Which Claude Opus 4.8 settings should marketers use?

Transcript

Claude just shipped a dial that's going to quietly cost most marketers either quality or money — and almost nobody is going to set it right.

It's not a new prompt. It's not a new tool. It's a setting. And once you understand what it actually controls, the way you use AI for marketing work changes that afternoon. I'm going to show you exactly where to put it for the work you actually do.

Here's what dropped. Opus 4.8 — the new top model — now gives you two controls right in the interface: an Effort slider, Low through Max, and an Adaptive thinking toggle.

Effort defaults to High. The official line is simple: higher effort means more thorough answers, but it's slower and it burns through your usage limits faster.

That sounds like a quality knob. It's not. It's a budget knob. And that reframe is the whole video.

Most marketers think about AI output quality at the prompt level. Better prompt, better answer. True — but incomplete.

There's a second layer now: how much thinking you're paying for on every single message. And here's the trap. If you leave it cranked, you're spending deep reasoning — slower responses, faster limit burn — on tasks that need none of it. Rewriting a subject line does not require the model to think for thirty seconds. But you'll wait for it, and you'll pay for it in usage.

The pro move is to treat effort like ad spend. You don't put your whole budget behind a retargeting banner. You match the spend to the value of the task.

So here's how I'd map it to a marketing team's actual week.

• The Low to Medium setting — your execution tier. Subject lines, ad variations, social captions, reformatting, cleaning up copy. High-volume, low-reasoning. Run it here and you'll work faster and stretch your limits dramatically.
• The High setting — the default, and your daily driver. Newsletter drafts, positioning, campaign architecture, prospect research. This is where most strategic work lives.
• Extra and Max — reserve these. This is for the genuinely hard, high-stakes single shot. A full analysis you want to nail in one pass. A messy data problem. A complex build. Don't default here — you'll hit your limits fast for almost no gain on routine work.

Eighty percent of marketing AI work is execution, not reasoning. If your team is running all of it on High, you're overpaying on every task and getting nothing extra for it.

Now the toggle. Adaptive thinking.

When it's on, the model only reasons deeply when the task actually calls for it. Simple lookup? It answers straight. Complex multi-step problem? It thinks first. It's auto-throttling the effort for you, message by message.

My recommendation: leave it on. A normal working session bounces between "fix this one line" and "build me a launch plan." Adaptive thinking means you're not wasting deep reasoning on the trivial turns in between.

The only time to switch it off is a dedicated strategy session where you want every question — even the small-looking ones — getting full horsepower.

The marketers who win the AI era aren't the ones with the cleverest prompts. They're the ones who understand their tools at the control level — not the surface level.

Knowing when to spend compute and when to conserve it is an operations skill. It's the same instinct as knowing where your budget actually moves the number.

This is what I mean when I talk about building marketing operations that hold up. The model will keep changing. The skill of matching your resources to the task does not.

So do this now:

1. Drop your default to Medium for execution work. Notice how much faster you move.
2. Turn Adaptive thinking on and leave it.
3. Pick the one task each week that genuinely deserves Max, and save it for that.

I'm Nick. I write Systems & Signal — a newsletter on building marketing operations that survive the AI era — and on this channel I break down moves like this one in plain language, for marketers who'd rather understand what's happening than panic about it.

Subscribe, grab the newsletter in the description, and I'll see you in the next one.