Roastmyskin Review: clear offer, uneven execution (74/100)

Fact	Detail
Domain	roastmyskin.com
Category	AI-powered personalized skincare analysis for men
Pricing	$9 entry price; one-time paid digital report
Pages crawled	15
Crawl date	2026-10-22

Roastmyskin is easy to understand and easy to start: it presents a one-time AI skincare analysis for men, a $9 entry price, and a two-minute PDF promise. That clarity is reflected in strong scores for usability (87.1/100), audience and messaging (84.2/100), first impressions and positioning (82.7/100), and technical SEO (89.2/100). The homepage also recorded a Lighthouse accessibility score of 91.

Execution is less consistent beyond the core offer. Design execution scored 41.5/100 after a confirmed 390px overflow on the blog template, Lighthouse contrast failures, and skipped heading structure. Performance scored 54.4/100: the mobile article run recorded an 18.2 s LCP, a 3,079 KiB page, and an estimated 7,040 ms of rendering work. Docs and self-serve help scored 63.5/100, while writing quality (83.3/100), risk and stability (83/100), and editorial QA (77.8/100) show a stronger content and operating foundation.

Roastmyskin's positioning is clear: it offers men a one-time AI skin analysis built around one selfie, the top three skin problems, exact ingredients, and a two-minute AM/PM routine. The hero makes the audience and category legible immediately, while the $9 entry price and PDF format make the commitment concrete. The sharper differentiation sits in the mechanism rather than the headline: the site contrasts a focused dossier with 47 bottles and 200 TikTok routines. Proof is credible as product demonstration, but the sample report is explicitly illustrative and the crawl supplies no attributable customer outcome. Keep the dermatologist boundary, move the top-three-to-routine sequence beside the CTA, and add dated user examples when available.

Roastmyskin understands its intended buyer and gives that buyer a compact job to hire: identify the top three skin problems, choose exact ingredients, and follow a two-minute routine. “Built for men only,” the $9 one-time price, and “PDF in 2 min” answer audience, commitment, and delivery in one flow. “How it works,” “Sample roast,” and “Pricing” also use goal-led labels. Trust is less complete: the About page names the founder and methodology, and the homepage routes cystic acne, rosacea, and skin disease to a board-certified dermatologist, but the supplied crawl gives no attributable outcome or equally specific answer for failed uploads, delivery, refunds, or support. Add that small operational FAQ near purchase.

In the simulated walkthrough, Roastmyskin's primary journey is easy to complete: the homepage states the offer, shows pricing from $9, and presents “Roast my skin” as the action. Blog discovery is also direct, with dated cards and substantive article content visible from the landing page. The friction appears when a skeptical visitor wants business proof. Contact, Terms, and Privacy are footer-oriented, while About is reached through the Blog footer at link depth 2, so trust verification takes a detour from the main path. Terminal-style controls such as “Deploy_Analysis” and “Generate_Report_Now” add interpretation cost even though the underlying destinations are clear. Keep the visual treatment, add plain-language labels, and expose About or Contact earlier.

The homepage has a solid semantic baseline and scored 91 in its Lighthouse accessibility audit, but the purchase path contains a critical naming gap. The captured #bundleForm markup shows email and card-number inputs identified by placeholders, with no associated label, aria-label, or aria-labelledby in the supplied excerpt. Add persistent labels, appropriate autocomplete tokens, and programmatically associated validation messages first. Across the sample, the PSI audit also reports color-contrast failures, and several pages lack a main landmark; the homepage heading sequence skips from h2 to h4. These are confirmed findings, while focus-trap behavior, keyboard order, and screen-reader announcements remain outside this static sample. Restore one main landmark per template and correct the heading levels after the payment fix.

Roastmyskin has a deliberate visual language—black surfaces, neon green accents, large italic display type, and a dominant homepage CTA—but the mobile implementation is not yet dependable. At a 390px viewport, the homepage measures 390px while the blog template reaches a 456px scroll width, creating 66px of horizontal overflow. The same sample records small tap targets across the homepage, Blog, and About pages, plus a 15.4px homepage input. Lighthouse also flags contrast, but the supplied audit does not include the failing color pairs, so exact replacements should wait for affected-node details. Fix the shared blog offender, raise the input to 16px, and establish 44px hit areas before adding more visual surface.

The representative article is the performance bottleneck in the supplied lab runs: mobile LCP reached 18.2 seconds, with 3,079 KiB of total byte weight and an estimated 7,040 ms of render-blocking savings. The image path is a clear first target: hyaluronic_acid_cover.png is listed at 1,610,251 bytes for a 401×252px rendered slot, while PSI estimates 1,538 KiB of image savings. The homepage is better on desktop at 1.4 s LCP, but its 530 ms TBT still signals work on the main thread. Trace the five blocking article requests, make the above-fold asset discoverable early, serve appropriately sized image variants, and recheck mobile lab results. No CrUX field data was supplied.

Roast My Skin's writing is strongest when it gives a practical answer, a concrete condition, and a usable limit. The guides name ingredients, concentrations, prices, and time windows, and the salicylic-acid article offers a decision-ready comparison rather than a generic ingredient overview. Credibility leaks at the edges. “The single best ingredient” is broader than the supplied use cases support, while “90% of results” and “adds 1–2 points” need a source or narrower claim. The affiliate page has a simpler contradiction: metadata says 30% commission, while the hero and feature card say 40%. Tighten universal claims, verify numerical promises, reconcile the payout rate everywhere, and keep the concrete claim-plus-condition pattern that already works.

The available evidence shows a live, indexable site without a confirmed sitewide traffic suppressor, but it does show two practical risk areas. The representative article recorded 18.2 s mobile LCP, 9.7 s FCP, and a 3,079 KiB payload in a lab run; that is a plausible engagement and organic-performance risk, not measured traffic loss because GSC and analytics are not connected. Separately, /blog and /blog/ both return 200 even though the inventory identifies the slash form as canonical. Consolidate the route with a one-hop redirect and retain one form in internal links and the sitemap. The domain has only one Wayback snapshot, so the crawl cannot establish a long-term stability trend.

Roast My Skin has a credible editorial foundation, but the final-pass gate is incomplete. Long-form guides open with direct answers, use question-led H2s, and the hyaluronic-acid article includes Article schema with named PubMed citations and visible numbered references. The remaining QA work is specific: audit “single best,” “90%,” and “1–2 points” against sources; reconcile the affiliate page's 30% metadata and 40% visible payout; and add a title-length check for the 77–112-character article titles. Repeated shell anchors—“chevron_left the_protocol” and “deploy_analysis” each appear 9 times—also crowd out descriptive contextual linking. Keep the current source-rich voice, but make claim verification, metadata consistency, and anchor variation mandatory before publishing.

Roastmyskin has a useful machine-readable orientation but no established self-serve answer hub in the supplied crawl. The 3,740-byte llms.txt states the $9 one-time report, selfie workflow, audience, founder page, and Protocol articles. The 15-page inventory did not discover a dedicated docs, help, or knowledge-base subtree, and it does not establish a searchable FAQ or explicit support escalation. A compact, prominently linked FAQ would close the most important gap: cover selfie upload, report delivery, privacy, refunds, analysis limitations, and a human support route. Keep llms.txt synchronized with those canonical answers so the human and machine-facing surfaces agree.

Roastmyskin's technical SEO foundation is strong: robots.txt allows the public site and declares the sitemap, 15 sampled URLs return 200 with HTTPS canonicals, and raw and rendered word counts match across all 15 pages with js_dependent: false. The main weakness is page experience, not crawlability. The representative article records 18.2 s mobile LCP, 9.7 s FCP, a 3,079 KiB payload, and an estimated 7,040 ms of render-blocking savings. Add Organization and WebSite JSON-LD, consolidate the blog URL, and fix the shared mobile layout after the article critical path.

Roastmyskin earns its 74/100 by making the core purchase understandable: men 18–55 can identify the audience, the skincare job, the $9 one-time commitment, and the promised PDF quickly. Its editorial content adds concrete guidance, including ingredients, concentrations, prices, time windows, and trade-offs.

The next gains are practical. Fix the confirmed 390px overflow on the blog template, address the Lighthouse contrast failures and skipped heading structure, and reduce the mobile article's 18.2 s LCP, 3,079 KiB page weight, and estimated 7,040 ms of rendering work. The product is a good fit for men who want a simple, low-commitment routine and are comfortable uploading a selfie; it is less polished for readers who enter through the article experience.

This 13-dimension review uses a crawl of 15 pages from 2026-10-22, the supplied public dimension score table, product facts detected from JSON-LD and page text, a homepage Lighthouse accessibility result, simulated usability walkthroughs, and raw-versus-rendered page comparisons where supplied. Scores and quoted metrics are preserved from those inputs; this article does not re-score the site.

Two dimensions were excluded as not applicable: 12 · Decision-support surfaces and 13 · Review-content integrity. Google Search Console was not connected or available, so related enrichment remains pending. Pricing is described using the supplied $9 entry price and one-time paid-report model. See How SiteList scores for the review method and data notes.