# The Low-Contact Idea Validation Workflow

## Purpose

Use this workflow for every new product idea. It is designed for a technical founder who prefers structured, asynchronous research over frequent live outreach. Its job is not to prove an idea is good; it is to decide, cheaply and quickly, whether the next level of effort is justified.

**Principle:** AI finds hypotheses. Real behaviour tests them.

## The model: Signal → Proposition → Commitment → Learning → Decision

```mermaid
flowchart LR
  A[Idea] --> B[Find existing pain]
  B --> C[Define a narrow buyer and job]
  C --> D[Make a concrete offer]
  D --> E[Ask for a small commitment]
  E --> F[Learn from evidence]
  F --> G{Evidence strong enough?}
  G -->|Yes| H[Run a concierge pilot]
  G -->|No| I[Refine, park, or stop]
  H --> J[Build only what the pilot proves]
```

## Stage 0 — Frame the idea

Do not start with a feature list. Complete this sentence:

> For **[specific person]** who needs to **[job they are already trying to do]**, this helps them achieve **[measurable outcome]** without **[current frustration or cost]**.

Write down:

- The buyer: one role, company type, or life situation.
- The user: if different from the buyer.
- The job: a repeated, time-sensitive task or risk they are already handling.
- The current alternative: spreadsheet, agency, incumbent product, manual process, or doing nothing.
- The consequence: wasted time, lost money, delay, risk, status, or stress.
- The earliest believable price or pricing model.

**Exit condition:** you can name who feels the pain, what they do now, and why that is inadequate.

## Stage 1 — Research existing pain

Use AI as a research assistant, not as the judge. Search for evidence that people are already struggling or spending.

### Evidence sources

| Source | What it reveals | Search pattern |
|---|---|---|
| Product reviews | Incumbent gaps and switching triggers | `"[competitor]" review "wish"`; `"[competitor]" "too expensive"` |
| Reddit and specialist forums | Unfiltered problem language and workarounds | `site:reddit.com "[job]" "how do you"`; `"[problem]" forum` |
| LinkedIn posts and comments | Professional pain, role language and visible communities | `site:linkedin.com/posts "[problem phrase]"` |
| Job adverts | Evidence that a company pays humans to solve the task | `"[job]" "responsibilities"` |
| App stores and software directories | Demand, complaints, categories, competitor positioning | search the category and read low-star reviews |
| Google suggestions and People Also Ask | How potential users phrase the job | type the job, problem, and alternatives into search |
| Public help centres and communities | Repeated support friction | `"[competitor]" "community"`; `"[competitor]" "help"` |

### Capture evidence consistently

For each useful item, record the exact problem in your own words, the source, the persona, the workaround, the emotional or commercial consequence, and a confidence rating. Seek at least ten independent pieces of evidence before treating a pattern as real.

### AI research prompt

> Act as a market-research analyst. For the audience **[persona]** and job **[job]**, identify repeated complaints, current workarounds, paid alternatives, and the exact language people use. Separate direct evidence from inference. Return a table with source type, insight, problem phrase, willingness-to-pay clue, and community or channel to investigate. Do not invent quotes or sources.

**Exit condition:** a pattern appears across more than one source type, and it describes an existing pain rather than a hypothetical preference.

## Stage 2 — Locate communities without cold networking

The objective is not to find a large audience. It is to find places where your specific people already exchange help, recommendations, and complaints.

### Community discovery ladder

1. **Name the identity.** Search the role, not merely the problem: “independent bookkeepers community”, “operations managers forum”, “parents of children with dyslexia group”.
2. **Follow the tools.** Search each incumbent or workaround plus `community`, `forum`, `Slack`, `Discord`, `Facebook group`, `Reddit`, `webinar`, or `newsletter`.
3. **Follow the events.** Search the role plus `conference`, `meetup`, `association`, `podcast`, or `newsletter`. Event sponsors and speakers often reveal communities.
4. **Follow the language.** Use exact complaint phrases from reviews or posts in quotes. A phrase often surfaces niche groups more reliably than broad keywords.
5. **Follow the creators.** Identify writers, YouTubers, podcasters, and newsletter authors who cover the job. Read their comments and resource pages.

### Score each community

Give each option 0–2 for each criterion:

- **Density:** are enough target people present?
- **Relevance:** are they discussing the actual job or pain?
- **Permission:** can you ask a research question or share a pilot without breaking rules?
- **Activity:** is the conversation current and substantive?
- **Access:** can you participate without paying or being a recognised insider?

Prioritise communities scoring 7 or more out of 10. Read their rules first. Add value before asking for anything; a concise request for pilot applicants is usually better than a broad survey.

### Low-pressure outreach templates

**Pilot post**

> I am exploring a small tool for **[specific job]**. It is aimed at people who currently **[manual workaround]** and want **[outcome]**. I am looking for a few people to try an early pilot and tell me, asynchronously, where it fails. If this is relevant, reply or use **[link]**. I will not add anyone to a mailing list without permission.

**Three-question form**

1. When did you last deal with this problem?
2. What did you do instead, and what did that cost in time, money, or risk?
3. If a tool solved it reliably, what would need to be true for you to try or pay for it?

## Stage 3 — Make a testable offer

Create a single page, not a full product. It should include:

- A clear audience and painful job in the headline.
- The outcome, in plain language.
- A brief explanation of how it works.
- An indicative price or pricing basis.
- A single commitment action: apply for pilot, join a waitlist, book an asynchronous setup, upload a sample, or place a refundable deposit.
- A note on who it is not for, where useful.

Avoid words such as “revolutionary”, “AI-powered” or “all-in-one” unless they clarify something concrete. Do not ask “Would you use this?”

## Stage 4 — Test for commitment

Order signals by strength:

| Signal | Meaning |
|---|---|
| Like or friendly comment | Curiosity only |
| Email address | Mild interest |
| Completed detailed application | Problem recognition |
| Calendar booking or asynchronous setup | Time commitment |
| Sample data, introduction, or workflow access | Trust and urgency |
| Deposit, pre-order, or paid pilot | Willingness to pay |
| Repeat use and renewal | Product value |

Set a decision threshold before promotion. Example: “I will run a manual pilot only if five people from the target role complete the application and at least two accept the stated price range.” Adjust the numbers for your market; the point is to prevent building because the idea feels exciting.

## Stage 5 — Run a concierge pilot

Deliver the promised outcome manually or with lightweight tools first. This exposes the real workflow, exceptions, and value drivers before software hardens assumptions.

Track:

- Time from request to outcome.
- Manual steps and recurring exceptions.
- Where participants hesitate or abandon.
- Whether the result changes a decision or saves effort.
- Whether they ask to continue and will pay.

Ask for feedback in writing: “What was useful?”, “What nearly made this not worth using?”, and “What would you do if this disappeared tomorrow?”

## Stage 6 — Decide deliberately

At the end of a short test cycle, choose one outcome:

- **Proceed:** repeated pain plus meaningful commitment; build the narrowest repeatable part.
- **Refine:** the pain is real but the audience, offer, channel, or price is wrong; change one variable and retest.
- **Park:** some interest but no urgency; record the evidence and revisit only if a new trigger appears.
- **Stop:** no meaningful commitment after a fair, targeted test; preserve the learning and move on.

## Reusable one-page worksheet

| Field | Answer |
|---|---|
| Idea name | |
| Target buyer/user | |
| Job and triggering moment | |
| Current workaround | |
| Cost of doing nothing | |
| Evidence collected | |
| Top three community candidates | |
| Offer and indicative price | |
| Commitment action | |
| Test threshold | |
| Pilot learning | |
| Decision and next test | |

## Weekly cadence for multiple ideas

- Monday: frame one idea and complete the evidence sweep.
- Tuesday: shortlist and score communities.
- Wednesday: publish the smoke test or pilot invitation.
- Thursday: collect form responses and clarify asynchronously.
- Friday: review commitment evidence and make one explicit decision.

Run no more than one or two ideas at once. A consistent record of rejected ideas is valuable: it prevents returning to them later without addressing the original evidence.

## Free or low-cost tool stack

- Research: search engines, Reddit, public review sites, app directories, AI-assisted synthesis.
- Capture: a spreadsheet or Notion-style database.
- Forms: Google Forms or Tally’s free tier.
- Landing page: a simple static page, GitHub Pages, Carrd free tier, or a shared document.
- Scheduling, only if wanted: a booking link with an “asynchronous questionnaire preferred” option.
- Testing services to study: User Interviews, Respondent, Wynter, PickFu and Pollfish. Their value is recruitment and structured testing; replicate the method first with public communities before paying for panels.

## Guardrails

- Do not misrepresent a prototype as a finished product.
- Follow community rules and do not scrape private spaces.
- Treat AI-generated claims as hypotheses until a source or human action supports them.
- Do not over-value compliments from friends, peers, or people outside the target audience.
- Keep an evidence log so future decisions are reversible and explainable.