Documentation

• Any business — everything the sources can find in that niche and place, ranked by how strong a prospect they are. The honest default.
• Has a website — only businesses with a reachable site. Use it when you need to see the business online before you decide anything.
• Outdated or broken site — the site is fetched and audited, and only businesses with a real, checkable defect survive: no HTTPS, a certificate a browser refuses, not mobile-friendly, dead homepage links, a free-builder page, 2000s-era markup. Worst sites first, because those are the easiest conversations to start. This is the redesign seller's filter.
• No website — businesses that exist on the map with no site at all. This is the most expensive search to run, because each business gets its own web lookup for an Instagram, phone or email — offline businesses usually have a social page even with no site. This is the web designer's filter.

The filters unlock by plan: Any business and Has a website from Free, Outdated or broken site from Starter, and No website on Pro.

Searches reuse matching fresh cached records and fetch missing or expired data from the source. Three sources, each with different strengths:

• Web search (AI) — reads the open web for the niche and place; often returns contact details and company context directly. Best default.
• OpenStreetMap — open map data; strong on local physical businesses (which is exactly who needs a website), though many entries carry no site.
• Companies House (UK) — the official register, for companies that incorporated recently and mostly have no site yet. It runs on the deployment's own register key, so there is nothing for you to set up.

Each business is then enriched: its site is visited for an email, Instagram, LinkedIn and phone, including de-obfuscation of tricks like info [at] company [dot] com. Every site is audited, and the findings are attached to that lead.

The score is evidence, not opinion. Every site is graded on real measurements — does it answer at all, is the certificate valid, does it load on a phone, do its links resolve, what platform is it built on. The score is 100 minus the weight of what actually failed. One grade is named unknown and it exists on purpose: it means we could not judge, and we will not pretend otherwise.

A tool like this fails in one specific way: it accuses a healthy business of being broken, and the person using it loses the deal on the first line. Every check below exists to stop that, at some cost in raw quantity:

• Bot walls are not defects. A site that answers 403, 429 or 451 to our check is graded unknown — "refused our automated check". Plenty of healthy sites refuse datacenter traffic.
• Timeouts are never a finding. A timeout may be our network, so it is recorded and never used as a reason to contact anyone.
• Soft gaps stay soft. A stale copyright year, a missing meta description, no analytics — recorded as findings, but never a flag and never a "needs work" verdict. A 2016 footer doesn't make a business broken.
• JavaScript apps are not thin. Sites built as app shells are recognised, so a perfectly good modern site is never flagged for "almost no content".
• Only things a visitor can click. Broken-link checks ignore <head> metadata and vendor-injected links (feeds, API endpoints, email-obfuscation helpers) — those 404 to a machine but no human ever clicks them.
• Numbers have to be dialable. Anything that looks like a phone number but isn't — copyright ranges, IP addresses — is dropped rather than shown to you.
• Location is enforced from the address. Every business's own address is compared with the place you typed, and anything that places itself elsewhere is left out. You always see how many were left out and why. Records with no usable address are kept, not quietly dropped.

The direction of every one of those decisions is the same: we would rather show you one fewer lead than point you at a business that is doing nothing wrong.

• Order — Best prospects balances everything we know; Easiest to reach puts businesses with an email, social handle or phone first; Worst site first is pure audit score, the redesign seller's order.
• Expand any row for the full audit: every check, what was measured, and the verdict. The short list on the row is what to say out loud.
• Pitch column copies that one lead's block — business, contact, score, the concrete problems — so you can work the list one at a time.
• Export writes the whole list as CSV with every way to reach the business, the score, the problems and the reason to make contact. Header names are stable, so re-importing the file lands in the right columns.
• My leads holds what you saved. A business you already imported is hidden on repeat searches and never counted twice.
• Find the owner — one click searches for the decision-maker behind a business and attaches their name and role to the lead.

See pricing for the current numbers. Two things are metered: audited results per calendar month, and which filters your plan unlocks. Owner lookups count as results.

The important part is how a limit behaves: nothing breaks and nothing is lost. At the monthly limit, the Lead Finder asks you to upgrade and stops until the 1st. A result you never saw — one left out because it wasn't in your location — never counts against the month. Your live usage is on the Plan & usage page in the app.

• Verification email didn't arrive — check spam, then use Resend from the in-app banner. Mark it "Not spam" so the next ones land properly.
• Fewer results than I asked for — the note above the results tells you which case it is. Either the source genuinely ran out for that niche and place (try a broader niche, a nearby bigger city, or another source), or some businesses were left out because their address placed them elsewhere — the count is shown.
• A site reads "unknown" — we could not judge it (refused our check, or never answered). That's information, not a defect: nothing was accused, and it is not counted as a business that needs work.
• "The site may be slow" — heavy pages get flagged because slow sites lose customers; that one is a measurement, and worth mentioning only if it is comfortably above the usual range.
• "Out of its free daily allowance" — the shared web-search quota for the whole day is used up, so it returns at 00:00 UTC. Every other source keeps working meanwhile, and OpenStreetMap and the UK register need no search at all.
• Anything else — email support; we usually answer within a business day.