Weekly Finds – Week 43, 2020



Updated on November 15, 2020
Topics:
3 min well spent

Listen to Weekly Finds

Quick Finds

SEMrush and Yoast Announce Integration Partnership

  • SEMrush now integrates with Yoast, a very unexpected partnership.
  • WordPress already drives > 33% of the web. Yoast is one of the most popular WordPress plugins, SEMrush one of the most popular SEO tools.

Introducing the new Google Analytics

  • Google announced GA 4
  • Lots of machine learning for insights and predictions
  • GA will even be able to predict churn and LTV
  • Measure audiences across apps and the web
  • Track Youtube actions
  • Another showcase of Platform Confluence: measure across the whole Google ecosystem.
  • Apparently, Google is trying to fill the gaps from vanishing cookies with Machine Learning

How Google autocomplete predictions are generated

  • Google explains how autocomplete works
  • Signals include commonly searched combinations, location, language, search history, and freshness

How AI is powering a more helpful Google

  • Instead of a developer conference, Google published a 20-minute Apple-like summary of what’s coming to search called “search on 2020”. The announcements were heavily discussed, though none of them should have come as a surprise.
  • BERT is now used in almost every English query.
  • Google will index passages instead of whole websites. Cindy Krum called it #fraggles.
  • You can now hum songs to Google and it tells you what it is.
  • Google wants to show a greater diversity of search results but it’s unclear how. They say they use neural nets to understand sub-topics of interests – whatever that means.

Weekly Finds

Measuring the returns from content is hard (which is why I published a full guide around it on Monday). Most companies measure conversion with last-click attribution (= user converts on the current page), but this creates a flawed view.

Instead, GA’s Model Comparison Tool provides first-click attribution, and the folks from Grow and Convert explained how to interpret the data right. GA’s time lag report shows you how many days it takes for people to convert. Path length shows you how many interactions it takes. The trick for comparing attribution models between pages is to select “landing page URL” as a dimension.

Also, cross-device tracking doesn’t reaaally exist yet, so tons of conversions are attributed to your homepage. The same is true for interaction gaps longer than 90 days or Word of Mouth (even though Reforge recently introduced the word of mouth coefficient).

Speaking of interpreting data, Certain SERP features like image packs show website URLs but lead to another Google search page (in this case, image search). That leads to a lot of confusion because Google Search Console counts the impressions and rankings for such URLs but, of course, you don’t get the clicks. My key takeaway is that Search Console becomes less valuable as Google moves further away from a “10 blue links” design of the SERPs.

More SERP Features make it more important to maximize your keyword CTR. There can be many reasons for low CTRs. Aleyda gives us a flow chart to help decide what to do about it:

  • Optimize meta-title and description
  • Go after Rich Snippets and Featured Snippets
  • Differentiate pages that might compete against each other more clearly

In many cases, adding or refining content goes a long way already. The article features a cool tool to compare CTRs in different industries and for different types of keywords that I didn’t know is AWR’s CTR Study.

Refining content is not just for SEO but also copy. Joanna Wiebe, founder of Copyhackers, explains how to rock landing page copy:

  • Focus your copy on value for the user, not your company
  • Varying sentence length creates “texture”
  • Add a strong adjective for “shine”
  • Address objections with “even if”
  • Create a landing page separate from your homepage

Good copy reflects the customer’s stage of awareness, from (problem) unaware to product-aware.

Discuss!

This site uses Akismet to reduce spam. Learn how your comment data is processed.