Choose one customer and outcome
Use the ICP to keep the idea specific. If the audience is uncertain, Specky names that uncertainty instead of hiding it.
Idea validation / 02
Most ideas do not fail because the team lacked opinions.
Test the belief that could make the whole idea fail.
One connected plan for assumptions, interviews, experiments, and the decision that follows.
Choose an idea and ICP segment. Specky reads the real evidence, finds the riskiest unknowns, drafts research designed to avoid leading questions, and gives the team observable rules for build, reframe, or stop.
- Evidence for and against stays visible
- Creates real Alex campaigns and experiments
- People own the final build decision
An idea moving through risky assumptions, customer interviews, a lean experiment, and a build decision
The job
Learn enough to make a smaller, better product bet.
A backlog item hides several beliefs: the problem exists, the right people care, the behavior can change, the business benefits, and the team can deliver. Idea Validation makes those beliefs testable.
Start with a question
“What must be true for this idea to deserve the next week of work?”
IdeaICPCustomer evidenceConstraints
Bring one idea, the customer you expect to help, what you already know, and the doubts or constraints the test must resolve.
Specky returns
A reviewable validation plan
- Risk-ranked assumptions with evidence on both sides
- Recruitment copy and behavioral interview guide
- A lean experiment and explicit decision rules
01
Frame
Name the audience, problem, and expected outcome.
02
Challenge
Find the belief with the greatest downside if wrong.
03
Test
Run the smallest credible interview or experiment.
04
Decide
Build, reframe, or stop against rules set in advance.
What goes in / what comes out
Clear inputs. Useful work back.
The feature should be understandable before anyone connects a source or runs an AI generation.
You give Specky
- A new idea or an existing feature, opportunity, roadmap item, strategy card, document, or experiment
- The ICP segment expected to benefit—or an honest note that the audience is unknown
- The customer problem, expected outcome, existing evidence, doubts, and product constraints
- Interviews, feedback, calls, support, analytics, and product context already in the Product Graph
Specky gives back
- Desirability, viability, feasibility, and usability assumptions ranked by risk
- Evidence for, evidence against, evidence gaps, and a confidence read
- A smallest-next-test recommendation with method, success criteria, and timebox
- LinkedIn recruitment post, follow-up, candidate outreach, and behavioral interview questions
- A real Alex campaign or Experiment Tracker record created from the plan
- Pre-agreed conditions for build with confidence, reframe, or stop
The workflow
From a fuzzy belief to work the team can use.
Each step leaves a reviewable artifact. People stay responsible for the claim, the decision, and what ships.
Put assumptions in risk order
Test the belief that would waste the most time or create the most harm—not the belief that is easiest to confirm.
Recruit without pitching
Invite the right people and ask about past behavior. The campaign seeks disconfirming evidence, not compliments.
Run the smallest credible test
Create the linked experiment, add a fake door when useful, and review customer words beside the behavioral result.
A concrete example
Should growth PMs receive a weekly explanation of signup changes?
The dangerous belief is not that Specky can generate a report. It is that a target PM has this recurring job, trusts a cited explanation, and changes a real decision because of it.
Starting point
- 1Idea: automatic weekly signup explanation
- 2ICP: growth PM at a 20–200 person B2B SaaS company
- 3Evidence: two calls mention reporting pain; no recurring-use proof
- 4Constraint: the brief must disclose missing analytics data
Useful result
- Critical assumption: PMs will use the brief without recreating the investigation
- Five behavioral interviews about the last signup review
- A two-week concierge test with a pre-agreed reuse criterion
- Build if three target PMs use it in a real review and request another; reframe or stop otherwise
Questions
The practical details
What does Idea Validation do?
It turns one idea and its available evidence into a ranked assumption board, a neutral customer-recruitment kit, behavioral interview questions, the smallest useful experiment, and explicit build, reframe, and stop conditions.
How is this different from the Experiment Tracker?
Idea Validation decides what needs testing and why. The Experiment Tracker runs and records a specific test. Specky can create that real experiment directly from the riskiest assumption, so the two stay connected.
Can Specky create customer interviews from the plan?
Yes. Create a real Alex research campaign from the validation plan. Specky carries the hypothesis and target participant into the campaign; the generated LinkedIn post, follow-up, outreach, and interview guide remain available for review and reuse.
Can we test an idea with a fake door or waitlist?
Yes. Create the experiment from the assumption, then attach a Validation Page for a fake-door, waitlist, feature-interest, prototype, or concierge test. Responses and events flow back into the Product Graph.
Does AI decide whether we should build?
No. Specky organizes the evidence, uncertainty, and pre-agreed decision rules. A person reviews the actual signals and records build with confidence, reframe, or stop.
Give the next idea a fair test—not a free pass.
Start with the risky belief. Keep the evidence honest. Make the build decision while it is still cheap to change.