Loopcard — financial viability (deep)

Date of record: 18 September 2026
Analyst posture: adversarial finance. No hopium. No invented ARR.
Product: Loopcard (Virtus Labs) — shareable coding-agent bill cards. Preview: https://cryptobook.space. New domain planned (loopcard.ai / .app / getloopcard.com).
Price on the spec: free at launch; $19/mo Pro and $49/mo Studio “later.” No Stripe account exists today.
Bar inherited from the challenge pass: ~$5,000 MRR by month 12, one already-paid VPS, PLG / social only, no enterprise sales.
This file does not underwrite that bar. It tests whether Loopcard can clear it, and whether it is worth running even if it cannot.

Verdict (read this first):

Nobody dies on COGS. The product dies on willingness to pay for a second copy of a screenshot they already posted, against a $0 CLI that 18k GitHub stars already run, and against a free-at-launch decision that gives the viral cohort the whole job for $0.

This is not audited accounting. VPS is sunk. No analogue below discloses ARR that we opened. Where a vendor only publishes a price, that is a price, not a market.

0. Method, bar, and what we refuse to count

0.1 What “viable” means here

Same bar as docs/04-financial-plan.md and challenges/finance.md:

“Viable as a standalone SaaS” means: a base case that reaches that headcount and keeps it against churn, with named distribution, without ads, without a salesperson, without assuming a miracle conversion rate the analogues refute.

“Viable as a Virtus marketing asset” means: incremental cash cost near zero, a screenshot people actually post, and a plausible (not proven) path to inbound for Virtus Labs. That is a brand bet, not a P&L.

0.2 What we will not do

• Invent ARR, G2 scores, or paid-user counts for SuperPenguin, Tokenade, ShieldMCP, MCP Observatory, Snappify, or SessionWatcher. We opened prices and traction signals. Signals are not deposits.
• Treat Bain’s ~$100B “agentic glue” or MarketsandMarkets’ $19.33B “agentic AI” as a customer list. See docs/02-market-study.md.
• Treat cryptobook.space leftover traffic as a PLG channel. Seven days to 18 Sep 2026: ~1,503 hits, ~65 unique IPs, almost all old social paths.
• Treat Product Hunt upvotes as conversion. AI Toolbox 3.0: 410 upvotes; maker wrote 23 people reached the pricing page (docs/02-market-study.md).
• Use the killed Slip stretch model. It assumed WTP for unwatermarked receipts. ShieldMCP’s actual paid object is a $49 PDF with 5 PH upvotes and 0 reviews.

0.3 Timing constraint that the SKU itself created

Launch is free only. Billing is “later.” That is a virality choice and a finance tax:

1. The people who feel the $6,000 overnight or the $1,263 Cursor month will complete the job (download PNG, post it) before a paywall exists.
2. Turning Stripe on in M4–M7 converts the residue, not the spike.
3. Watermarking a card after the internet already has unwatermarked demo cards of the same $6k story is not a cash event. It is a brand preference.

All paid P&L below assumes Stripe exists from the start of month 4. If billing slips past the launch cohort (M7+), haircut paying counts another 30–50%. If billing never ships, MRR is $0 by construction and the rest of this file is a Virtus-asset memo.

1. Job-to-be-done vs willingness to pay

1.1 The job, stated without decoration

From docs/11-loopcard-product.md:

Job: “Make my agent bill into a card people will share.”
Aha: one click, no account, no API keys.

That sentence is the whole product. It is also the whole problem.

The aha is the PNG. The PNG is free. Files parse in-browser. Publish stores aggregates only. Demo needs no file. The user who arrived from a screenshot of someone else’s $6,000 overnight can leave with their own card in under a minute without creating an account.

Jobs that convert a Visa in this band, from the demand pass (research/demand-who-pays.md):

Loopcard is not in rows 1–5. It sits next to Carbon / Ray.so (pretty shareable images) and next to ccusage (local cost tables). Those two categories have opposite economics, and Loopcard picked the worse of both: the shareable-image job, priced like a SaaS sidecar.

1.2 What the $19 is actually selling

Compare to products that do collect $19–$30 from the same humans:

• Tokenade Pro $24.90/mo excl. tax — “For $100–$200/mo Anthropic plans.” The meter is tokens saved. Leaderboard claim: −38.9% mean cost vs plain Claude Code. The buyer can do arithmetic: if Max is $100–$200, 39% is tens of dollars a month. WTP is a fraction of the insured bill.
• SuperPenguin Growth $30/mo — up to $5k managed spend, 3 seats, Slack/email/Discord alerts. The buyer is attributing a real invoice, not decorating it.
• SessionWatcher Solo $6.99 one-time (Pro $59 lifetime or $24/yr) — glanceable rate-limit rings so you do not lose a coding session. The testimonial on their own page: “the one time payment is the most worth it money I've spent this year.” Revealed WTP for usage visibility in this niche is coffee-to-dinner, once, not $228/year.

Loopcard’s $19 does not stop a runaway loop, does not cut tokens, does not attribute spend to a PR, and does not sit in the menu bar. It removes a watermark from a PNG the user already posted, and keeps 90 days of numbers they can see in Anthropic’s dashboard and in npx ccusage.

1.3 Who feels the pain, and what they already do

Public stories the product is built on (labeled recreations, not customer accounts):

• Overnight Claude Code: ~$6,000
• Cursor month on HN: $1,263 — founder shipped OSS, not a $19 gateway
• Uber: entire 2026 AI budget in four months on Claude Code
• 30-team audit: median $480/mo, 90th $1,650

Those are real invoices. They are not Loopcard invoices. The incumbent for “what did I spend?” is:

1. The vendor dashboard (Anthropic Usage, Cursor usage, Claude /usage)
2. ccusage / cctally / claude-code-cost — local, $0
3. A screenshot of (1) or a terminal dump of (2), posted to X

Loopcard competes with the screenshot button. That is a brutal incumbent. It is installed on every phone, it is $0, and it is how the $6,000 story spread in the first place.

1.4 Free PNG vs $19/mo — the conversion wall

Artisan Strategies (opened 18 Sep 2026): consumer utilities sit at the low end of freemium because the free tier is “good enough indefinitely.” Content/creative tools (the honest category for a PNG) convert in the middle, and only when the upgrade is export / storage / collaboration. Developer tools sit at the upper end only when the paid trigger is seats, usage, or private repos.

Loopcard’s paid trigger is unwatermarked export of a thing free already exported. That is the stall pattern Artisan lists explicitly: “Free is the whole product — no natural upgrade wall.”

ChartMogul × Growth Unhinged × ProductLed (200 products, Jan 2026, opened):

• Median free-to-paid across all models: 8%, and almost nobody is at the median (bimodal).
• Freemium: 25% of products convert below 2.5%.
• “Good” classic freemium: 3–5%; “great”: 8–12%.
• Ungated freemium (Loopcard’s actual UX): 7% visitor→signup, 8% free→paid of those who signed up — for products where paid is still the job (Lovable credits, etc.).
• Per 1,000 visitors, ungated freemium: 70 signups, 5.6 paying. That is a 0.56% visitor→paid on products that work.

Loopcard is ungated and the ungated object is the product. Apply the ChartMogul ungated funnel without a haircut and you still need ~47,000 visits in a 6-month conversion window to get 263 payers — before churn. Apply the haircuts this SKU deserves (unknown domain, OSS alternative, vanity paywall, billing delayed): visitor→paid lands in 0.05–0.20%, not 0.56%.

Willingness-to-pay, stated harshly:

Insurance WTP is 5–15% of the insured monthly bill when you sell a kill-switch. Loopcard sells a portrait of the corpse. The $6,000 overnight victim does not need a $19 subscription. They need Anthropic’s spend cap, which is already in the console.

1.5 The $19 price is the worst number again

Same finding as the Permit Inbox pass, for a weaker SKU:

• $19/mo is Cursor/Claude Pro money. Buyers compare it to the agent, not to a card.
• Stripe on $19 is heavy (~4.5% domestic, 6–8% with international + Billing; see §5).
• SessionWatcher revealed the ceiling for “see my usage prettily”: $6.99–$59 once.
• Snappify, the only durable paid “pretty developer image” we opened, starts at $5/mo (watermarked free, 40,000+ users claimed, ARR not disclosed). They sell presentations and video, not a bill.
• ShieldMCP asked $49 once for a PDF and got 5 upvotes, 0 reviews.

If Loopcard ever charges, the less-wrong SKUs are $5–$9 one-time PNG pack or $49 Studio as a weekly team ritual. Recurring $19 for an individual is a category error.

2. Analogues (opened 18 September 2026)

None of these disclose ARR on the pages we opened. Treat prices as prices.

2.1 ccusage — $0, and it already has --compact for screenshots

• Site: https://ccusage.com/guide/
• Repo: https://github.com/ccusage/ccusage (org listing 18,574 stars, 836 forks, MIT; README: npx ccusage@latest)
• Job: local token/cost reports across Claude Code, Codex, OpenCode, Amp, Droid, Copilot CLI, Gemini CLI, Grok Build CLI, etc.
• Price: $0. Sponsors on the README (Lineman.io as Team/Enterprise for Claude Code; CodeRabbit; Blacksmith). That is the SSL Labs pattern inside the cost-tracker category: the CLI is free; the upsell is someone else’s team product.
• Feature that should make a Loopcard PM wince: --compact — “Force compact table mode, perfect for screenshots and sharing.” The incumbent already optimized for the viral unit.

Implication: the accounting job is solved at $0, locally, with 18k social proof. Loopcard cannot win on depth. It can only win on beauty. Beauty is a Carbon/Ray.so business, and those did not become $19 SaaS.

2.2 SuperPenguin — $0 / $30 / $200, YC, not a card

Opened https://superpenguin.ai/

Mac app for Coding ROI is free for the engineer; the company pays on managed spend. SuperPenguin is not in the request path (not a proxy). It sells names on dollars.

No ARR disclosed. YC badge is distribution, not a Loopcard analogue. The $30 SKU exists because a team is already spending $2k–$5k on models. Loopcard’s user may have spent $6k once and then turned the agent off.

2.3 Tokenade — $0 / $24.90, sells savings, not a portrait

Opened https://tokenade.net/pricing and https://tokenade.net/en

npm README still lists $19.90/mo and $0.20/M overage; the live pricing page is $24.90 / 10% of API cost. We use the page we opened.

Leaderboard on their homepage: Tokenade −38.9% mean cost vs plain Claude Code. That is a labor/money replacement SKU. Freemium is designed so light users stay free forever and heavy users (the $100–$200 Anthropic plans) hit the cap.

No ARR, no paid-user count opened. Directory-badge carpet on the homepage is a launch tactic, not traction.

Implication: the humans who will pay $20–$25 around agent bills pay to make the bill smaller. Loopcard makes the bill legible and pink. Legibility without a kill-switch is a tweet.

2.4 ShieldMCP — $49 PDF, 5 upvotes, 0 reviews

Opened https://www.producthunt.com/products/shieldmcp

• Free scan: score + flagged issues, ~60 seconds, no account.
• Paid object: $49 one-time for the full PDF with copy-paste JSON fixes.
• Traction on the page we opened: 5 upvotes, 6 followers, 0 reviews, launched ~3 months earlier.

This is the closest paid object to Loopcard Pro: free grade, pay to take the watermark/PDF home. It is also the closest failure signal. A $49 one-time PDF in a scary category (OWASP MCP Top 10, secrets in claude_desktop_config.json) could not get a review. Loopcard’s PDF-equivalent is a flex, not a security report. If ShieldMCP could not sell fear, Loopcard will not sell vibes at $19/mo.

2.5 MCP Observatory — $29/mo for the same paywall Loopcard wants

Opened https://mcp-observatory.com/

• Local scan: free, npx -y @kryptosai/mcp-observatory@latest, no account.
• Hosted: one latest snapshot free.
• Individual Pro $29/month: 90-day history, hosted CI ingestion, regression markers, artifact downloads.
• Teams with a real release decision: $15,000 Release Gate Pilot (1–3 critical servers, ten business days).

Loopcard Pro is “unwatermarked PNG + public URL + 90-day history.” Observatory Pro is “90-day history + hosted CI” on top of a free local scan. Observatory at least ties history to CI and a production approve/gate/defer. Loopcard history is a gallery of old bills.

No self-serve paid-user count opened. Logo row (“developers at Accenture / Cisco / Oracle”) is the pattern our competitive landscape already forbids copying unless true. The existence of a $29 button is not evidence it is pressed.

2.6 SSL Labs — the honest economic analogue

Opened:

• https://www.ssllabs.com/
• https://www.ssllabs.com/about/assessment.html
• https://www.qualys.com/free-services

SSL Labs is explicitly a non-commercial research effort run by Qualys since 2009. The unit of growth is a letter grade people paste into tickets, RFPs, and tweets. There is no $19 SSL Labs. There has never been one.

Every SSL Labs page we opened carries Qualys CTAs: “Qualys Free Trial,” “Try Qualys for free!,” CertView, Cloud Platform. Qualys lists SSL Labs under Free services next to Community Edition, CertView, BrowserCheck.

What Qualys actually bills (reseller / analyst pages, 2026; Qualys does not list VMDR on a public self-serve grid):

• VMDR planning range $199–$250 per asset per year (CyCognito, Beagle Security, UnderDefense, CiphersSecurity — secondary).
• SME packages on qualys.com/subscriptions/sme: VMDR TruRisk starting at $2,195.
• Q2 2026 earnings recap (Investing.com transcript, 4 Aug 2026): full-year 2026 revenue guidance $732–$738 million. SSL Labs is not a line item. It is a top-of-funnel monument.

Loopcard-as-SaaS is trying to be SSL Labs and charge Qualys-Consumer prices. That is not how the monument works. The grade is free; the platform is sales-led and four figures.

Loopcard-as-Virtus-asset is the only SSL Labs reading that is not a category error. Even then: Qualys had a vulnerability-management company before the grade. Virtus is not Qualys. A pink bill card will not conjure a VMDR.

2.7 Extra analogues we opened because the category is “shareable card,” not “MCP”

Analogue scoreboard, one line each:

If you only remember one row: the things that look like Loopcard are free; the things that charge $19–$30 do a different job.

3. Unit economics: social-only CAC, and free→paid 0.5 / 1 / 2.5 / 5%

3.1 Cash CAC if social-only

Paid ads are $0 by policy (docs/04-financial-plan.md). Domain leftover is not a channel.

Social-only does not mean free. It means the operator is the media buy.

Viral screenshot math is also not a CAC of $0 in customers. A card that gets 200k impressions and 4k site visits can still produce zero subscriptions if the visit’s job is “download PNG.” That is the designed funnel.

3.2 Funnel assumptions (named, not vibes)

Which free→paid % is honest for this SKU:

3.3 Conversion grid — M12 paying accounts and MRR

Identity (1-month lag, billing from M4):

new_paid_t = signups_{t-1} × free_to_paid
paying_t = paying_{t-1} × (1 − churn) + new_paid_t

Traffic ramps as in §4. Visitor→signup and churn follow the column (bear/base/bull). The row is free→paid. ARPU follows the column. Numbers rounded to whole customers.

M12 paying accounts

M12 MRR

Read the grid as a warning label, not a menu:

• The cells that clear $5k are bull × 1% or better, or base × something north of 5% (not shown; would need ~227 paying / $22 ≈ 5% of a 20k-visit, 3.5% signup world after 8% churn — it does not close).
• Base × 1% = ~30 paying, ~$660 MRR. That is the cell this file will defend as “competent execution, honest WTP.”
• Bull × 2.5% = ~580 paying, ~$16k MRR is what a pitch deck would circle. It requires ~538k year-1 visits and a conversion rate the SKU’s job does not support and 5% churn on a one-shot PNG. We will not underwrite it.

3.4 Steady-state identity (why $5k is a replacement problem)

To hold 263 Pro accounts at 8% monthly churn you need ~21 net new paying customers every month, forever.

Base M12 traffic is 20,000 visits. Even the 2.5% row is a stretch against that ramp. The 0.5–1% rows need a media site, not a card generator.

At 12% churn (bear, one-shot), 263 payers need ~32 new paying / month. That is a different company.

3.5 LTV (cash only)

LTV ≈ ARPU × gross margin / monthly churn
Gross margin after Stripe ~90% if we never proxy tokens (we must not).

A $19 SKU with 8% monthly churn is a leaky bucket of people who already tweeted. LTV does not rescue a vanity paywall.

4. 12-month and 24-month P&L — bear / base / bull

4.1 Traffic ramps (social-only, new domain; leftover 65 IPs/week is a rounding error)

Bear Y2 decays (novelty tools die; leftover domain does not compound). Base Y2 crawls (posting cadence + a little SEO on the new domain). Bull Y2 mean-reverts (viral peaks do not persist; we do not gift a second hockey stick).

Bull M12 at 80k visits/month is “PH featured + two or three cards that actually get screenshotted + a Reddit thread that is not removed.” It is possible. It is not the median of social launches, and it is not in the operator’s current 65 IPs/week.

4.2 Scenario definitions (one conversion rate each)

Bull uses 2.5% because that is the highest rate we will even print as a scenario, not because we believe it. A “mechanical bull” that also holds 5% conversion is in the grid in §3 and is rejected.

4.3 12-month P&L (cash)

Paying counts are end-of-month. MRR is paying × ARPU. Stripe ~5% blended on $22, ~4.5% on $28, ~6% on $19 (see §5). VPS $0 incremental. Support hours are founder time, not a cash line — they are shown so nobody confuses 90% gross with a business.

Bear

Base

Bull (not underwritten)

Bull clears $5k MRR around M7–M8 on these levers. That sentence is only true if you accept 2.5% free→paid on a free PNG. We do not. A bull that is honest about WTP (1% free→paid, same traffic) lands at ~230 paying, ~$6,400 MRR — still above $5k, still requiring the traffic miracle. A bull that is honest about both WTP (1%) and churn (8% not 5%) lands closer to ~150 paying, ~$4,200 MRR and misses the bar.

Base misses $5k by a factor of eight. Bear is a rounding error.

4.4 24-month P&L

Assumptions: no second product, no ads, no sales team, no Studio miracle. Y2 is the hangover.

Bear Y2. Traffic decays. Paying oscillates between 2 and 6. M24 MRR $40–$110. Cumulative two-year cash collected ≈ $800–$1,500 before Stripe. Not a company. A side project that forgot to die.

Base Y2. Visits crawl 20k → 25k. New paid ~7/month. Steady state at 8% churn ≈ 7 / 0.08 = 88 paying. From M12’s 30, M24 lands ~70–85 paying, $1,500–$1,900 MRR. Cumulative two-year cash ~$18k–$25k. That does not pay a salary. It does not pay back 10 h/week of social. It is a hobby with Stripe.

Bull Y2, mechanical (rejected). If 80k visits and 2.5% conversion and 5% churn held, steady state would be 100+ new paid / month ÷ 0.05 ≈ 2,000+ paying, $50k+ MRR. We will not print that as a forecast. Screenshot toys do not hold peak traffic for 12 silent months.

Bull Y2, mean-reverting (the only bull we will look at). Traffic falls 80k → 40k. Conversion fades to 1% as the audience is no longer “I just got billed.” Churn rises to 8% as one-shot posters cancel. M24 paying ~120–180, MRR $2,600–$5,000. You touch $5k in the lucky half of that band, then yo-yo. Cumulative two-year cash maybe $80k–$140k — still a micro-SaaS, still founder-time negative if you cost the hours, still one viral drought from $2k.

4.5 Cumulative cash (approx., post-Stripe, no salary)

Two-year base cash of ~$22k is less than one month of a junior contractor, and it consumed ~1,000 founder hours of posting. At $50/h that is $50k of time to make $22k of Stripe. Negative on a fully loaded basis.

4.6 Kill tests (write these on the wall)

If billing is on by M6 and any of these fire, stop treating Loopcard as a SaaS and keep it as a free Virtus card:

1. Paying accounts < 10.
2. Free→paid of accounts < 0.8% after 90 days of a visible paywall.
3. Support + CSV-parse tickets > 6 h/week under $1k MRR.
4. Social hours > 12 h/week with < 8k visits/month (the channel is not working; posting more will not create WTP).

5. COGS

5.1 Cash COGS — this VPS is enough, and that does not matter

Cash gross margin on a live $19 SKU: ~90–95%. Correct, and irrelevant. Helicone-class death is not the risk. WTP and distribution are the risk.

5.2 Founder-hour COGS (the real one)

Ticket mix we should expect:

• “My Cursor CSV doesn’t parse” / “Claude export is a zip” / “ccusage JSON vs Anthropic CSV”
• “Why is the number different from /usage?” (ccusage already warns: transcript-derived ≠ invoice)
• “Remove the watermark” from people who will not pay
• “Take this PNG down” (someone else’s bill posted with a name)

Load model: 0.15 tickets / paying / month × 12 minutes + 0.02 tickets / free account / month × 8 minutes, docs-gated.

CSV parsers for vendor exports are a moving target. Anthropic and Cursor will change columns. That is unbounded, quiet COGS. Cap it: if parse tickets exceed 40% of support, freeze new vendor formats.

Social GTM is not COGS; it is CAC. At 10 h/week it dominates the P&L on any fully loaded view until MRR is several thousand, which base never reaches.

5.3 What we must not add (COGS landmines)

• Server-side PNG farm
• LLM “explain my bill”
• Full JSONL ingest of ~/.claude/projects (disk + privacy)
• MCP / token proxy (bandwidth + CFAA-adjacent + Helicone graveyard)
• SOC2 to close Studio deals (kills the 12-month clock)

6. Path to $5k MRR — required? plausible?

6.1 Required (arithmetic, not aspiration)

Required visits at honest 1% free→paid and 3.5% signup: ~60,000 / month, indefinitely (path A). That is 3× the base M12 ramp and 24× leftover-domain reality.

6.2 Plausible?

Required? Only if Loopcard is supposed to be the company. The operator already has TokensMarketCap and Virtus on this VPS. $5k MRR as a hurdle for a sidecar is optional. As a hurdle for “this is the pivot,” it is the bar we were given, and Loopcard does not clear it on base.

Plausible? No. The only $5k pictures require a viral distribution and a conversion rate the free PNG refutes and retention the one-shot job refutes. That is three miracles. Finance will not sign two, let alone three.

7. Alternative monetization

Ranked by honesty, not by how they look on a pricing page.

7.1 Nothing (keep it free) — default

SSL Labs. ccusage. Carbon. Ray.so. mcp-scan.

Cash: $0. Cost: VPS already on, a few hours/month of parser bitrot. Benefit: every posted card can carry a Virtus / Loopcard URL. This is the only model that does not fight the job-to-be-done.

Do not put a fake “Pro coming soon” on the card. That is a conversion tax on the viral unit. If Pro is not live, the card is a free monument.

7.2 Virtus lead-gen — the floor, and only a floor

Mechanism: card footer “Loopcard · Virtus Labs”, link to virtusblockchainlabs.com, optional “get a spend review” that is a human offer, not a $19 login.

What this is worth: unknown. Qualys can do this because VMDR is a $199–$250/asset product with $732M+ guidance. Virtus is not that machine. A realistic year-1 yield if cards actually spread: 5–20 conversations, 0–2 paid engagements. We will not invent the engagement value.

Risk: CryptoBook leftover brand. H1 must say Loopcard, not CryptoBook. Preview on cryptobook.space is a conversion fine until DNS moves.

This is MARGINALLY VIABLE as marketing: cheap, on-brand pink, screenshot-native. It is not a substitute for a Virtus sales motion that does not exist in this memo.

7.3 One-time PNG pack — least-wrong paid SKU

Price: $5–$9 once (Snappify Starter is $5/mo because they have storage/video; we have a PNG). Unlock: unwatermarked export, 4:5 and 16:9, maybe a 3-card pack (overnight / month / team).

Why this is less wrong: SessionWatcher, Token Pacer, ShieldMCP, and the entire “I made a Mac usage app” category priced once. A tweet-shaped job wants a tweet-shaped price.

Ceiling math (not a forecast): 200 packs/month × $7 = $1,400 GMV, not MRR, with the same traffic problem. Does not clear $5k. Might clear coffee money. Stripe on $7 is 7.2% ($0.50) — ugly, like every low-ticket.

Do not dress this up as SaaS.

7.4 Team / Studio $49 — keep as a door, do not lead with it

Only if a studio actually posts weekly cards to clients. That is a content ritual, not a cost-control ritual. No analogue converts this. SuperPenguin’s $30 is the team SKU that exists, and it is attribution + alerts.

Ship Studio only after 10 organic “can my team…?” requests. Until then it is a row on a Notion page.

7.5 What not to do

8. Verdict

8.1 Standalone SaaS: NOT VIABLE

Not “risky.” Not “needs a growth hack.” Not viable at $19 / $49 self-serve on this VPS, on social-only distribution, as a shareable bill card.

Reasons, in order of damage:

1. The free object is the product. ChartMogul ungated math assumes paid is still the job. Here it is not.
2. The accounting incumbent is $0 and famous. ccusage, 18.6k stars, --compact for screenshots.
3. The share incumbent is the OS screenshot. The $6,000 story already spread without Loopcard.
4. The paid analogues that look like us do not convert (ShieldMCP 5 upvotes) or charge $5–$9 / one-time (Snappify, SessionWatcher), not $19/mo.
5. The paid analogues that charge $19–$30 do another job (save tokens, attribute spend, host CI history).
6. Base M12 is ~30 paying, ~$660 MRR. Two-year cash ~$22k. Fully loaded time CAC does not pay back.
7. $5k MRR needs 263 humans at $19 and ~60k visits/month at honest conversion. That is not this domain, this brand, or this job.
8. Free-at-launch then bill later donates the only high-intent cohort to the commons.

Gross margin is fine. WTP is not. Distribution on cryptobook.space is not. A new domain removes a tax; it does not create a market.

8.2 Virtus marketing asset: MARGINALLY VIABLE

Keep the card free. Move the domain off CryptoBook. Put Virtus on the footer. Boom screenshots. Do not staff it like a startup.

Why only marginally:

• SSL Labs worked because Qualys already sold expensive scanning. Virtus does not have a VMDR.
• Lead-gen yield is unmeasured. This memo will not invent it.
• Parser maintenance and social hours are still a tax. Cap social at the hours you would spend anyway on Virtus presence.
• If M6 visits are still in the low thousands and cards are not being posted unprompted, the asset is also dead. Kill it without a funeral.

8.3 What to do with money (there is almost none to spend)

8.4 One paragraph for the operator

Loopcard is a good tweet and a bad SKU. The $19 Pro / $49 Studio grid was copied from a killed permit product that already failed WTP. The market that pays around agent bills pays Cursor $20, Claude $20–$100, Tokenade $24.90 to shrink tokens, or SuperPenguin $30 to name the dollars — or it pays SessionWatcher $6.99 once to see a ring in the menu bar. It does not pay a monthly fee to unscrew a watermark from a PNG it already posted. Run the card as a free Virtus monument in the SSL Labs tradition. Do not tell yourself it is a path to $5k MRR. The arithmetic does not permit it, and the analogues are not shy about saying so.

Sources opened (18 September 2026)

Primary vendor / product pages:

• https://cryptobook.space (preview; nginx unique-IP residual in market study)
• https://ccusage.com/guide/
• https://github.com/ccusage/ccusage
• https://superpenguin.ai/
• https://tokenade.net/pricing
• https://tokenade.net/en
• https://www.producthunt.com/products/shieldmcp
• https://mcp-observatory.com/
• https://www.ssllabs.com/
• https://www.ssllabs.com/about/assessment.html
• https://www.qualys.com/free-services
• https://snappify.com/pricing
• https://sessionwatcher.com/
• https://stripe.com/pricing (geo-served EEA on this pass: Billing 0.7%; EEA card 1.5%+€0.25)

Conversion / churn / fees (opened or used as dated secondary; First Page Sage JS-blocked on fetch, not used as a primary cite):

• https://chartmogul.com/reports/saas-conversion-report/ (Kyle Poyar × ProductLed, 200 products, Jan 2026)
• https://www.artisangrowthstrategies.com/blog/freemium-conversion-rate-benchmarks
• https://churntools.com/churn-rate/developer-tools (median 3.8% monthly; web-only 4–6% in our finance file)
• https://churntools.com/churn-rate-by-size (consumer/prosumer 6–12% monthly)
• https://churncost.com/b2b-saas-churn-benchmarks-2026 (under $1k ACV 6–10% monthly)
• https://retentioncheck.com/churn-benchmarks/developer-tools (2.7% embedded; 4–6% web)
• https://feeprobe.com/stripe-us/ and https://feebreaker.com/us/stripe (US 2.9%+$0.30, citing Stripe; this pass did not get the US Stripe grid because stripe.com geo-redirected)

Qualys / SSL Labs economics (secondary for VMDR $; Qualys itself is quote-ware):

• https://www.qualys.com/subscriptions/sme/ (starting at $2,195)
• https://ca.investing.com/news/transcripts/earnings-call-transcript-qualys-tops-q2-2026-estimates-and-lifts-outlook-93CH-4775928 (FY2026 revenue guidance $732–$738M)
• https://beaglesecurity.com/blog/article/qualys-pricing.html
• https://www.cycognito.com/learn/attack-surface/qualys-pricing

Internal, same date:

• docs/04-financial-plan.md
• docs/11-loopcard-product.md
• docs/02-market-study.md
• docs/03-competitive-landscape.md
• challenges/finance.md
• research/demand-who-pays.md

Not ARR. Not a customer list. Not a reason to open Stripe this month.