This week the digital excitement knows no bounds as we talk about link-building in 2024; LinkedIn and their AI-infused collaborative articles; fixing unassigned data in GA4; the demise of four attribution models in Google Ads & Analytics and; a long, exciting piece on how retrieval systems work for AI-driven Search Generative Experience and the like. What a time to be alive! Dive in and enjoy this week’s Digital Update from JB.

Link building still ain’t dead.

This is a handy reminder that even now in late 2023, heading into 2024, links are still important for SEO. 

It’s interesting how things have moved from link-building into link-earning over the last few years – despite the fact that really, it’s always been about gaining links because you offer something people want to link to. 

The simple fact of the matter is that Wikipedia has never built a link, nor created content to earn a link, and yet, it’s probably the most linked website on the planet. 

Admittedly, if you are the 854th site in the niche, then you are going to struggle to gain links in a very natural & authentic fashion, although it’s not impossible – just think about how many sites / apps have been built after you thought something was completely done and dusted. 

So, try not to buy links, try not to get all sorts of low-value links, create good content which is worth a link, and build networks with people who you would link and who would link you back.

Give me a shout about SEO, link-building and creating great content strategies.

LinkedIn Collaborative “Articles”

LinkedIn is launching / promoting  / reinvigorating what I think is its 18th creator / content  / information platform: Collaborative Articles. 

These are articles sourced initially by AI and then somewhat rounded out by collaboration input from people who may or may not be subject matter experts. LinkedIn is supposed to be giving them greater visibility in search, the feed and in notifications. The benefits for the contributor include greater visibility for your contributions and some alleged SEO benefit. 

The articles that have popped up in my feed have been a definite mixed bag. The AI side of it has usually been (like a lot of AI output) about 60-80% right, but it misses the nuance of a situation which a proper expert can bring out. Annnnnd then you have a real mixed bag of comments in the follow ups. 

The other massive issue with collaborative articles is that I see them in my feed and then they disappear. I know I’m supposed to respond instantly, but that’s just not me. Oh well, it’s probably for the best. 

So, second-rate content being puffed up by second-rate contributors, what could possibly go wrong? (Remember my AI Loop). 

Social Media is fun! Nearly as fun as content creation for it, or advertising on social. 

Fix Your Unassigned Data in GA4

The architecture of GA4 occasionally shows its strict app roots, where a capital can make all the difference. In this case, if your traffic source, or medium does not match a strict set of definitions, or regular expressions, you will find your traffic going to Unassigned. This is the GA equivalent of it throwing its hands up in the air and just giving up. Note, it is not the dump bin of direct. This is traffic that has a source, but GA doesn’t recognise it. 

Generally, the way to fix it is to correct the issue causing it in your UTM tags – read the GA4 guide to channels and sources and take your UTM creation strategy from there – making sure you observe the rules around exact matching. 

It is normal for Google not to back fill this information, so creating proper tags will cause Unassigned to drop over time, but you won’t have historic data. One thing that is identifiable is the Campaign name can be used to identify and analyse traffic, but it does make it more difficult to see a channel as a single entity. 

I’ve helped clients work this analytics & GA4 oddity out before, contact me if you want help. 

Four Attribution Models Pining for the Norwegian Fjords

Google Ads & Google Analytics are ditching the four attribution models most people knew about, even if they didn’t really use them. First Click, Linear, Time decay and Position based are going to way off the dodo, heading off into the sunset, never to be seen again. 

When I’ve talked performance and conversions with clients, we’ve often had a quick look at these models, maybe even clicked into them once or twice, possibly fiddled with the settings and then put them away for another six months. 

The mothballing of these is caused by the above and Google’s AI-infused Data-Driven model being its replacement. It’s not entirely clear how this model works we just have to “trust” Google, because that’s what we all do, right? 

If I’m honest, the figures in the data-driven version don’t look too bad. My general belief is that when someone is fiddling with an attribution model, it’s usually to their advantage. Google Ads seems to convert well all of a sudden. 

Let’s talk conversions, analytics & performance and Google Ads any time you like. 

SGE & Retrieval Augmentation Generation

Have you got half an hour to spend? Then this is one of those times to take a bit of a dive into how information retrieval kind of works for AI / SGE (Search Generative Experience). 

Of course, all of this is changing the paradigm etc and search is forever about to be changed and yada, yada, yada. Maybe, or maybe not.

I’m still struck by the realisation that we are the 1% in the rarified atmosphere of having an interest in search and tech. Most people don’t care, and don’t care to know either. AI is already, or will pass them by without them noticing. Something like 40-50% of businesses still do not have a website (admittedly I think that’s a rubbery figure created for shock purposes), but the general point stands. If the web, coming at us since the mid-90s, hasn’t penetrated that far, then what hope does AI have? 

Now, back to the article. I actually enjoyed it because I like to read about retrieval systems. But, if you have any inkling of how AI works, it is not a surprise, with a few layers of retrieval, composition, checking and screening. All great and shiny and fabulous. Until you see the dross that AI puts out, and yes it’s great for some things, but it has a little way to go yet. Perhaps in web terms, AI is now at the My Favourite Pizza Toppings stage of development, ready to move to MIDI music and dancing chipmunks. 

Let’s talk SEO, technology, AI and development. Call me. 

State of SEO 2017 Machine Learning and AI

Latest Digital Marketing Articles

In the last week, we have also written these three articles about my VCMO service (Virtual Chief Marketing Officer), Analytics & Performance, and Email & CRM. Use the links beneath to read them if you missed them the first time around :-)

Belmore Digital Square Icon