Hey Reader!

There was a huge shake-up in the SEO world recently that revealed something pretty alarming about our analytics data. Google made changes to how ranking tools can scrape search results, which caused a massive drop in impressions in Google Search Console across the board.

Brodie Clark has a great breakdown of the technical details, and it reminded me of something I posted on LinkedIn earlier this year about how you could clearly see spikes in impressions in GSC for low-volume terms as evidence of rank trackers artificially inflating those numbers. But I don't think any of us realized exactly how bad the issue actually was.

This whole situation got me thinking about a bigger question: how do we know if a visitor is a bot or not? Right now, GA4 tells us they filter out bots, and LLM crawlers don't execute JavaScript (yet), so most analytics setups are relatively safe. But it's only a matter of time before LLM bots start polluting our analytics data.

To start preparing for this future now, create baselines of your data. What is your typical session length, page depth, engagement patterns, and so on? When AI bots start executing JavaScript they’ll likely have different engagement patterns, and so we should (one hopes) be able to tell them apart.

If you have suggestions or thoughts on this, just hit reply. I'd love to hear what you're thinking about bot detection and data quality protection.

Dana DiTomaso Founder ​dana@kpplaybook.com​

Fix Your GA4 "Unassigned" Traffic Problem

One of the most frustrating things about GA4 is when productive conversations about data get derailed by GA4’s odd terminology choices. In particular, the Unassigned channel, which is where GA4 sends any traffic that doesn’t follow the channel rules.

I just published a step-by-step guide on how to rename and reorganize your Unassigned channel so that it can be named whatever you want it to be named and remove Unassigned from GA4 forever!

​Read the full guide →​

Why Full-Funnel Testing Should Be Your Priority

Kaitlin R. McMichael published a piece on Better Marketing on why you're not doing your job if you're not doing full-funnel testing. Now, the headline is a bit dramatic, but the core point is solid.

Most of us get caught up testing individual touchpoints like landing page copy, ad creative, and email subject lines. But ultimately, what matters is understanding how changes at one stage affect performance at every other stage. That headline change might improve click-through rates but tank conversion rates. You'll never know if you're only measuring (and reporting on) the first part of the funnel.

The Reality of Analytics Processes

The Analytics Power Hour crew tackled the processes of analytics and it’s a great listen! They dug into the gap between what we think our analytics processes should look like and what actually happens in practice.

What I appreciate about this discussion is the honest acknowledgment that most analytics work is messier than we'd like to admit. If you're struggling with inconsistent reporting, unclear data governance, or just feeling like your analytics processes are held together with bubblegum and hope, you're not alone.

Do LLMs Actually Use Google for Search?

Celeste Gonzalez at Ricky Roo investigated whether LLMs actually use Google to answer questions, and the findings are more nuanced than you might expect. This ties directly back to my earlier point about preparing for AI bot traffic.

The short answer is: sometimes, but not always in ways we can easily detect. This matters because it affects how we think about organic search strategy and what kind of traffic we might start seeing as AI tools become more sophisticated in their information gathering.

Start Small + Get Specific: What Marketers Get Wrong About AI

Join me and AI educator Britney Muller on Monday, October 21st, for a conversation that tackles one of the biggest challenges facing marketers today: turning AI ambitions into actual results.

Even seasoned marketing professionals are struggling to identify the right AI tasks and break them down into actionable workflows. You're not alone if you've ever asked an AI to do something like "improve this content!" and wondered why it didn't work out.

We'll share real examples of how breakthrough moments happen when marketers learn to leverage LLMs for what they actually do well, plus give you a practical framework for identifying your most valuable AI applications: the ones tailored to the unique tasks you actually do every day.

​Secure Your Spot →​

That's it for this edition of The Huddle. As always, if you have questions or want to share what you're working on, just hit reply. I read every email!

Want more? You got it!