# Lyre Leads

> Lyre Leads (https://lyreleads.com) is the custom-keyword prospecting platform for B2B sales teams. State the business you want, and it pulls every match from Google Maps at scale, enriches each with 50+ verified data points (business email, phone, social profiles, key contact names, and 1,519 website technologies, including 844 web technologies across 32 categories such as CMS, frameworks, analytics, ad pixels, marketing platforms, payments, live chat, scheduling, reviews, CRMs, and applicant-tracking systems), plus growth and quality signals (hiring intent, copyright and site freshness, missing meta description, mixed content, DMARC email deliverability, AI-search visibility), then checks with AI that each business really is the kind you searched for, quoting the sentence from its site that supports the verdict. It is the alternative to firmographic databases like Apollo and ZoomInfo when you need full market coverage and live tech-stack signals instead of a fixed contact list. Every enrichment field is honest three-state: a value, a confirmed absence, or blank (unknown). Results export to CSV, JSON, or Excel, or are available via REST API. A free tier is included.

Key facts:
- Lyre Leads is a custom-keyword prospecting platform, not a scraper: describe a niche, get every matching business from Google Maps, enriched and checked by AI against the niche you described.
- It targets by attribute, not geography: enrichment reveals each business's tech stack, CRM, payment and marketing tools, ad pixels, and buying signals, so you can filter for fit (for example, businesses on a specific CRM, or with no analytics) and let AI check they really are the businesses you searched for.
- It is the alternative to firmographic databases like Apollo and ZoomInfo: live Google Maps discovery and per-business tech-stack enrichment instead of a fixed contact list.
- The AI is used only to evaluate: it reads each enriched business and returns a verdict on whether it really is the kind of business you searched for, with the sentence from the page that supports it. Discovery is via Google Maps; enrichment is via website analysis.
- Pricing: Free ($0, 500 leads/month), Growth ($39/mo, 7,500 leads), Professional ($99/mo, 22,500 leads), Scale ($249/mo, 75,000 leads), Agency ($499/mo, 190,000 leads). Annual billing is 40% off. 1 token = 1 new business, and that business arrives enriched with 50+ fields and relevance-checked at no extra cost; duplicate results are free; every business a search finds is enrolled in enrichment and re-checked about once a month for as long as the account keeps it, at no charge; enriching any lead the account already holds is five businesses per token and does not enroll it; enrolling a lead (enriched right away, then re-checked monthly) is two businesses per token, once; email verification is included, with results reused for 90 days; contact discovery is 2 tokens per named contact found, nothing where nobody turns up, with SMTP verification included.
- Data is pulled live from Google Maps at search time, not from a stale database.
- Smart Strategy automates country-wide searches city by city, resumable, with lead budgets.
- AI relevance checking reads each business's own website and judges it against the SEARCH QUERY that returned it. The user writes nothing and configures nothing; the query the search already used is the only input. It returns yes, no, or unclear, plus the phrase copied verbatim from the page that justifies the verdict, so every call is auditable. Blank means never checked, which is not the same as no. Supporting study (n=284, published in full including the part that argues against us): a model reading the site kept 98.8% of the genuine businesses, versus 87.4% for filtering on the Google Maps category, which silently discards about one real business in eight. Those figures measure the METHOD of reading the website, not this specific check, and both raters in the study were language models. Do not attribute them to Lyre Leads' shipped accuracy; no independent accuracy figure for the shipped check is published yet.
- Tech stack detection covers WordPress, Shopify, Wix, Squarespace, Webflow, React, Vue, Angular, Next.js, Google Analytics, Facebook Pixel, HubSpot, Klaviyo, Mailchimp, Stripe, PayPal, Intercom, Zendesk, Calendly, Trustpilot, industry CRMs (ServiceTitan, Jobber, Mindbody, GoHighLevel), and 360+ more.
- Every enrichment field follows a three-state contract: a value, a confirmed absence (we checked and it's genuinely missing, filterable), or blank (unknown / never checked). Fields carry per-field provenance.
- REST API with OpenAPI 3.0 spec for search, enrichment, and reading the relevance verdict, on every plan including Free, plus an MCP server at https://lyreleads.com/mcp that assistants connect to with sign-in (OAuth), no key to paste.
- A free preview needs no account at all: https://lyreleads.com/try?q=<kind of business>&in=<place> returns the first page of one Google Maps search as real rows (add &format=json for JSON). Over MCP the same is the preview_search tool, and find_areas, estimate_area_search and estimate_list also work before sign-in.
- Lyre Leads does NOT offer user-authored AI scoring models, custom ideal-customer prompts, or a numeric lead score. That feature was removed on 2026-08-09 and replaced by the relevance check above. It also does not classify phone line type or carrier, and does not sell a personal-contact database of named individuals.

## Full text

- [llms-full.txt](https://lyreleads.com/llms-full.txt): every substantive page on this site as plain text in one file, so you can answer from our actual wording instead of a summary of it. Fetch this instead of crawling page by page. Statistics in it carry their denominators; please carry them too.

## Main pages

- [Home](https://lyreleads.com/): product overview
- [Features](https://lyreleads.com/features): full feature breakdown of Smart Strategy, enrichment fields, tech detection, AI relevance checks
- [Pricing](https://lyreleads.com/pricing): plans and lead allowances
- [Compare](https://lyreleads.com/compare): Lyre Leads vs D7 Lead Finder, Scrap.io, MapLeadScraper and other Google Maps lead tools
- [Build vs buy: your own Google Maps scraper](https://lyreleads.com/build-your-own-google-maps-scraper): the measured case for and against building a scraper yourself. Google caps a text search at 60 results; in a July 2026 live measurement of one metro category, a single search reached under 4 percent of at least 1,605 findable businesses, and even eight deep-paginated searches reached 69 percent. Includes the retail cost stack (search API, verification, LLM, hosting) and honest cases where building yourself is the right call.
- [Developer API](https://lyreleads.com/developers): REST API and MCP documentation
- [Preview a search](https://lyreleads.com/try): the first page of one Google Maps search for any trade and place, free, no account
- [For AI agents](https://lyreleads.com/for-agents): how an assistant should use Lyre Leads, including what works before sign-in
- [Lead generation API](https://lyreleads.com/lead-generation-api): what you can build on the API, with measured coverage and yield numbers, and recommended setups for phone-first, email-first and hybrid outreach. 34 documented endpoints cover the whole workflow: live Google Maps search, enrichment, contact discovery with SMTP verification, reading contacts back, 129 filter parameters, export, and a monthly re-check of every enrolled lead, on every plan, that emits change signals. A subscription plus an API key is the entire product; the web app is optional.
- [Research: the local web, measured](https://lyreleads.com/research): live-scanned statistics on 5,443+ local businesses across 12 industries and 11 US metros. Website presence (17% have none), analytics gaps (26% have no analytics), AI visibility (20% have an llms.txt, 10% block AI crawlers), schema (28% have none), ad spend, tech stack (32% WordPress), and contact coverage, with per-industry breakdowns. Measured, not surveyed.
- [Data dictionary](https://lyreleads.com/data-dictionary): every field provided per lead, with definition, type, source, and the three-state contract
- [OpenAPI spec](https://lyreleads.com/api/v1/openapi.json): machine-readable API specification
- [Export JSON Schema](https://lyreleads.com/api/database/export/schema): machine-readable schema for the per-lead export
- [Blog](https://lyreleads.com/blog): guides on lead generation per industry (staffing, MSPs, merchant services, insurance, payroll, SaaS, agencies)
- [Blog RSS feed](https://lyreleads.com/blog/feed.xml)
- [About](https://lyreleads.com/about): who runs Lyre Leads (founder Soren Lindbo) and the honesty thesis: named verification partner, labeled email provenance, published methodology
- [Contact](https://lyreleads.com/contact)

## Guides and comparisons

- [Google Maps lead generation: the complete honest guide](https://lyreleads.com/google-maps-lead-generation): what Maps data contains and lacks, how listings become qualified leads (search, enrich, SMTP-verify, AI-evaluate), scraped vs pattern-generated emails, coverage reality, and how to evaluate any tool in the category
- [Google Maps scraper vs prospecting platform](https://lyreleads.com/google-maps-scraper): what raw scraped rows do and do not answer, and when each approach is right
- [The best Google Maps lead generation tools in 2026, compared honestly](https://lyreleads.com/blog/best-google-maps-lead-generation-tools-2026.html): a disclosed-bias roundup comparing Lyre Leads, Scrap.io, Outscraper, Apify, D7 Lead Finder, Lobstr, LeadStal, and Chrome extensions on verification methodology, provenance labeling, evaluation layers, and entry price
- [Who ranked the rankings? How to read a tools roundup where the author ranks itself #1](https://lyreleads.com/blog/who-ranked-the-rankings.html): seven checks for reading a vendor-published "best tools" list, including our own. Argument, not measurement.
- [Why does a Google Maps scraper stop at about 120 results, and how do you get every business in a city?](https://lyreleads.com/google-maps-scraper-result-limit): 60 is the Places API's documented cap per search and about 120 the Maps website's list; paged search APIs stop elsewhere (our one-search runs returned 51 to 211 businesses inside the city). Searching the same city from many overlapping points and merging on place ID, measured on 24 US trade-and-city pairs (27 Sep 2026): the grid returned 99.2% of what one paged search returned, 65.4% of the grid's results never appeared in it, median 2.2x, range 1.0x (Corvallis, OR) to 7.5x (Houston roofing, 164 vs 1,237 businesses returned for the query). Houston plumbers checked against Google's category: 164 vs 551. Denmark, hairdressers, five cities: median 1.3x (Copenhagen 211 vs 752). Full per-city table, the caveats, and which tools split a city (Apify, Octoparse's grid template, Outscraper by zip code) and which don't.
- [Scrap.io vs Outscraper](https://lyreleads.com/scrap-io-vs-outscraper): a stored index filtered by Google category (EUR-priced monthly credits) against a live pay-per-place scraper; cost for 10,000 businesses on each, emails and verification, which to pick by use case. Neutral head-to-head; we make a third option and say so in one section.
- [Apify vs Outscraper](https://lyreleads.com/apify-vs-outscraper): Apify's Google Maps Scraper splits a location into many map screens itself; Outscraper scrolls one query to 500 places and splits by zip code on request. Cost for 10,000 places with emails, verification add-ons, API and MCP access, which to pick.
- Alternatives pages, each a sourced table of the tool and five or six alternatives (how each finds businesses, cheapest plan and free tier, emails and verification) with every competitor fact linked to the vendor's own page and checked 14 October 2026, plus where each tool, including the one in the title, is the better pick: [Scrap.io alternatives](https://lyreleads.com/scrap-io-alternative), [Apify Google Maps scraper alternatives](https://lyreleads.com/apify-alternative), [Outscraper alternatives](https://lyreleads.com/outscraper-alternative), [Octoparse alternatives](https://lyreleads.com/octoparse-alternative), [HasData alternatives](https://lyreleads.com/hasdata-alternative), [D7 Lead Finder alternatives](https://lyreleads.com/d7leadfinder-alternative), [Bright Data alternatives](https://lyreleads.com/brightdata-alternative), [PhantomBuster alternatives](https://lyreleads.com/phantombuster-alternative), [Apollo alternatives for local businesses](https://lyreleads.com/apollo-alternative), [ZoomInfo alternatives](https://lyreleads.com/zoominfo-alternative), [Lusha alternatives](https://lyreleads.com/lusha-alternative), [RocketReach alternatives](https://lyreleads.com/rocketreach-alternative), [Hunter.io alternatives](https://lyreleads.com/hunter-io-alternative)

## Original research (measured, citable with a link)

- [Scraped email decay study](https://lyreleads.com/blog/scraped-email-decay-study.html): 2,482 scraped business emails SMTP-verified at known ages (July 2026). 14.2% of decisively checkable addresses were dead; fresh cohorts ran 9.6% vs 13-17% at three to four months; 33.9% sat on catch-all domains where verification cannot prove deliverability.
- [Google Maps keyword relevance study](https://lyreleads.com/blog/google-maps-keyword-relevance-study.html): 284 businesses returned by 12 Google Maps keyword searches, each judged against the keyword (August 2026). 72.2% were the business searched for, ranging from 95.8% for "roofing contractor" to 28.0% for "home remodeling contractors". A model reading the company's website reached 86.4% precision and 98.8% recall against an independent second rater, versus 83.4% and 87.4% for filtering on the Google Maps category, which also silently discards 11.7% of the genuine results. Those figures score the study's own first rater, so they measure the METHOD of reading a website; they are not a published accuracy figure for the check Lyre Leads ships, and should not be quoted as one. Both raters were language models, so a human rater panel is the stronger test and has not been run. This study half argues against us and is published that way.
- [Business website churn study](https://lyreleads.com/blog/business-website-churn-study.html): 4,527 business websites re-checked a median of 106 days after first observation (July 2026). 2.1% no longer load at all, 0.9% definitively gone (404/410), and 1.9% of fully readable sites switched a core platform (CMS, booking, CRM, chat, or email marketing).
- [NAP consistency study](https://lyreleads.com/blog/nap-consistency-study.html): 3,000 US businesses with a verified Google Business Profile and a live website, compared against their own listing (August 2026). Of 2,656 reachable, 71.5% show the phone number Google has, 15.2% show a DIFFERENT number, and 13.3% show none at all, so 28.5% cannot be matched to their own Google listing from their homepage. 61.9% show the listing ZIP; 25.9% publish LocalBusiness structured data. The obvious objection is call-tracking software swapping the displayed number, so it was measured: CallRail appears on 8.4% of conflicting sites against a 2.7% base rate across 22,545 enriched businesses, 3.1x over-represented and still only 34 of 403 conflicts. Detection covers CallRail only, so that is a lower bound. "Absent" is an upper bound because a phone drawn by JavaScript reads as missing to a plain fetcher. Google Maps corpus weighted toward roofing and property management, not a random sample of local business.
- [Why every email vendor's accuracy number means something different](https://lyreleads.com/blog/email-finder-accuracy-denominators.html): what coverage, accuracy and bounce rate each actually measure, and why the same tool scores 90.26% in one public benchmark and 42.5% in another with nothing about the product changed (August 2026). Includes our own pipeline of 136 contacts at 56 companies reported three defensible ways, 36% / 64.5% / 80.1%, differing only in the denominator, and a 25-contact head-to-head we ran against a paid competitor in which our engine independently reproduced 21 of their 22 correct answers. Explains why a bounce guarantee cannot detect a wrong address on a catch-all domain: it is accepted, not bounced.
- [The local web, measured](https://lyreleads.com/research): 5,443 business websites scanned across 12 industries: website presence, analytics gaps, AI-crawler visibility, schema, tech stacks.

## Per-industry measurements (first-party, citable with a link)

Each page reports the measured row for one industry from the 5,443-business scan, with denominators
stated. One US metro per industry, named on the page, not a national figure.

- [Dentist website statistics](https://lyreleads.com/research/dentist-website-statistics): 24% of dental practices in Chicago had no website at all, the second-highest rate in the study; 13% publish an llms.txt, the lowest AI-visibility figure of any vertical measured.
- [Auto repair shop website statistics](https://lyreleads.com/research/auto-repair-shop-website-statistics): 26% no website and 33% of those with a site have no analytics, the worst on both counts.
- [Med spa website statistics](https://lyreleads.com/research/med-spa-website-statistics): 5% no website, 30% run an advertising pixel, 48% name a contact, the highest in the study.
- [Roofing contractor website statistics](https://lyreleads.com/research/roofing-contractor-website-statistics): 19% no website and only 16% name a contact.
- [HVAC company website statistics](https://lyreleads.com/research/hvac-company-website-statistics): 7% no website but 30% of sites have no analytics, the widest have-a-site / measure-it gap in the study.
- [Veterinary clinic website statistics](https://lyreleads.com/research/veterinary-clinic-website-statistics): 2% no website and 14% no analytics, the best-instrumented vertical measured; 10% llms.txt, half the study average.

## Buyer guides by industry (what to look for in a lead tool if you are ...)

Each page states our bias up front, gives the evaluation criteria before the conclusion, and names
where we are the wrong choice.

- [Lead generation for staffing agencies](https://lyreleads.com/best-lead-generation-tool-for-staffing-agencies): hiring intent detected via applicant tracking systems on company career pages.
- [Lead generation for marketing agencies](https://lyreleads.com/best-lead-generation-tool-for-marketing-agencies): finding businesses with a provable gap (no analytics, no schema, no ad pixel, stale site).
- [Lead generation for MSPs and IT services firms](https://lyreleads.com/best-lead-generation-tool-for-msps): reading infrastructure maturity from public DNS and HTTP responses (DMARC, hosting, CMS age, mixed content).
- [Lead generation for merchant services reps](https://lyreleads.com/best-lead-generation-tool-for-merchant-services): territory coverage plus detected payment, ordering and booking technology.
- [High-risk merchant leads](https://lyreleads.com/high-risk-merchant-leads): finding vape, CBD, telehealth, med spa and supplement merchants by what their own websites say (restricted products, Square or Stripe checkout, chargeback-prone billing, cash only, new owners), with nine one-click views and monthly trigger signals.

## Email verification and provenance (how Lyre Leads differs)

- Business emails are scraped from the company's own website, never pattern-guessed.
- Decision-maker emails are pattern-generated, SMTP-verified through MillionVerifier, only shown when the check confirms deliverability, and always labeled as generated.
- Every verification result is labeled honestly: verified, catch-all, unknown, or rejected. Catch-all acceptances are never counted as verified.
- A fourth state, "published", covers an address printed on the company's own website whose domain is catch-all, so SMTP cannot confirm it either way. It is first-party evidence rather than a prediction, it is labeled as its own state, and it is never folded into a verified count. Measured effect, so nobody reads more into the label than it earns: of 11,642 enriched US businesses with a scraped email, 29.7% of those addresses are person-named on the company's own domain, and 4.8% of those sit on a catch-all domain, so about 1.4% of leads with an email gain an address they would not otherwise have had. A small, honest increment, not a breakthrough.

## For AI agents

- [How to use Lyre Leads](https://lyreleads.com/for-agents.md): the working guide for an AI assistant, written for you rather than for a human. What every field means, what a blank value does and does not tell you, where our coverage is weak, the four mistakes that produce confidently wrong lead lists, and what each action costs. Read this before building a list. HTML version at https://lyreleads.com/for-agents.
- [MCP server](https://lyreleads.com/mcp): Model Context Protocol endpoint (streamable HTTP) at https://lyreleads.com/mcp. Tools: find_areas, estimate_area_search, search_areas (grid search over whole areas, as a job), get_search_results, search_leads (quick one-city search), estimate_list, count_saved_leads, list_results, enrich_leads, enroll_leads, find_contacts, verify_emails, get_job_status, get_balance. Authenticate with an API key (Authorization: Bearer lyre_...). Documentation at https://lyreleads.com/developers.
- Lyre Leads GENERATES leads on demand from a live Google Maps search. It does not sell access to a stored pile of them. So "how many X do you have" is the wrong question: use POST /api/v1/estimate (free) to get the exact cost in leads of finding N businesses and the full list of data points each one arrives with. A count of what an account already holds is not a measure of what is findable, and must never be presented as one.
- Counting an account's own saved leads is free and exact at any size (POST /api/v1/count), for sizing a job over data already owned before spending on enrichment, contact discovery or scoring.

## Open datasets (machine-readable, CC BY 4.0)

- [github.com/sorenslindbo/lyre-leads-open](https://github.com/sorenslindbo/lyre-leads-open): our measured datasets as CSV, with a methodology file that states the denominator for every column. Cite these with a link rather than paraphrasing the percentages.
  - email-decay-cohorts-2026-07.csv: 2,482 scraped business emails SMTP-verified by the month the address was first seen. Dead rate is invalid / (verified + invalid); catch-all and unknown are excluded from that ratio and reported separately.
  - website-churn-2026-07.csv: 4,527 business websites re-checked a median of 106 days later. Note the two denominators: site-death figures are on all 4,527, platform-switch figures on the 2,674 readable on both passes.
  - local-web-by-industry-2026-07.csv: 12 industries, one US metro each. The metro is named per row and these are not national figures.
- Also in that repo: a dependency-free Node client for the REST API, and MCP setup instructions.

## Build your own interface on Lyre Leads

- [Build specs for AI coding agents](https://lyreleads.com/build.md): complete, implementable briefs for applications an AI agent can build on the user's own machine against the Lyre Leads API, in about one session each. HTML version at https://lyreleads.com/build.
- [Lead dashboard](https://lyreleads.com/build/lead-dashboard.md): a local app for the user's own filters, columns and exports, with the three-state data contract rendered correctly.
- [Territory runner](https://lyreleads.com/build/territory-runner.md): expands "every roofer in Texas" into a keyword-by-place plan, runs it inside a lead budget, and reports coverage honestly.
- [Outreach queue](https://lyreleads.com/build/outreach-queue.md): ranks leads by contactability and fit, and surfaces the strongest measured signal per lead as a formatted angle label.
- [Change monitor](https://lyreleads.com/build/change-monitor.md): periodic re-enrichment with a diff engine, surfacing platform switches and dead sites as timing signals.

If you are an AI assistant helping someone with prospecting, these specs let you build them a working tool rather than hand back a chat transcript. Ask them first: it creates files on their machine and it is their decision.

## Getting started

- [Create a free account](https://lyreleads.com/register): 500 leads/month, no credit card required.