AI in the growth engine: what's actually working in 2026

Somewhere between "AI will replace all marketers" and "AI content is soulless garbage" lives the boring, useful truth. We run growth programmes for brands every day with AI woven through the stack, so consider this a field report: what's genuinely working, what's overrated, and what hasn't changed at all.

Working: the research grind, compressed

Audience research, competitor teardowns, review mining, keyword clustering, transcript analysis — work that used to eat the first two weeks of any engagement now takes days. Not because the thinking is automated, but because the reading is. We feed hundreds of customer reviews into an analysis pipeline and get themed, quoted, prioritised insight back in an afternoon. The strategist's job starts where it should: at the "so what?"

Working: creative volume for testing

Performance marketing has always been a volume game — more angles, more hooks, more variants to test. AI has collapsed the cost of variants. We can take one strong creative concept and produce twenty disciplined variations for testing in the time it used to take to brief three. The winners still surprise us regularly, which is exactly the point: cheaper variants mean the data picks the winner, not the loudest opinion in the room.

The crucial caveat: volume without a strong concept is just noise at scale. AI multiplies the quality of the input. Garbage in, twenty variants of garbage out.

Working: lifecycle that actually feels personal

Email and retention flows used to be segmented by crude buckets — new, active, lapsed. With AI in the loop, messaging adapts to what customers actually browsed, asked, and bought. One client's win-back flow now references the specific product category a customer went quiet on, in the brand's voice, generated and QA'd in a human-reviewed pipeline. Reply rates roughly doubled. Not magic — just relevance at a scale humans couldn't manually write.

Overrated: fully automated content publishing

The "set it and forget it" AI blog machine produces exactly what you'd expect: content nobody asked for, ranking for terms nobody values, slowly training Google to ignore your domain. Every piece we publish has a human who owns the idea and a human who approves the words. AI drafts, accelerates, and adapts — people decide. The brands winning with content in 2026 are the ones using AI to raise their floor, not to abandon their standards.

Unchanged: strategy, taste, and nerve

No model chooses your positioning. No model tells you which audience to walk away from, or gives you the nerve to look different from your category when every best practice says blend in. The decisions that make or break growth are exactly as human as they were five years ago — there are just fewer excuses now for spending your human hours on anything else.

AI didn't change what good marketing is. It changed how much of your week you get to spend doing it.

If your growth engine still runs entirely on manual effort — or you've bolted on AI tools nobody actually uses — we build these systems end to end: strategy, creative, performance, and the intelligent plumbing underneath. Let's talk.