How to Find and Validate Startup Ideas Using Real Search Demand

Updated

Start with what people already search for, then filter by two numbers: monthly search volume and competition score. On MuckBrass.com, each idea is shown with both, so you can drop directions that are too small or too crowded before writing any code. The method works best for software and micro-SaaS ideas where search behavior reflects a real, recurring problem.

Why interest is a bad starting point

The site opens with a Paul Graham quote built on the Yorkshire saying "where there's muck, there's brass": unpleasant work pays, and work people enjoy often doesn't, because supply and demand push the price down. The extreme example given is programming languages, which people build for free because they like it.

The practical translation: if you start a company to do something cool, aim to make money and maybe be cool — not the reverse. Your own enthusiasm is not evidence of demand. Search behavior is.

The two filters: volume and competition

MuckBrass.com lists ideas with a monthly search volume and a competition score out of 100. The site's own framing is "build what they Google, not what you think they need," and its workflow is: AI analyzes trending keywords, you build a product around one, then traffic follows.

Use the two numbers as a first pass:

Signal What it tells you What to do
High volume, low competition Demand exists and the field is open Shortlist for deeper validation
High volume, high competition Demand proven, but you need a sharp wedge Only proceed with a specific angle
Low volume, low competition Possibly an untouched niche, possibly no market Check whether the searchers are buyers
Low volume, high competition Weak on both counts Skip

The site's examples show how wide the range is. An "AI girlfriend" idea shows 21K searches per month at competition 19/100. A "no-code website builder" for SMEs shows 13K per month but competition 59/100 — real demand, crowded field. "AI food critic" and "faceless content" both show 10 searches per month at competition 0/100: essentially no measured demand, so a zero competition score is not automatically good news.

A customized hypertrophy workout planner for fitness athletes shows 92K per month at 19/100 — the highest volume in the sample with modest competition, which is why it reads as the strongest candidate there.

Prefer "it hurts" problems over fun projects

The site's section heading "When it hurts, they search" captures the selection rule: pick problems painful enough that people look for a fix, rather than ideas that are merely interesting. Pain shows up as repeated search queries with commercial language — "app," "software," "calculator," "plans," "builder." Those words suggest the searcher wants a tool, not an article.

The workout example illustrates it: the listed keywords include "hypertrophy training app," "one rep max calculator," and "strength training software." Someone searching those is close to paying for a product. Compare that with a broad curiosity query, which tells you little about willingness to pay.

Turn a keyword into a testable hypothesis

A search volume is not a business. Convert each shortlisted idea into a one-sentence hypothesis with a specific user and outcome, then test it cheaply.

  1. Write the hypothesis. For the workout idea: "Hypertrophy-focused lifters will pay for a tool that turns their one-rep max into personalized MEV/MAV training plans."
  2. Check the intent behind the keywords. Confirm the listed terms are product-seeking, not informational. If most queries are "what is," demand may be for content, not software.
  3. Build a landing page, not the product. State the outcome, show a price or a waitlist, and drive the exact keywords you found. The site itself offers a "Test my idea" path alongside "Find ideas," which reflects this order: validate before building.
  4. Set a decision threshold in advance. For example: a defined number of sign-ups or pre-orders from a set amount of traffic within two weeks. Decide the number before you look at results.
  5. Talk to a handful of searchers. Ask what they use today and what they pay for it. Search volume shows the problem exists; conversations show whether your fix fits.

Common traps

  • Treating zero competition as a green light. Ten searches a month at 0/100 competition means almost nobody is looking. Low competition only matters when volume is real.
  • Ignoring the crowded high-volume ideas entirely. 13K searches at 59/100 competition is a proven market. The right response is a narrower wedge, not abandonment.
  • Building before testing. The site's own sequence — analyze keywords, then build — only works if the validation step between them actually happens.
  • Confusing a keyword list with a product. The listed terms are evidence of demand, not a feature spec. Your hypothesis has to name a user and an outcome.

Where to start

Open MuckBrass.com's idea list, sort mentally by the table above, and pick two or three candidates with meaningful volume and competition you can plausibly enter. Write one hypothesis for each, put up a landing page, and let the response — not your gut — decide what you build.