Ways To Deal with, Sorry Cope With, Privacy
You know, even though Think With Google is touting there own stuff (and who doesn’t?), every now and again there’s a really useful piece. This is one of them, at least as a thought-starter. Covering three main strands of the future of digital marketing as we barrel towards a cookie-less world, it chats about how GA4 will save the day, especially in audience-building, before moving onto using enhanced conversions in Google Ads (I told you this was spruiking their services). Finally, there’s a brief allusion to using Consent Mode to uplift conversions via modelling.
So, essentially, we’re losing the ability to track users down to the nth degree, but in its place modelling, consent and predictive modelling will come charging to save the day. Maybe. And maybe that looks good on a deck in some snazzy offices, but I’m yet to be convinced the replacements are as good as the originals.
Read this, but with a weather eye, as you should anything from Google. Instead you should talk to me about measuring your business with my Analytics & Performance services.

Large Language Models and Long Posts

You had better put the kettle on. And the coffee-maker. And possibly make a sandwich. This is a very long look into large language models – one of the things that power the recent wave of AI. And for the most part it’s actually incredibly interesting, both for what it shows and what it doesn’t.
The thing it really shows is that language while complex, has a rough set of rules you can use to process the words you hear / read to form them into some semblance of sense. This is the kind of thing our brains have learnt to do relatively easily in most instances, unless partners are taking to us. So far, so good – this has been the basis of search evolution over the last 15-20 years. The thing it doesn’t show, or admit, is how over-blown most of Google’s statements on the wondrousness of its search algorithm have been up to now, and likely to continue into the future.
Great read. Enjoy it and learn. After that, come chat to me about SEO and Content Strategy and Creation.
Some Really Useful Email Automations
4 of these automated emails are really useful an absolutely critical to have for any ecommerce operation: Post-Purchase (or follow-up), Welcome, Abandon Cart, We Miss You. They are so critical, I can instantly think of a whole bunch of operations who didn’t / don’t do it, and for the life of me I can’t thin why. It’s not quite money on the table, but it’s definitely down the back of the sofa and eminently gettable. The last two emails I’m not so sure about: a birthday email is nice, providing it comes with a decent or targeted offer: ‘Happy Birthday come shop with us’, does not cut it as a birthday present; a browse abandonment is definitely a maybe in my books. Often I abandon my browsing because the product isn’t right, or I’m checking a detail, or I just don’t want to buy it right then. “hey you were looking at this, here’s 50 irrelevant products you might also consider’ is not a helpful interjection – sending me something so I can remember which products I looked at is useful.
Don’t just read this, but go check your Email / CRM workflows and see if you can integrate these into your own operations. Or have me look at them as part of my Email & Customer Relations consultancy services.

Finally! Facebook Spots AI, But Not In The Metaverse

Finally! Facebook awakens from its slumber and introduces AI into its panoply of tools. I say introduces, I mean “rolls out a smallish feature set to a small number of advertisers as a test run”. This is actually a useful set of tools, AI image cropping and ad text improvement among them.
What I’m confused by is how Microsoft, Open AI and others caught Facebook, Google, Amazon and everyone else napping and completely by surprise recently. It’s like nobody was at the wheel, everyone was in the back partying away. In place they have, or are in the process of rolling out semi-formed tools with a lot of catching up to do.
One thing that is unclear is whether Meta has dropped the Metaverse and is pivoting to AI (although it claims it’s been using it since 2005 – “like, before you were born, man”). Stories abound that Zuck has soured on the Verse after pouring many small countries’ GDP into it, but Meta is still pumping out PR about how glorious the future is. Hmmmm.
Still it’s a good time to chat about PPC & Paid Social advertising, or just good old Social Media Management.
INP is the new FID
FID is out INP is the new cool lad on the street! You what? Google’s Lighthouse and PageSpeed devs have been working on transitioning from a key metric in the PageSpeed algo FID: First Input Delay to INP: Interaction to Next Paint. This will start now and the transition will be complete in 2024 – roughly 12 months time.
For those who are not JavaScript cognoscenti, First Input Delay is a measurement of the time necessary from when a user first clicks an element on page, to the time the browser is able to respond to that input (hence First Input Delay, neat, huh?). Generally less than 100ms is good, over 300ms is poor. Interaction to Next Paint (who thinks up these snazzy titles?) is essentially an extension of this, in that it include the time for the responding action to complete and display on screen, ready for the next task to begin.
Imagine if you will an expanding accordion tab. You click it. The time at the start of the accordion expansion is FID, the time at the end of the tab expanding and you being able to click something else is INP. Easy, eh?
Of course, a far simpler solution is to stop overcrowding web pages with snazzy and useless JavaScript functionality which looks pretty but serves no purpose. Or pre-render the JS server-side, so you’re not clogging the browser or device memory and processors with useless compute tasks.
If you want to work on PageSpeed to improve Conversions, or SEO, or just for a better web property, talk to me about Technology and Development consultancy.
