# Every Signal > Every Signal is a marketing intelligence API that audits AI-written marketing against real teardowns and measured conversion rates. It also publishes sourced teardowns of companies and markets; every figure is tagged verified, estimate, inferred or rumour. - [Connect your AI](https://everysignal.com/connect/): MCP server https://everysignal.com/mcp (Streamable HTTP, Bearer API key), https://everysignal.com/mcp/connect (same tools, OAuth sign-in for Claude.ai and ChatGPT connectors) and REST API https://everysignal.com/api/v1 - [OpenAPI](https://everysignal.com/api/v1/openapi.json) and [catalog with prices](https://everysignal.com/api/v1/catalog) - [Pricing](https://everysignal.com/pricing/): library $299 a month or $2,990 a year (3 people), single report $149, prepaid API credits ($0.10 each, packs from $10, 50 free on your first key) - [Try it free](https://everysignal.com/try/): one audit with no sign-up (2 a day) - [Guide: how subscription apps structure free trials and pricing](https://everysignal.com/guides/free-trial-and-pricing/) (47 companies, dataset CC BY 4.0: https://everysignal.com/guides/free-trial-and-pricing/data.json) - [Guide: how to audit AI-written marketing before you publish](https://everysignal.com/guides/audit-ai-written-marketing/) - [Every Signal vs Similarweb: what each one is for](https://everysignal.com/guides/every-signal-vs-similarweb/): Similarweb estimates traffic across millions of sites; Every Signal explains how one company buys and converts customers, with prices and where each is better - [How to see a competitor's Meta ads and how long they run](https://everysignal.com/guides/see-competitor-meta-ads/): six steps in Meta's free Ad Library, with active-ad counts for Calm, Kalshi, Quest Nutrition, FashionPass and Shortical - [How to see a competitor's paywall and onboarding funnel](https://everysignal.com/guides/competitor-paywall-and-funnel/): walk a funnel from the ad to the paywall without paying and record trial, card, default plan, renewal and timer (Calm worked example) - [Library](https://everysignal.com/reports/): all public reports - [Full text for LLMs](https://everysignal.com/llms-full.txt) - [Claude skill](https://everysignal.com/skill/SKILL.md) ([zip](https://everysignal.com/skill/every-signal-skill.zip)) and the official MCP registry entry com.everysignal/everysignal ## Tools (MCP and REST) - audit_work (30 credits, $3): audit copy, a landing page, an email, an ad, a funnel or a paywall your AI wrote; returns score, findings with KB record ids, fixes and a rewrite - analyze_landing_page (100 credits, $10): tech stack, copy, speed and funnel signals for one URL against how the sector sells; ranked fixes and a hero rewrite - analyze_ads (150 credits, $15): ad creatives (image URLs, ad texts) or a brand's live Meta ads scored against real ad patterns; fixes and new hooks - growth_audit (150 credits, $15): what a domain ranks for, its search and AI-answer footprint and tracking, against the sector's growth moves; a ranked growth plan - intel_report (490 credits, $49): automated snapshot of a company (tracking stack, live Meta ads, search footprint, homepage) with a machine summary - kb_sectors, kb_brief, kb_playbook (free): sectors, sector briefs, task playbooks (homepage, email, paid social, subscription funnel, checkout, geo) - kb_search, kb_examples, kb_benchmarks, kb_get (1 credit each): records, real copy, measured conversion rates - list_reports, get_report, job_status, credits_balance (free) ## Reports ### [Packdraw: how does a crypto-cashout pack casino reach $15-25M a month with no Meta ads?](https://everysignal.com/reports/packdraw/) (2026-10-29) Packdraw sells virtual packs from $0.11 to $22,972 and keeps its edge in the odds, not in ads. On 11 packs the expected value is 86.8-90.8% of price, and items sell back at 100% of Packdraw's own displayed value or leave as crypto. The report puts pack volume at roughly $15-25M a month and the take at $1.5-3M a month (inferred, October 29, 2026). Traffic comes from sponsored YouTube and Kick creators, with zero Meta ads, and the legal posture is weak. - On 11 packs from $0.53 to $22,972, the expected value computed from Packdraw's published odds is 86.8-90.8% of price. (verified) - Items sell back to balance at 100% of their displayed value, and crypto withdrawal starts at $15. (verified) - The home page counter read 78,726,793 packs opened, and the live feed showed about 1,500 opens an hour. (verified) - Pack volume is roughly $15-25M a month and the take is $1.5-3M a month, mostly from recycled credit. (inferred) - Semrush puts the site at 3.58M visits a month, 90% direct and 40% US. (estimate) - Packdraw ran zero Meta ads and zero TikTok ads on October 29, 2026, six Google text ads on the brand term, and no Meta, TikTok or Snap pixel. (verified) - A Minnesota suit cited by review sites could not be confirmed. (rumour) **How does Packdraw make money?** Through the odds. On 11 packs the expected value is 86.8-90.8% of price, so one open, sell-back and re-open loop keeps about 9-13% for the house at Packdraw's own item values. The report estimates a take of $1.5-3M a month (inferred). **Is Packdraw legit?** The report found no gambling licence (review sites), and the terms promise that the value of items "and/or XP" in a pack "shall be equivalent to or surpass" the price while the published odds say most pulls are worth less. Trustpilot shows 3.4 from 478 reviews, 31% one-star, with complaints about "rigged" games, withdrawals and bans. No lawsuit or regulator action against Packdraw itself was found. **Who owns Packdraw?** The Terms name PackDraw US, LLC, a Delaware company, as operator. The footer calls PackDraw a brand of PackDraw Group, made up of Packdraw Limited in Cyprus and the Delaware LLC. The public face is the YouTuber DOOM13. **How does Packdraw get customers?** Through creators, not ad platforms. The report counts about 65 sponsored videos across 11 YouTube and Kick channels, each with a code worth +5% on a deposit up to $200, plus Discord wager leaderboards. There are no Meta or TikTok ads. **Can you cash out of Packdraw?** Yes. Items sell back to balance at 100% of displayed value, or can be withdrawn in crypto from $15. A 1x play-through applies before withdrawal. ### [AI agent cockpits in 2026: is there room for another one?](https://everysignal.com/reports/agent-cockpits/) (2026-10-29) Not as a general product. Free open-source cockpits (Orca has 109K GitHub stars) cover the low end, and Coder ($90M Series C led by KKR, AWS agreement), Warp, GitHub Agent HQ and Slack Code (free on every Slack plan) cover the governed top. Shared rooms for humans and agents are only partly open, and seven tools in the space shut down, sold or pivoted in 2026. - Orca, a free multi-agent desktop app, had 109K GitHub stars in October 2026. (verified) - Coder raised a $90M Series C led by KKR and signed a strategic agreement with AWS on July 23, 2026. (verified) - Slack Code launched August 20, 2026 on every Slack plan at no extra charge. (estimate) - Conductor charges $50 a month for Pro and $60 per user for Teams. (verified) - Warp Business costs $50 per user a month; Max is $200 a month. (verified) **What is an AI agent cockpit?** A front end for running and supervising several AI coding agents (Claude Code, Codex, OpenCode and others) at once, usually with persistent sessions, a status view of which agent needs input, and phone or web access. **Which AI agent orchestration tools are free?** Orca, Happy, Paseo, Codeman, Agent Deck and Vibe Kanban are free and open source; Coder's Community edition is free for self-hosting with up to 5 agents. **Who is making money in AI agent orchestration?** Mostly governance and infrastructure companies: Coder, Warp, and the platforms that bundle agents (GitHub, Slack). Conductor is one of the few UI-first tools charging per user. ### [Why AI coding tool startups shut down in 2026](https://everysignal.com/reports/ai-coding-tools-shutdowns-2026/) (2026-10-29) Mostly for lack of paying users. Vibe Kanban (28K GitHub stars) said most users were free and it couldn't find a business model; the agent makers then shipped the same features (parallel sessions, desktop apps, cloud agents) inside their own products, and wrapping a fast-changing CLI proved fragile. Work is also moving from the laptop to cloud agents, which favours companies with infrastructure and distribution. - Vibe Kanban's company shut down on April 10, 2026, saying the vast majority of users were free. (verified) - Terragon released an open-source snapshot dated January 16, 2026 at the time of its shutdown. (verified) - Roo Code stopped its VS Code extension in April 2026 after 3M+ installs and archived the repo on May 15. (estimate) - Cursor acqui-hired Continue (34K stars) in June 2026 and shut the product down. (estimate) - Crystal was deprecated in February 2026 and replaced by Nimbalyst. (verified) **Why did Vibe Kanban shut down?** Its April 10, 2026 shutdown post said the vast majority of users were free and the team couldn't find a business model it could get excited about. The project continues as community-maintained open source. **What happened to Roo Code?** Roo Code stopped work on its VS Code extension in April 2026, archived the repository on May 15, and moved to a cloud agent; its founder called editor-based tools a transitional phase. **Which AI coding tool companies are still growing?** Infrastructure and governance players such as Coder and Warp, platforms that bundle agents (GitHub, Slack), and a few paid UI tools such as Conductor. ### [AI employee memory and handover tools in 2026](https://everysignal.com/reports/ai-employee-memory/) (2026-10-29) Partly. Capturing each person's AI and email activity is cheap and already sold in pieces (vendor compliance exports, AI gateways, email APIs), and Viven ($35M seed, Khosla) builds AI twins of each employee for large companies. No product we found turns AI-tool activity into a structured successor handover for small teams. The hard parts are rare use (few departures a year) and monitoring law: written consent and notice in New York, Connecticut and Delaware, impact assessments and works councils in the EU. - Viven raised a $35M seed from Khosla Ventures and Foundation Capital in October 2025 to build AI digital twins of employees. (verified) - Sensay runs AI-led exit interviews in Slack and Teams and says over 200 companies use it. (verified) - ChatGPT Enterprise's Compliance API holds conversation logs for about 30 days, so they must be exported continuously. (estimate) - Harmonic Security raised a $50M Series B on October 14, 2026 for employee AI-usage security. (verified) - The EU AI Act's high-risk rules for systems that monitor or evaluate workers now apply from December 2, 2027. (estimate) **Can a company record employees' AI chats?** In the US, on company-controlled accounts and with written employee consent and notice (New York, Connecticut and Delaware require notice). In the EU it needs a legitimate-interest basis, a data-protection impact assessment and, in Germany, works-council approval. This is not legal advice. **What is an AI digital twin of an employee?** An AI model built from a person's work (email, documents, meetings, chats) that colleagues can query. Viven sells this to large companies with privacy controls and audit trails. **How do companies keep knowledge when employees leave?** Options include AI exit interviews (Sensay), offboarding workflows (ServiceNow), company search (Glean, Microsoft), and per-employee twins (Viven). A handover built automatically from AI-tool activity is still largely unserved for small teams. ### [Kalshi vs Polymarket: who is winning prediction markets, and how does each buy users?](https://everysignal.com/reports/kalshi-vs-polymarket/) (2026-10-29) Kalshi is winning the US money; Polymarket wins global attention. Kalshi did roughly 70-80% of combined trading volume in recent months and earned about 1.15% of volume in 2025 ($263.5M on $22.9B), while Polymarket has more web visits (about 24M vs 19M over three months) but only began charging fees in 2026. Kalshi runs about 1,100 live Meta ads against Polymarket's roughly 100, all of them under 8 days old. - Kalshi held roughly 70-80% of combined Kalshi and Polymarket trading volume in mid-to-late 2026. (estimate) - Kalshi earned $263.5M in fee revenue in 2025, about 1.15% of its $22.9B volume. (estimate) - Kalshi had about 1,100 active Meta ads on one Page in late October 2026, while Polymarket had about 100, all started October 22-28 and offering $25 with code FREE25. (verified) - Similarweb counted about 19.4M visits to kalshi.com and 24.2M to polymarket.com over the three months to September 2026, with 86.1% of Kalshi traffic from the US against 28.4% for Polymarket. (estimate) - Trustpilot rated Kalshi 2.7 out of 5 from 407 reviews and Polymarket 1.3 out of 5 from 655 reviews in October 2026. (verified) **Who has more trading volume, Kalshi or Polymarket?** Kalshi, by a wide margin. The report estimates it did roughly 70-80% of combined volume in mid-to-late 2026, with a June 2026 split of about 70% Kalshi, 23% Polymarket global and 7% Polymarket US. **How does Kalshi get users?** Mostly paid media: about 1,100 live Meta ads, Google ads, connected TV, podcasts, Spotify and sports and crypto ad networks, landing on its sign-up page. Search traffic is 76.6% organic and 23.4% paid. **How do Kalshi and Polymarket charge fees?** Kalshi charges takers 0.07 x contracts x P x (1-P) on event contracts. Polymarket charges takers by category (crypto 0.07, sports 0.05, politics 0.04, geopolitics 0), makers pay nothing, and fees only began in 2026. ### [American Hartford Gold: how does the gold dealer sell and earn?](https://everysignal.com/reports/american-hartford-gold/) (2026-10-29) A high-margin phone sales operation, not a web business. Nothing is priced online: every product page says Call For Pricing, and the contract discloses spreads of 1-20% on bullion, up to 40% on exclusive coins and up to 60% on numismatics. On a claimed $0.8B a year of metal delivered, that implies roughly $60-200M of gross profit a year. Customers come from TV, radio, podcasts and Newsmax email, with only about 157K web visits a month and 6 Meta ads. - American Hartford Gold's September 2026 transaction agreement discloses spreads of 1-20% on bullion, up to 40% on AHG exclusive coins and up to 60% on certain numismatics. (verified) - On October 29, 2026 the company had 6 active Meta ads in 4 campaigns, all launched within the previous 30 days and all video. (verified) - On October 5, 2026 American Hartford Gold bought the assets and customer records of Rosland Capital out of Chapter 11, assuming none of its liabilities. (verified) - Similarweb put the site at about 157K visits a month for July to September 2026, down from about 218K. (estimate) - If the claimed move from $2B delivered in mid-2023 to $4B in January 2026 counts invoiced sales, gross profit is roughly $60-200M a year, at low to medium confidence. (inferred) **How does American Hartford Gold make money?** Mainly through the spread over its own cost, disclosed only in its contract PDF: 1-20% on bullion, up to 40% on exclusive and semi-numismatic coins and up to 60% on numismatics. Reps are commissioned on volume and margin. **Does American Hartford Gold guarantee a buyback?** No. Its marketing mentions a buy-back commitment, but the contract states AHG makes no guarantee of repurchase or of a repurchase price. Complaint sites cite buyback offers 30-40% under purchase price, which the report treats as a rumour. **How does American Hartford Gold get customers?** Through endorsers such as Bill O'Reilly and Rick Harrison, TV and radio, podcasts and Newsmax email, then a phone call to a commissioned rep. Its Meta spend is tiny and the site is secondary. ### [OpenRouter: can anyone still compete with the LLM router Stripe just bought?](https://everysignal.com/reports/openrouter/) (2026-10-28) Not head-on. Stripe agreed to acquire OpenRouter on Aug 19, 2026 (reported at $7-7.5B). OpenRouter makes money from a 5.5% fee on credit purchases, not model markups, routes roughly 70 trillion tokens a week across 457 models, and grows almost entirely through coding-agent integrations and its public rankings. The routing itself is easy to copy and free rivals exist; the openings are quality-based vertical routing, per-customer AI margin tools and regulated routing. - Stripe agreed to acquire OpenRouter on August 19, 2026. (verified) - Reported deal value: over $7B (Bloomberg) to about $7.5B (NYT). (estimate) - OpenRouter charges 5.5% on credit purchases (8% on Business) and says there is no markup on inference prices. (verified) - 457 models are listed in OpenRouter's public models API (October 2026). (verified) - Annualized revenue was about $160M in August 2026 (Sacra). (estimate) - The OpenRouter website runs no advertising pixels. (verified) **How does OpenRouter make money?** Mainly from a 5.5% fee on credit purchases (8% on the Business plan; 5% on crypto). It passes model prices through without markup and charges 5% on bring-your-own-key usage above $25K a month. **Who bought OpenRouter?** Stripe agreed to acquire OpenRouter on August 19, 2026. The price was not disclosed; Bloomberg reported over $7B and the New York Times about $7.5B. **What are OpenRouter's main competitors?** Vercel AI Gateway, Cloudflare AI Gateway and open-source LiteLLM (all 0% markup), Portkey and Helicone (subscription gateways), Requesty (about 5% markup), and the clouds' own multi-model services (Bedrock, Vertex, Azure). ### [Kalshi's perps referral program: who loses, and who gets paid?](https://everysignal.com/reports/kalshi-perps-referrals/) (2026-10-28) Partly. Kalshi's perps referral program pays promoters a share of their referrals' trading fees weekly, and fees (0.12% per side at the base tier) are charged on the full leveraged position whether the trader wins or loses. But most users were already losing before perps launched on June 3, 2026: Kalshi itself cited 2.9 losing users per winner in April 2026. Claims that Kalshi profits from putting customers in debt are unproven. - Kalshi filed a Perpetuals Referral Program with the CFTC on July 24, 2026 that pays referrers a share of referred users' net trading fees. (verified) - Kalshi's base-tier perps taker fee is 0.12% per side, charged on notional value. (verified) - Kalshi's spokesperson said there were 2.9 unprofitable users for every profitable one in April 2026. (estimate) - The Roosevelt Institute estimates market takers lost $583.5M on Kalshi from July 2021 to May 15, 2026; Kalshi disputes the method. (estimate) - 51% of about 150 surveyed prediction-market users told BadCredit.org they had borrowed to fund bets. (estimate) **Do most Kalshi users lose money?** Kalshi's spokesperson said there were 2.9 losing users for every winning one in April 2026, a one-month snapshot. The Roosevelt Institute estimates takers lost $583.5M from 2021 to May 2026; Kalshi disputes how that study classifies users. **How does the Kalshi perps referral program pay?** Referrers receive a share of the trading fees their referred users pay on perpetual futures, paid weekly; referred users get a temporary fee rebate. The exact percentage varies by campaign. **Can you lose more than you deposit on Kalshi perps?** Positions are liquidated at maintenance margin to keep balances from going negative, but Kalshi notes a fast price gap can still close a position at a worse price, including into a negative balance. Event contracts cannot go negative. ### [Astrology apps: how Co-Star, The Pattern, Mirror and CHANI grew](https://everysignal.com/reports/astrology-apps/) (2026-10-27) Mostly without ads. Co-Star reached 20M+ downloads without real marketing by writing daily lines people screenshot and by making friends the product, and Midjourney bought it in spring 2026 (about 4.3M monthly users). CHANI is the top-grossing US astrology app with no ads or investors, and The Pattern grew on word of mouth. The paid-funnel apps (Nebula, Astroline, Hint) instead rely on long quizzes, fake accuracy meters and weekly subscriptions. - Midjourney announced on July 24, 2026 that it had bought Co-Star; the deal closed that spring and terms weren't disclosed. (verified) - Co-Star raised about $21M in total before the sale. (verified) - Co-Star had about 4.3M monthly users at acquisition (Sensor Tower). (estimate) - CHANI was the top-grossing US astrology app in Q1 2026 and runs no ads. (verified) - Co-Star, The Pattern and Mirror all let you add a person by birthday without them joining. (verified) **Why did Midjourney buy Co-Star?** For consumer-app and design talent for its first standalone apps. Co-Star's 24 staff joined, founder Banu Guler became Midjourney's Chief Design Officer, and an astrology image product is planned. **How did Co-Star grow without advertising?** Its short, blunt daily notifications were screenshotted and shared, its design was built for phones and screenshots, and compatibility with friends turned users into referrers. **How do Nebula and Astroline get users?** Paid ads into long web quizzes (14-25 questions) with progress and accuracy meters, an email gate, then weekly subscriptions. ### [SaneBox vs Superhuman: how do the two email tools price and convert?](https://everysignal.com/reports/sanebox-vs-superhuman/) (2026-10-26) They are opposites. SaneBox is a lean, mature business with $9.49-42.99 a month plans, a 7-day trial without a card, 19 Meta ads and about 116K site visits a month. Superhuman Mail charges $25-40 per seat, takes a card up front for a 30-day trial, and runs 55 Meta ads across 3 Pages as part of a Grammarly suite. Neither shows a per-day price. - SaneBox plans in October 2026 ran from $9.49 a month for 1 account to $42.99 a month for 4 accounts, with a 2-year term pre-selected. (verified) - Superhuman Mail costs $30 a month or $25 yearly on Starter and $40 or $33 on Business, with a 30-day trial that takes a card and shows Due today $0.00. (verified) - On October 26, 2026 SaneBox ran 19 Meta ads on 1 Page, all video, while Superhuman ran 55 ads on 3 Pages. (verified) - SaneBox had 116K site visits in September 2026, down 24.5% on the month, while Superhuman's roughly 6M includes Grammarly and Coda traffic. (estimate) - An old founder interview put SaneBox profit just under 50%, churn near 1.5% a month and organic trial-to-paid at 20-25%. (rumour) **How much does SaneBox cost compared with Superhuman?** SaneBox costs $9.49-42.99 a month depending on accounts and features, or less on its 2-year term. Superhuman Mail costs $25-40 per seat per month, with annual billing pre-selected at 17% off. **Do SaneBox and Superhuman require a credit card for the trial?** SaneBox does not: its 7-day trial needs no card, which is taken later on the Billing page. Superhuman's 30-day trial takes a card at checkout through Stripe, with $0 due today. **Which email providers does each tool support?** SaneBox works with anything offering IMAP, EWS or ActiveSync, including Gmail, Microsoft 365, Outlook.com, Yahoo, iCloud and Fastmail, but not POP-only or ProtonMail. Superhuman Mail supports Gmail and Outlook only. ### [FashionPass vs Nuuly and Rent the Runway: who is winning clothing rental?](https://everysignal.com/reports/clothing-rental-fashionpass/) (2026-10-23) Nuuly. It has about 484K subscribers, $568M in FY26 revenue (up 50%) and 211 active Meta ads, while Rent the Runway is retreating with 141K subscribers, down 3.8%, after raising its plan to $164 plus shipping. FashionPass has the cheapest core offer, 10 items over 2 shipments for $125 with free shipping, but the smallest ad engine, with 21 Meta ads and 62 Google ads. - Nuuly took in $568M in FY26 revenue, up 50%, with about 484K average subscribers in the July 2026 quarter. (verified) - Rent the Runway ended its July 2026 quarter with 140,826 subscribers, down 3.8%, after raising its 10-item plan from $144 to $164 plus $1.99-4.99 shipping per shipment. (verified) - FashionPass charges $125 a month for 10 items across 2 shipments with free shipping, and its first month is $39 with code EARLY. (verified) - On October 23, 2026 FashionPass ran 21 active Meta ads against Nuuly's 211, and its ads said $75 while its landing pages said $39 then $125. (verified) - Semrush counted 62K-166K monthly visits for FashionPass in February to April, against 2.92M for Nuuly and 1.43M for Rent the Runway. (estimate) **How much does FashionPass cost compared with Nuuly and Rent the Runway?** FashionPass is $125 a month for 10 items with free shipping. Nuuly is $98 for 6 items, and Rent the Runway is $164 for 10 items plus shipping on each shipment. **Is Rent the Runway losing subscribers?** Yes. It ended its July 2026 quarter with 140,826 subscribers, down 3.8% year on year, and management guides to flat subscribers for the rest of the year. Its web visits are down about 40% year on year (estimate). **How big is FashionPass?** Its size is not public. It claims 100,000+ members, no funding was found, and the report infers revenue of roughly $12-45M a year. ### [Comfrt: how did a bootstrapped hoodie brand build a 600,000-affiliate machine?](https://everysignal.com/reports/comfrt/) (2026-10-22) With a three-layer creator machine. Comfrt combines commission-only TikTok Shop sellers, a 10% Ambassador link program and paid ads run through creators' own identities to advertorial pages. The founder claims $500-700M in 2025 with no outside funding; the report's traffic and ad-spend math supports roughly $400-700M a year. - Comfrt claims about 600K affiliate sign-ups and about 500 core creators. (rumour) - The Ambassador program pays 10% of revenue excluding shipping and tax, runs on Growi, and uses last-click attribution. (verified) - In one 30-day window about 9,393 creators posted Comfrt on TikTok, with about $9.4M in TikTok Shop sales. (estimate) - Meta Ad Library showed about 6,100 Comfrt ads all-time and about 5,900 active in October 2026, with about 50 creator Pages starting ads in September and October. (verified) - Traffic and ad-spend math supports roughly $400-700M a year, while the founder's claim of $1B in 2026 would need a much stronger fourth quarter. (inferred) **How does Comfrt's affiliate program work?** Three layers: commission-only TikTok Shop creators, a hand-selected Ambassador program paying 10% on Growi links with auto-minted discount codes, and publishers through a CJ affiliate network. **How much revenue does Comfrt make?** The founder claims $500-700M for 2025. The report's inferred range from website traffic, TikTok Shop sales and Meta spend is roughly $400-700M a year; the $1B 2026 claim is unproven. **Who sells Comfrt on TikTok?** Mostly small creators. Of 252 profiled, 106 were micro (10-50K followers) and 89 nano, with a median of 17K followers, though the top 10 affiliates drive over half of creator sales. ### [High Purity Labs: how does a five-month-old research peptide store grow with no ads?](https://everysignal.com/reports/high-purity-labs/) (2026-10-20) Through partners, not ads. High Purity Labs runs no ads on Meta or Google and no tracking pixels, and instead uses a private affiliate program on hidden subdomains, most likely paid through discount codes. It is small: Instagram has about 2.1K followers and the report models roughly $40K a month in revenue, in a range of about $8K to $150K. - highpuritylabs.com was registered in April 2026 and the store went live on May 20, 2026 with 67 products, mostly $65 to $95 a vial. (verified) - Meta Ad Library and Google Ads Transparency showed zero ads for the domain in October 2026, and the site carries no ad pixels. (verified) - Three subdomains (affiliates., go. and app.) serve a private Affiliate Operations sign-in with commission review and dual-control payouts. (verified) - Checkout accepts only Zelle, Cash App or Bitcoin, with no card processor anywhere in the code. (verified) - The report models about 15K visits a month and roughly $40K in monthly revenue, in a range of about $8K to $150K, because no traffic tool ranks the domain. (inferred) **How does High Purity Labs get customers?** Mainly through a private affiliate and referral network, an Instagram account with about 2,116 followers, and a manually approved wholesale tier. It buys no Meta or Google ads and has no SEO footprint. **How does High Purity Labs take payment?** Only by Zelle to a personal Gmail address, a Cash App handle or Bitcoin, which is how sellers in this category avoid card-processor underwriting. **How much revenue does High Purity Labs make?** Unknown. The report's inferred central case is about $38K a month, from 15,000 visits at 1.5% conversion and a $170 average order, with a range of $7.5K to $150K. ### [Shortical: does the micro-drama app really clear $6M a month?](https://everysignal.com/reports/shortical/) (2026-10-19) Not as profit. Everything visible from outside points to roughly $3-8M a month in gross revenue, bought almost entirely with Meta ads (3,651 active ads across 21 Facebook Pages in October 2026). Clearing $6M a month after ad spend would need about $12-20M a month in gross revenue and 15-40% of new users paying, against a 1-5% industry norm. The company's own line, about $100M revenue by year-end, averages roughly $8M a month gross. - Shortical had 3,651 active Meta ads across 21 Facebook Pages on October 19, 2026. (verified) - 1,570 new Meta ads launched October 12-18, 2026, up from 226 the week before. (verified) - Shortical took a $100M credit line from PvX Partners in July 2026 to fund ad spend. (verified) - Similarweb counted 1.1M site visits in September 2026, up 140% month on month. (estimate) - Gross revenue of roughly $3-8M a month fits the visible evidence; $6M a month profit after ads does not. (inferred) **How does Shortical make money?** Paid access to vertical micro-drama episodes: episode 1 plays free, then a paywall offers discounted first purchases and subscriptions, sold through app stores and two web checkouts. **How does Shortical get users?** Almost entirely through Meta ads: thousands of active ads across more than 20 Facebook Pages, with new creative launched in weekly waves. We found little Google or TikTok advertising. **Is Shortical making $6M a month?** Not as profit after ad spend, on the public evidence. As gross revenue it's plausible; the company itself says it's on track for about $100M in revenue by year-end. ### [Rotten: how does a better-for-you candy brand actually grow?](https://everysignal.com/reports/rotten/) (2026-10-19) Through one paid channel feeding three funnels. Rotten runs about 110 live Meta ads and nothing on Google or TikTok, sending traffic to its own Shopify store, an Amazon store selling roughly 3,000 units a month, and a retail rollout it claims reaches 6,000+ doors. It is a real, small-but-growing brand, but no priced round or SEC Form D is on file. - Rotten had about 110 active Meta ads from one Page, all started between June 25 and September 29, 2026, and zero ads on Google or TikTok. (verified) - Summing Amazon's own bought-past-month badges gives roughly 2,950 or more units a month, with the hero listing at 4.2 stars and 488 reviews. (verified) - Similarweb counted about 33K visits to eatrotten.com in September 2026, down 33% month on month. (estimate) - Rotten won the 2026 NCA Most Innovative New Product Award at the Sweets and Snacks Expo in May 2026 for Sour Tiniez. (verified) - No priced venture round or SEC Form D was found; the paper trail is a roughly $38K Kickstarter from 2021 and the brand claims 6,000+ retail doors. (verified) **Where does Rotten advertise?** Only on Meta. About 110 ads from one Page run across Facebook, Instagram, Messenger, Audience Network and Threads, landing on advertorial pages on shop.eatrotten.com. **Where does Rotten sell its candy?** Three places: its own Shopify store, an Amazon store with about 10 SKUs, and retail including Sprouts (Nov 2025), Kroger (May 2026) and Costco. **Has Rotten raised venture funding?** Not on any public record. The report found no priced round, SEC Form D or Crunchbase funding profile, only a roughly $38K Kickstarter in 2021 and undisclosed private capital. ### [Face-matched photo-sharing apps: why PicSee, Facebook Moments and Flock died](https://everysignal.com/reports/camera-roll-photo-apps/) (2026-10-19) Not as a standalone social network. Three teams built almost exactly this and shut down: PicSee in 2026, Facebook Moments in 2019 and Flock in 2014. The need is real but comes in bursts after a party, the app is empty until a whole friend group joins, and sending photos is already free through AirDrop, iMessage and Google Photos. Face-matched delivery has survived only as a paid utility for events and schools, such as Waldo. - PicSee raised $4.25M, reached about 10K Android downloads and 150K photos exchanged, then shut in July 2026. (verified) - Facebook Moments ran from 2015 to 2019 and only hit number one in the App Store when Facebook threatened to delete synced photos unless people installed it. (verified) - Google shut down Flock on January 31, 2014, after buying its maker Bump, and absorbed the idea into Google Photos. (verified) - Google paid a $100M Illinois biometric settlement over Photos face grouping, and Facebook paid $650M over tag suggestions. (verified) - Poparazzi, which raised $17M, fell from about 4M monthly users to roughly 2-3K before shutting in June 2023. (estimate) **Why did PicSee shut down?** The need for photos from friends is episodic, huge the morning after a party and gone two days later, and the app was empty until a whole friend group joined. PicSee returned 60-65% of its capital in July 2026. **Are face-matched photo apps legal?** They carry real risk. Illinois BIPA settlements reached $650M for Facebook and $100M for Google, and Texas CUBI payouts reached $1.4B for Meta. A 7th Circuit ruling in August 2026 (G.T. v. Samsung) helps on-device face templates, but only in that circuit. **Which face-matched photo apps are still live?** Waldo (event albums, 25K ratings, paid), SWSH (college albums), Kwikpic, Google Photos, Apple's iCloud Shared Library, and the new Japanese app Paparazzi - Shot by friends from September 2026.