My early review of How to Use a Fork by Orlando Swayne is live. I found it endlessly fascinating and moving. @picadorbooks #bookreview #bookblogger #booktwitter #medicalbook #OrlandoSwayne #Neurological booksbyyourbedside.org/2026/…

When your whole thing is being a growth agency it’s pretty exciting when two of the 15 clients you worked with last year made it to onto the Times Top 100 Fastest Growing Brands list Huge well done to MOTHER ROOT ✸ and Tilly Sveaas Looking forward to seeing what % of our roster makes the 2027 list 😊

Sometimes you see a demo and you’re like ‘wow, the future is here’. This is not that demo…

Holy shit… are you paying minimum wage? You have an entry level position, and that person works a ‘normal’ week - 9-5 plus an hour or two of extra work per day. 48 hours per week is £31,746 I think some startups have people working longer hours on less money…

This is what a properly built persona system looks like in practice. 3 micro personas. 4 angles each. 4 vehicles each. 48 genuinely distinct ads from a single macro persona. Take one macro persona: "The Conscious Parent." On paper, one audience. In reality, completely different people with completely different fears, motivators, and trigger points. That's the micro persona layer. And it's where most creative strategies stop being strategic. P1.1, P1.2, and P1.3 sit inside the same macro. Same broad demographic. But each one needs a different hook, a different setup, a different cognitive bias tapped to actually convert. An ad built for one will not land the same way with another. If you're writing to all three with the same message, you're reaching none of them properly. Here's what building to that level of specificity actually unlocks. Each one a different signal. Each signal giving Andromeda something new to work with. Each piece of creative capable of reaching a person the last one couldn't. That's how incremental reach actually grows. Not by producing more, but by covering more psychological territory with what you produce. Most accounts have far more creative territory available to them than they realise. The personas are there. The angles exist. They just haven't been mapped and briefed against each other yet.

There’s a well-known phenomenon in the facial aesthetics literature whereby “average faces” (that is, faces formed by superimposing many faces atop one another) tend to be more attractive than the average person. This may be counterintuitive, but it makes sense when you consider the following: Individual faces are all slightly flawed, from a beauty perspective, in idiosyncratic ways. And when you average lots of faces, you average out all of these minor issues. So, an “average face” is errorless and looks quite pleasant as a result. However, another thing you’ll notice about these “average faces” is that none of them could be models. They’re more attractive than the average human, yes, but less attractive than the most attractive humans. This is because extremely attractive faces tend to have certain features that are, mathematically, extreme. (For example, male models tend to have lower-set brows and larger jawbones than you would see in any average face.) Recently, I have begun to wonder if LLM-writing faces a similar challenge. It’s always “more attractive than average,” because all of the flaws of normal human writing have been averaged out. But it's also missing the unusual taste and style of the best human writers I've read. In my experience, it's only ever 85%-good; like an "average face," it's never flawed, but equally, it's never exceptionally beautiful.

What will be the impact of AI on jobs? I can’t tell you, but if you put your assumptions in here: claude.ai/public/artifacts/6… then it will calculate the impact over different time horizons.

Do you think at some point, saying you use ChatGPT will be like having a Hotmail email address?

Depends on the company

But also, why not 4:0? Because you could probably spend more on marketing, at a higher CAC and have higher overall profit. Ps. LTV isn’t revenue, it’s CM2 (revenue less costs of goods and delivery.

Why 3.0 LTV:CAC? Lots of people have heard they need to shoot for 3:1 in their LTV:CAC ratio but few seem to understand why. This is the explanation 👇 - At 1.0 LTV:CAC, you are just breaking even and not really creating incremental EV from getting the Nth customer, pretty much just recycling cash. - At 2.0 LTV:CAC, buying one customer returns 2× what you spent. The surplus is the “fuel” that can fund additional growth so acquiring customers actually starts to compound. - At 3.0 LTV:CAC, buying one customer creates 2× CAC of surplus. If payback is reasonable, growth begins to compound in a durable way, and EV can start scaling rapidly. More insights in newsletter

I flip cards of a randomly shuffled 52-card deck over one by one. At any point you can say "STOP" and you win if the next card flipped is black. What's your probability of winning? It turns out the answer is 50%. It feels like waiting to collect info should help but it doesn't - eg if you wait til the last card you will know for sure what it is, but it will still only be black 50% of the time. The analogy in startups is: waiting longer makes you more likely to know if you're right. But not more likely to be right. Just do things!

I really loved the idea of the iPhone asking cold callers why they are calling. But no one leaves a meaningful one.

The Report is live. Check out our insights, creative overview, and predictions for 2026 here: weareballpoint.com/resources…

The number of times I see numbers like this for businesses for sale is crazy. Slight decline for 3 years, but it’s all going to be good next year… hmm.

Playing around with the Spotify mix feature (that I’ve been asking for for ages). open.spotify.com/playlist/3d…