DistroFitThe resources page on this site lists tools by the question they answer. What it does not say is what order to ask the questions in, or how long to give each one before moving on. That is what this post is for.
The order matters more than the tools. Most ideas die at one of six questions, and the cheapest question to answer should go first, because a no on Monday saves the rest of the week. So the sequence below is arranged by how quickly each step can kill the idea, not by how interesting it is.
One rule before any of it. Keep a log. One line per finding, in the other person's words where you can, with a link back to where you found it. The log is the output of the week. A decision is what you do with the log afterwards, and a decision made without one is a guess with a calendar entry.
Monday: is anyone complaining?
The strongest evidence that a problem exists is people complaining about it in public, unprompted, before you turned up. Everything else is weaker, including surveys, and especially including your own conviction.
Start on Reddit. Find the subreddit where your users already spend time, sort by top, and read the threads where people are annoyed. It is slow and there is no shortcut. The threads you want are the ones where someone describes a workaround, because a workaround is proof that the pain was worth effort.
Then go to G2 and Capterra and read only the one and two star reviews of the products closest to yours. These come from people who already paid money, which makes them worth more than any forum thread. Both sites are lead generation businesses and the ordering of products is influenced by who pays them, but the review text itself is real.
For developer and founder problems, search Hacker News through Algolia and read the Ask HN threads. The audience is technical and contrarian, so do not mistake it for how normal buyers think, but it is unusually specific about what hurts.
Finish with Google Trends for one thing only: direction. It gives relative interest, never absolute numbers, so it can tell you whether you are early or late and nothing more. Exploding Topics surfaces things earlier and has a free tier. It is part of Semrush now and its lists are widely read, so the head start is smaller than it looks.
What a good Monday log looks like: a dozen complaints in other people's words, dated, most of them older than your idea. What a bad one looks like: your own description of the problem, restated a dozen ways.
Skip the tools that promise to mine Reddit for a startup idea. GummySearch was the best known of them and it closed in November 2025. What replaced it is mostly a wrapper around the public Reddit API with a language model on top, and none of it has data you cannot read yourself.
Tuesday: who already does this, and are they alive?
It almost always exists already. The useful question is whether the existing products are any good and whether any of them make money.
A category page on G2 or Capterra gives you the whole competitive set in about ten minutes. Coverage is strong for B2B software and weak for consumer, hardware and anything niche, so treat an empty category page on those as a gap in the directory, not a gap in the market.
Then read Product Hunt for the last few years of attempts, including the ones that died. A listing tells you nothing about whether the product survived, which is exactly why the graveyard is worth reading. Every dead attempt is a free post-mortem if you can work out what it got wrong.
For the ones still alive, Similarweb gives you a free per-site view without logging in: rank, traffic direction, bounce rate, and the split between direct, search and referral traffic. The numbers are estimates and they get unreliable on small sites, so read them as shape rather than fact. Write the traffic split down anyway, because you will want it on Friday.
BuiltWith tells you what a competitor runs, which is a decent read on how serious they are and roughly what they spend. It works by fingerprinting, so it misses anything server side or self hosted. Ahrefs says its free backlink checker shows the top 20 backlinks, which is enough to know whether a competitor has any real search presence at all. Crunchbase tells you whether anyone raised money to attack the same problem. Its coverage of bootstrapped competitors is poor, and those are usually the ones you are actually up against.
Tuesday's log should name every live competitor, what each one charges, where its traffic comes from, and one thing its own customers say it gets wrong.
Wednesday: count, do not estimate
This is where most people invent a number and move on. The fix is to count real things.
For a B2B market, the honest way to size it is the US Census SUSB. It is free and downloadable, and it gives you counts of real firms by industry and employee band, down to county level. Counting the companies that could actually buy from you beats any multiplication you can do on a slide. It is US only and it lags by a year or two, which is fine, because you are after an order of magnitude.
For anything found through search, Google Keyword Planner is the best free source of absolute volume, and the suggested bids tell you what commercial intent looks like in your space. You need a Google Ads account to reach it, and the volumes come in wide ranges until you are actually spending. The Ahrefs Keyword Generator is the no-signup sanity check: ten seed terms, volume estimates, done in a minute.
Point Similarweb at the category leader. Modest traffic for the biggest player in your space means your share of it is smaller still, and that is worth knowing before Thursday.
Statista has a free basic tier. The Starter plan is listed at $199 a month billed annually, which is out of reach for most people doing this on their own. Plenty of its headline figures are re-published vendor estimates, so check the source line before you quote one.
Skip the TAM, SAM and SOM calculators. They all do the same thing, which is turn assumptions you made up into a number that looks researched. A market size that did not come from counting firms, search volume or competitor traffic is a guess with a chart on it.
Thursday: the price, and what it forces
Open five competitor pricing pages. This is the single most useful hour of the week and it needs no tool. Write down the tiers, what gates each one, and whether the price is per seat, per month or per sale.
Then look at Baremetrics Open Startups, a public board of small companies publishing real monthly revenue. It is a small, self-selected sample skewed towards founders who benefit from being seen, but it is actual numbers from businesses at a comparable size, which is rarer than it should be. For your own list, Tally runs a willingness to pay survey for nothing: unlimited forms and submissions under fair use, with conditional logic and calculations not paywalled. Remember that what people say they would pay is always higher than what they pay.
Now do the sum, because the price decides most of what happens next. Say you charge £5 a month and want £100,000 a year. That is 1,667 paying customers, and at 5% monthly churn you need about 83 new ones every month just to stay where you are. Charge £99 for the same product and it is 85 customers, with about four a month to hold them. Same product, same target, a completely different job. And if you want acquisition cost back within six months, £9 a month gives you £54 to win each customer, which closes paid search and paid social before you have written a single ad.
Customers needed runs that arithmetic for any price, and the channel report can run it for a one-off price as well, since a product that is paid for once has no base to hold: the year's sales have to be found afresh, a twelfth of them every month.
Skip AI pricing optimisation and dynamic pricing tools. At small scale you do not have the traffic to run a price test that means anything, so you would be paying for precision you cannot reach. Read the five pages, pick a number, change it later.
Friday: how would anyone find out, and does the site work?
Two questions, because by Friday you either have a site or you are about to build one.
For distribution, Google Search Console is not optional. It is the only place your real query, impression and click data exists. It keeps roughly sixteen months of history and hides low volume queries, so start it early. SparkToro answers where your audience already spends attention: which sites, podcasts and accounts they follow. The free plan is three previews a month, which is a teaser, and paid starts at $150 a month for the Business tier, so treat it as a one month subscription for a research sprint rather than something you keep. Go back to the Similarweb traffic split you wrote down on Tuesday. Where the leader in your category gets most of its traffic tells you more than any advice post will.
Indie Hackers has channel post-mortems from people at your scale. The featured stories carry heavy survivorship bias, so read the failures more carefully than the wins. BuzzSumo finds high performing content and the people who write it, and it starts at $199 a month, which is not a solo builder price. Use the free trial for one focused week and cancel.
For the site itself, Search Console's URL Inspection is the only authoritative answer to whether Google has indexed a page. PageSpeed Insights combines lab results with real Chrome user data where there is enough traffic to have any; a single score is noise, so read the diagnostics under it. Lighthouse is the same engine, scriptable, and able to run against localhost and pages behind a login. Screaming Frog crawls up to 500 URLs free, which covers most solo projects, and finds the broken links and duplicate titles you will never spot by hand. The licence is £199 a year, and the free version cannot render JavaScript, which matters for a single page app. Check structured data twice, with the Rich Results Test for what Google will actually do with it and the Schema Markup Validator for whether it matches the spec. Give Bing Webmaster Tools ten minutes, because Bing's index increasingly feeds AI answer surfaces.
Skip the all in one growth suites and the social schedulers. Scheduling is a solved commodity and none of them solve distribution. They solve posting, which is a problem you do not have yet. Skip the free SEO score widgets too. A number out of 100 followed by an email field is a lead capture form in a diagnostic costume.
I will be straight about the obvious conflict here, since this site runs a free diagnostic as well. The channel report reads your site and says which of the 19 channels are open, closed, or would open with one specific product change, and every line cites the signal that produced it. It does not estimate traffic, because nobody can do that accurately for free. Where your audience already spends attention is a question SparkToro answers and this site does not.
What the log should say by Friday evening
Read it back and it will fall into one of three shapes.
Complaints that predate you, live competitors with visible traffic, a countable market and a price that clears the arithmetic. Build it, and start with whichever channel Friday said was open.
Everything except the complaints. Stop. A market with competitors and no unprompted pain is a market where the existing products are good enough, and you found that out for the price of a week.
Everything except the price. Change the price, not the plan. Thursday's sum is the one part of the week you control completely, and it is usually the part that was never done.
A note on money. Every price quoted here was taken from the tool's own site when the resources page was written. They change constantly and free tiers get quietly reduced, so check before you commit to anything. Nothing on this page pays me and there are no affiliate links.