Hamsterdam Part 61: Weekly SEO & AI News Recap (6/3 to 6/9, 2024)

By Ethan Lazuk

A weekly look-back at SEO & AI news, tips, and other content shared on social media & beyond.

Opening notes, thoughts, and other musings:

• Welcome to another week of Hamsterdam!
• I’m publishing this recap a little later in the day. Hope that’s ok!
• I greatly appreciate you reading it, though. 🙂
• There’s a LOT to get to this week. So let’s hit it!

Feel free to jump ahead to the news recap if you’re in a hurry.

If you’d like these delivered every week, subscribe to my free newsletter here.

---

Marketing word of the week: “Demographics”

Demographics refers to particular subsets within a population that share attributes in statistical data.

Huh?

In the context of marketing, demographics are groups of people who share characteristics, like age, gender, income, or education.

Oh, ok.

They are used to identify a target audience or segments for tailored messaging and can inform buyer personas, channel selection (organic search, social, email, etc.), and content.

Specifically in digital marketing, demographics can determine targeted advertising with Google Ads or paid social, as well as organic strategies based on personalized campaigns.

In the context of SEO, demographics might influence keyword research, content creation, and even link-building (digital PR) outreach. They can also impact page experience decisions (like UX and design).

My 2 cents on demographics are …

I think the way they’re are commonly thought of is anachronistic — i.e., old school.

I hope to write more Hamsterdam Marketing articles involving social sciences topics soon, but in our first one about personas, we talked about using data and avoiding stereotypes.

Aside from eligibility (for example, “our service is designed for women age 25-45 living in California”), I think demographics should go out the window and instead be based on dominant shared traits, as determined by machine learning models.

These would be things like, “eats plant-based foods” or “enjoys living near water.”

For example, I started watching this free TV app called Tubi.

It didn’t have a user profile for me, so all of the ads were for low-cost products or how to save money. You know, the kind of topics you’d stereotypically expect to appeal to someone who watches a free TV app.

When in actuality, I just wanted to watch full episodes of Dog the Bounty Hunter for nostalgic purposes!

Maybe they would have served me up some more relevant content with more of my history, but that’s where topics like user interaction data and personalization, including for LLMs, come into play, and where I think the future of search is likewise headed.

---

AI word of the week: “Decision boundary”

A decision boundary is a separator between classes, either for binary or multi-class classification problems, in a machine learning model.

It helps the model determine which class a new data point belongs to based on its features or characteristics.

More technically, a decision boundary is called a problem space, where the output label of a classifier is ambiguous, i.e., it’s the dividing line between different predicted classes.

A classifier (algorithm) learns a decision boundary during training and then uses it to make predictions.

For example, if we trained a classifier to say whether a document was “helpful” or “unhelpful” based on labeled examples, it could then predict if a new document was helpful given where it fell relative to the decision boundary, as based on relevant attributes.

If the model learns the training data too well, overfitting can occur, where the boundary becomes highly complex and irregular, reducing its accuracy. The same can be said for underfitting.

What if data falls on the decision boundary? It depends!

The model might use random assignment, provide a confidence score, adjust itself, or use a soft margin, where it tolerates some data points being on the wrong side with the goal of a more generalizable model.

When it comes to classification algorithms, decision boundaries are a handy concept to know.

---

This week in SEO history: “DMOZ” (1998)

On June 5th, 1998, two developers for Sun Microsystems named Rich Skrenta and Bob Truel created a non-commercial and multilingual catalog of websites called the Open Directory Project (or ODP).

It’s better known as DMOZ, named for its original domain: directory.mozilla.org. Here’s a screenshot from 1999 via WayBack Machine:

The content of the catalog was maintained by a community of volunteers, and AOL owned it for most of its history, until the project ended on March 17th, 2017.

It’s obituary (sort of) was written on Search Engine Land by Danny Sullivan:

“DMOZ — The Open Directory Project that uses human editors to organize websites — is closing. It marks the end of a time when humans, rather than machines, tried to organize the web. …

DMOZ was born in June 1998 as ‘GnuHoo,’ then quickly changed to “NewHoo,” a rival to the Yahoo Directory at the time. Yahoo had faced criticism as being too powerful and too difficult for sites to be listed in.

It was soon acquired by Netscape in November 1998 and renamed the Netscape Open Directory. Later that month, AOL acquired Netscape, giving AOL control of The Open Directory.

Also born that year was Google, which was the start of the end of human curation of websites. …

DMOZ will live on in one unique way — the NOODP meta tag. This was a way for publishers to tell Google and other search engines not to describe their pages using Open Directory descriptions.”

– RIP DMOZ: The Open Directory Project is closing (Danny Sullivan, SEL)

The site today is 400 bad request error.

There’s another website called DMOZTools.net that says it’s “an independently created static mirror of dmoz.org and has no other connection to that site or AOL,” but it appears to be broken.

There’s also ODP.org, which says, “As former directory editors, we were sad to see DMOZ announce their closure on short notice, so we decided to use their RDF file to preserve the work & create a static snapshot of their web directory.” It’s not HTTPS, so I won’t link to it here, but it does appear to work.

The ODP.org About page has some additional history, including DMOZ’s influence on Wikipedia:

“AOL did not state a reason for closing the directory. However the directory (like other general purpose web directories) saw declining influence and usage as user search behavior shifted away from web directories to search engines. This shift was accelerated by the shift to mobile devices, as searching on mobile devices made instant answers a far better user experience than a list of websites. Many major search services like Google & Bing offer instant answers & knowledge panels in their search results which aim to answer frequently asked questions.

While searchers are still interested in a diverse range of content, online usage behavior has concentrated on a few major portal sites like Facebook, Instagram, YouTube, Amazon, eBay, Netflix, Google, Twiter, and TikTok.

The web directory model of listing websites struggled to compete on relevancy with search engines that quickly surfaced the most relevant pages (or even the most relevant answers) from widely trusted websites.

Larry Sanger claimed DMOZ was an inspiration driving the launch of Nupedia, which led to the founding of Wikipedia. Content from Wikipedia is used widely by search engines in instant answers and knowledge graph results. Search engines pay to license the content from the Wikimedia Foundation.”

– ODP.org About page

It’s an interesting history DMOZ has, born as a non-commercial and volunteer-supported project.

Speaking of volunteering and projects …

Let’s get to our introduction this week, talking about hard conversations.

---

Introduction to week 61: “hard conversations”

Location, location, location.

There’ve been some good local SEO articles published since the Google API Content Warehouse leak (that tiny event mentioned in last week’s intro). 😉

Andrew Shotland wrote a post for Search Engine Land.

Here’s a snippet of it:

“In theory, if you have titles throughout your site that are relevant to the user queries (a.k.a. ‘relevance’) + a strong localityScore (a.k.a. ‘proximity’), you should have a decent shot at good local rankings, assuming you are prominent enough.”

– Andrew Shotland, SEL

That made me think about my own situation.

I’ve run my business since January and also added much of my site’s content since then.

I put my site into Google’s NotebookLM experiment last night as a test, and this was the summary:

That’s pretty good. As you can see, it says I work with businesses “across the globe.”

Now for Search …

I created a service page for Arabic-based businesses, which basically says I’m an American SEO with cultural knowledge of the Arab world.

I haven’t spent as much time on the page as I’d like, and you can see it ranks for likely location-based queries that don’t apply:

On the other hand, a lot of my incoming calls these days are from Central Florida businesses, which makes sense because I’m pretty easy to find in Orlando:

I’m happy about that, but I also plan to update that page this week.

Why would I do that if it ranks well?

Well, as I often remind businesses, it’s not about the keyword rankings; it’s about what they get you, business wise.

Which brings me to this week’s topic of “hard conversations.”

I’m fairly new to doing consultations with small businesses. I’ve helped plenty over the years, but that’s after they were qualified and the whole nine.

This scene from The Wire came to mind …

Stringer Bell: “Now, you need to call Mr. Goose, Mr. Faucet, whoever the f*ck you gotta call because I gave you money to run, remember that?”

Clay Davis: “It ain’t like no drսg deal, String. Hell, man, you don’t put your money on the street and have it come right back. It don’t work like that.”

Stringer Bell: “Oh, it don’t, huh?”

Clay Davis: “Patience, my man.”

– The Wire, Season 3, Ep. 9

Ok, bad example, haha …

But seriously, many of these small Florida businesses that I’m speaking with have the same story — they paid someone who wasn’t a knowledgeable SEO professional and got burned in one way or another.

By the time they come to me, they’re fed up and skeptical and like, “Why didn’t this work?!”

It’s never one or two fast and easy things. It’s invariably a collection of micro-improvements that will add up over time.

It’s a marketing strategy …

I don’t mind spending a little effort to put together recommendations these businesses can bring to their teams, whether we work together or not.

That’s just being a decent human being, imo.

But truthfully, it’s still a drop in the bucket.

Even as I was writing this — like literally — a Screaming Frog message popped up to tell me I’m running out of memory on a crawl.

It was for a small business that emailed me. I looked at their site and saw maybe 10 pages.

Screaming Frog found 150,000 URLs …

These are hard conversations to have.

But all we can do every day is wake up, brush our teeth (and floss!), and be honest about our capabilities and true to our ethics.

I just hope more of the rocket ship emoji crowd remembers that. 😉

Thank you for supporting Hamsterdam and the cause of SEO & AI learning.

Missed last week? Don’t worry, I got you! Read Part 60 to catch up.

Other great sources of weekly SEO news:

• The SEO Weekly – Garret Sussman, iPullRank
• SEOFOMO – Aleyda Solis
• Weekly Video Recaps – Barry Schwartz, SER
• Weekly SEO News YouTube channel – Olga Zarr, Seosly
• Niche Surfer – Yoyao Hsueh

---