Evidence first.
Local questions next.
Find customer problems that sustain businesses elsewhere, then investigate whether a specific German customer is still poorly served.
What this collection covers
The collection began with six selected businesses from the public IndieHackers catalogue. Smith.ai, MyClean and Edusign test service and administration models. Typebot and Ghost test established software markets. LabGPT tests how the process handles an extraordinary directory revenue figure and limited public proof.
The next batch adds every Stripe-labelled entry: 23 records, including Ghost and Typebot from the pilot. All 528 catalogue records are retained in the work queue. This publication contains 27 company investigations and linked opportunity assessments. It does not claim to represent the whole German market.
Three separate assessments
- Product existence
- What can be observed about the team, website, product, documentation and public operating history?
- Commercial success
- What evidence supports paying adoption or sustainable operations? A company-published metric and an audited result are different types of evidence.
- German opportunity
- What specific customer problem appears underserved after checking direct competitors, international suppliers and ordinary substitutes?
How to read evidence labels
- Observed
- A page, repository, document or publicly visible product behaviour inspected during research.
- Company claim
- A claim made by the company, including customer counts, performance results and financial figures.
- Corroborated
- Additional evidence supports the statement. The cited sources show the extent of that support.
- Analysis
- Our interpretation of the evidence, including transferability and proposed customer tests.
- Unresolved
- Public evidence is incomplete, contradictory or inaccessible.
What the assessments mean
- Ready for a customer test
- Evidence supports a bounded customer experiment. It still does not establish willingness to pay.
- Investigate further
- A narrower hypothesis is worth investigating, but a material assumption is untested.
- Broad market already served
- The broad category already has credible alternatives. Any narrower proposal needs new evidence.
- Evidence still insufficient
- The overseas success or local gap has not been established well enough to use as a precedent.
We use reasoned assessments, without invented numerical confidence scores or revenue forecasts.
Sources, screenshots and dates
Searches use German and English descriptions of the customer problem. Sources include company pages, customer cases, product documentation, public repositories, German competitors and official regional statistics. Public founder or social accounts are checked when relevant and accessible. Login walls, incomplete archives and unverifiable claims remain explicit research limitations.
Website captures use a 1,440 by 1,000 CSS-pixel desktop viewport at 2× device pixel density. Full-page originals are retained without resizing. Social evidence uses native-resolution crops of public posts, excluding account navigation and personalised profile context. Pages are scrolled to trigger lazy content. A screenshot establishes what a page displayed; it does not verify the company's claims or test the paid product. Cookies, changing content and animations may affect a capture.
Publication dates, observation times and the reference periods of statistics are recorded separately. Regional counts retain their original geographic boundaries. Household or professional counts do not prove demand or willingness to pay.
How updates work
A persistent queue holds initial research and later rechecks. Each research revision keeps its sources, findings and capture timestamps. Repeated imports of an unchanged packet do not duplicate the research. New evidence adds a revision and regenerates the affected pages.
The six-company pilot is complete. The completed Stripe batch covers all 23 Stripe-labelled directory records, including Ghost and Typebot from the pilot, for 27 distinct companies in total. The remaining catalogue and any recurring schedule are separate run scopes. Promising hypotheses should next be tested through customer conversations or a small paid pilot.