fall into the trap of "how do I increase this number" and forget the human on the other side. Myopically chasing "user experience" metrics can lead to dark patterns. darkpatterns.org

12/n

The next version of browsers, a new privacy law, a change in how web analytics providers "calculates" stuff are all things I've seen in my career. This is a problem that's imperfect and evolving. Adapt and move on. Create better user experiences as they your bottomline.

13/end

I reference these documents all. the. time. No Shame.

GA: support.google.com/analytics/answ…
GSC: support.google.com/webmasters/ans… and support.google.com/webmasters/ans…
SEMRush: semrush.com/kb/998-where-d…
Ahrefs: help.ahrefs.com/en/articles/78…

Remember, you're not alone when your analytics looks wacky.

Alright, back to the rabbit-hole:

https://twitter.com/victorpan/status/1299401354709000192