Hint intelligence

A hint describes a conversation, not a query

A keyword catches what someone types into a box. A hint catches what someone is telling ChatGPT about their situation. Nobody types "plumber Stockholm cheap" into a conversation — they write "there's water coming through my kitchen ceiling, what do I do?" A good hint sounds like that: scenario plus intent, in the words a real person would use mid-problem.

What AthillyAds does with that

We read your site and your brand profile, then write hints in groups — one group per intent. People describing an urgent problem think and talk differently from people comparing three quotes, and both differ from people who already know your name. Each group gets its own ad copy, because the same card can't speak to all three.

Every group carries a one-line explanation of who it reaches and what to expect — "These reach people weighing options: expect lower CTR but cheaper clicks." You never have to reverse-engineer why the machine did what it did.

Exclusion hints: who you'd rather not pay for

Your brand profile records who you're not for — renovations you don't take, customer types that never convert. That feeds exclusion hints on every ad group: conversations we deliberately keep your ads out of. It's negative targeting the platform's own ads manager doesn't even have an input field for yet. We store it, apply it to generation, and export it — so it's working for you the moment the API accepts the field.

Five to eight hints, on evidence

The API accepts up to 2,000 hints per ad group. Accounts that stuffed the field watched delivery stall. Our generator writes five to eight specific hints per group because that's what the early data supports — and we say so in the product, marked as the working hypothesis it is. As real campaigns run through the platform, anonymized benchmarks confirm or kill it. Your hint library shows how each hint performs against those benchmarks, not just against your own average.