Claude for
product work
Research synthesis, PRD drafting, competitive teardowns, and the messy work between scattered feedback and a coherent plan.
Product work is a lot of turning unstructured input into structured thinking, which is exactly where Claude is most useful. But the failure mode is subtle: it will happily produce a beautifully structured PRD that misses the actual insight, and you'll only notice on the second read. The habits on this page are mostly about protecting against that.
Three places to start
Three workflows that change the rhythm of product work without replacing the thinking.
1. Synthesising a stack of user interviews
You've done eight interviews and have pages of notes. Claude will pull out the themes, count how often each appeared, surface direct quotes, and, most importantly, flag the places where interviewees contradicted each other. The contradictions are usually the most valuable part of the output, because they're where real insight hides.
Interview synthesis prompt, annotated
Four elements applied to the research synthesis ritual.
Where this falls over
Claude's default is to smooth over disagreements by merging them into a single theme. You must tell it explicitly: do not smooth over disagreements, surface them as tensions.
2. Drafting a PRD from rough bullets
You've got a feature idea as scribbled bullets. Paste them, paste an example PRD from your team as a voice reference, and ask for a first draft. The critical instruction: do not add anything I did not say, and leave a clearly marked TBD for anything ambiguous in my bullets. The output is a draft that tracks your actual thinking plus a list of decisions you hadn't made yet. Those TBDs become your next hour of work.
Rough bullets into a PRD, two ways
The difference between a PRD that drifts from your thinking and one that tracks it.
A beautifully-structured PRD with sections on "User Personas" (three invented personas), "Success Metrics" (a made-up 20% improvement target), "User Stories" that are plausible but not things you said, and a confident "Proposed Solution" that commits you to the checklist option without flagging that you hadn't decided yet.
A PRD draft that tracks your actual thinking, with clearly marked TBDs for everything you hadn't decided:
(1) Target metric for "too long". You said "too long" but not against what. Is this a time metric, a step count, a drop-off rate?
(2) Shorter tutorial vs checklist. You floated both. Which are we proposing? I've left both in a "two options to pick between" section.
(3) Who is the primary user? Your bullets don't say. I've drafted this assuming a mixed user base but it may need narrowing.
Where this falls over
Without the "do not add anything" instruction, Claude will produce a PRD that sounds plausible and contains three invented user needs. The drift is subtle. Always include the instruction.
3. Competitive teardowns
Give Claude a competitor's website, your existing notes, and ask for a factual teardown: target customer, core value prop in their words, three features they're proud of, pricing if visible, what they're conspicuously avoiding, and the two or three things you do differently. It won't replace the qualitative work of actually using the product, but it will give you a skeleton to fill in.
Did a teardown of one of our competitors in twenty minutes. Pasted their landing page, pricing page, and three G2 reviews into Claude. Asked for: target customer, three features they lead with, pricing model, what they are conspicuously not talking about, and two or three things we do differently.
Got exactly that. Factually grounded, no speculation, every claim tied back to something in the source I pasted. Dropped it straight into the product strategy doc. First time a competitive teardown has not eaten my Wednesday afternoon.
Where this falls over
Claude will make claims it can't verify unless you explicitly tell it to stick to what's in the source material. Always include that constraint.
Worked examples
Before and after, on tasks from product work.
Prompts to steal
Copy any of these, fill in the square brackets, run.
Try this week
Pick one. All of these are under an hour.
The test of whether Claude is helping your product work is whether your thinking is getting sharper or just getting produced faster. The first is the goal. The second is how you end up with a lot of polished PRDs that ship features nobody wanted.
If you want to go broader, the Beginner guide covers the habits that apply everywhere. The prompt library has the full set of role-tagged prompts in one place.