The SPARKS framework
From The 3-level ladder that separates AI prompters from AI thinkers published on AI Lab, engineered by ActiveCampaign
Get the framework
The big idea
Bryan Cassady’s SPARKS method is a six-step thinking discipline for working with AI on a single output—a blog post, a campaign brief, a landing page. It exists because AI fills in the blanks you should have filled yourself: vague input produces competent but generic output. SPARKS forces clarity at every step, from the messy first articulation to the final quality check.
The framework
| Letter | Step | What it forces |
|---|---|---|
| S | Speak it out | Articulate your messy thinking before AI helps. Rough input produces better clarification. |
| P | Pivot roles | Make AI interview you before it generates anything. |
| A | Ask for more | Force the critique. “What’s weakest?” beats “Is this good?” |
| R | Reframe | Change the angle of attack, not just the phrasing. |
| K | Keep asking | Iterate with surgical, specific feedback—not “make it better.” |
| S | Stop and think | Define what excellent looks like and grade output against it. |
When to use this
- You’re starting any single AI-assisted output—a brief, a draft, a campaign concept
- Your first AI outputs feel competent but generic, and you can’t articulate why
- You’re spending more time fixing AI output than the work would have taken yourself
- You need a repeatable method instead of random prompting
Common mistakes
- Skipping S (speak it out) and presenting polished thinking → Dump it messy. The mess is the point.
- Adding “…but go ahead and write a draft” to the P (pivot) step → Hold the line on the interview.
- Asking “is this good?” at the A (ask) step → Ask “what’s weakest?” instead.
- Treating R (reframe) as a synonym for “try again” → Change the viewpoint entirely.
- Vague feedback at K (keep asking) → Name the exact problem and the missing constraint.
- Accepting “done enough” at the final S → Write what excellent looks like, then grade against it.
Quick-start
Before you open any AI tool, run the Hallway Test: imagine asking the smartest intern in your company the question you were about to type. If they’d push back (“Who’s it for? What’s at stake?”), the prompt isn’t ready. SPARKS is what closes that gap. Start with the first S—speak your thinking out, messily—and the rest of the method falls into place.
Related
More data from the AI Lab.