Hey, SEO Agencies: You’re Doing Keyword Research Wrong!

November 3, 2022.

In my quest to serve more clients, I’ve been interviewing with agencies looking to hire me outright or else “white label” my SEO/content marketing services. One of them gave me a “test” to take home and complete (which I share below).

Now, in theory I get it. Before you waste your time interviewing people who may or may not be any good at marketing, you need a quick way to weed out the less-talented. Hiring directors and everyone else involved are busy folks!

But I’ve seen these kinds of tests on UpWork and other job sites … A good SEO can tell when a test seems to have been based on best practices and “common SEO knowledge” from 10-plus years ago.

And that makes it hard. Hard because that’s when you find yourself purposely choosing the “wrong” answers because you can tell that the test-writers think they’re the right answers. And you don’t want to answer correctly when you know it’s going to count against you, because then you’ll have a 70% or 80% proficiency score for all the world to see—in a field where you know more than most! (And where you also know that there isn’t always a right answer, since the best SEO approach is often dictated by vast number of variables at play with the client, their industry, and their website.)

How NOT To Assign A Test To Potential SEO Partners

Here’s the test assignment I got from a web design company this week:

  • Review and provide keyword research for https://www.google.com/photos/about/
  • Please provide 15 keywords (and their global, english avg. monthly search volume) that would be a good fit for page optimizations
  • Please steer clear of branded keywords (such as google photos)
  • Indicate which 3 keywords from your 15 would be your final recommended target keywords for the content
  • Provide content optimizations for https://www.google.com/photos/about/. This should include drafting updated metadata and suggested page copy edits to incorporate your 3 selected keywords.

So What’s the Problem Here?

Setting aside the fact that the web design agency making this assignment should probably give folks tests that simulate actual, real-world client situations (sorry, guys, you won’t be tasked with helping Google rank their own pages any times soon … AND IF YOU WERE, IT WOULD BE AN ENTIRELY DIFFERENT APPROACH FROM THE NORM).

But here’s the thing … One would never go seeking 15 keywords to optimize a single page for—especially not an existing page.

The correct approach would be to create new content optimized for the keywords surfaced by the ongoing process of keyword research (more specifically, user observation and topic modeling, which brings us to the next point …)

Keyword Research Has Changed, And Most Agencies Don’t Know It

This focus on specific keywords, driven primarily by reported search volume, is an obsolete approach to planning, creating and ranking content. It ignores the fact that Google’s AI now looks at what searchers do AFTER their initial search. Here’s a typical user journey: 

  • Step One: “High volume” keyword is searched
  • Step Two: User does not like results
  • Step Three: User performs additional search(es) until finding the content they need

In this scenario a supposedly “low volume” phrase would have provided a better user experience and would be more likely to draw the ideal user to the site than a high-volume “head” term.

Search any specific keyword phrase and you’ll see that many (often most or all) page-one results do not even have that phrase … That’s because the AI has processed vast amounts of user data and determined what the user “really” wants.

Today, good SEOs are not looking for specific high-volume keywords to stuff onto the page and metadata; they are clustering topics based on intent and creating content that satisfies all related queries.

For more, see Bernard Huang’s discussion with Ross Hudgens at Siege Media:


… as well as his Twitter thread about this:



As always, hit me up if you have any ideas for future topics.

Leave a Comment