Hilm Review: strong bilingual builder, weak self-serve help (82/100)

Fact	Detail
Domain	hilm.ai
Category	Arabic-first AI website builder for small businesses, professionals and bilingual company profiles
Pricing	Freemium: Free at USD 0; Pro at USD 5/month billed yearly (USD 60 yearly) or USD 10/month billed monthly
Pages crawled	40
Crawl date	2026-10-21

Hilm's public surface is strongest at explaining and starting the product: Usability and Accessibility both score 100/100, while Technical SEO scores 97.3/100. The first screen makes the category, Arabic/English promise and ask-to-edit workflow legible, and the product answers practical questions about pricing, bilingual publishing and undoing changes.

The main weakness is self-serve help. Docs & self-serve help scores 6.3/100 because the complete 40-page crawl did not discover a public docs, help or knowledge-base surface. Design execution scores 73.3/100, with tap-target compliance as the clearest mechanical defect. Performance remains strong overall at 88.1/100, but mobile is slower: the homepage reaches its largest content at 3.38 seconds, and representative Arabic landing pages reach 3.22 seconds and above. Writing quality scores 80.7/100, and decision-support surfaces score 76/100.

Hilm makes its category and core workflow legible within the first screen, but the first buyer remains broad. The hero says, "Your website, ready in minutes. Edit it just by asking," and explains that a business description becomes an Arabic or English site. Visitors can try a live demo before signup, browse 98 templates, and see Free at USD 0 with Pro at USD 5/month yearly or USD 10/month monthly. That is strong product and price proof. The gap is positioning focus: the hero presents Home kitchen, Wedding photographer, and Law firm paths without choosing one lead segment, while the crawl found no comparison or alternatives page. Keep the broad catalog, but give the first screen a clear segment path and add a named outcome beside its strongest demo.

Hilm answers the practical questions of a small-business prospect unusually well, while leaving the broader choice unresolved. The homepage explains what the product does, who it serves through business examples, and how to start: describe the business, pick a look, edit by asking, then publish and share. Pricing gives Free at USD 0/month and Pro at USD 5/month when billed yearly or USD 10/month monthly, with up to three sites and Arabic and English on Pro. Vertical FAQs add concrete questions about menus, WhatsApp orders, bookings, consultations, and domains. The remaining gap is why this route should beat other approaches: the crawl found no comparison or alternatives page, and the company-profile page mainly contrasts a link with a heavy PDF. Add a concise, evidence-led alternatives comparison and named customer outcomes.

Hilm’s public journey is easy to understand and start, with choice density as the main usability risk. The hero states the Arabic/English website promise, and desktop and mobile captures expose a prominent "Start free" action. Homepage links to Pricing, How it works, and Why hilm are each depth-one routes, while the pricing page presents Free and Pro with clear actions. The supplied mobile evidence preserves the hero, walkthrough, templates, and pricing without an observed clipped or overlapping target. The homepage inventory contains 29 outgoing links across templates, verticals, countries, product variants, legal pages, and repeated calls to action. Group secondary routes by intent while keeping Start free dominant. No live form submission or inline validation run was supplied, so those states remain unassessed.

Hilm’s sampled public surface has a strong automated accessibility baseline, with live interaction still requiring confirmation. The PageSpeed Insights audit reports 100 for the homepage. Its DOM exposes header, nav, main, footer, and section landmarks; declares lang="en"; includes a skip link; reports zero missing image alts across 31 images; and reports seven inputs with detected labels or ARIA names. The hero textarea has a visible label associated with textarea#hero-prompt. Pricing and privacy retain core landmarks, while sampled pages report zero positive tabindex values and no generic link texts. The demo-cafe sample reports hasSkipLink=false, so confirm an equivalent bypass mechanism there. A keyboard and assistive-technology pass is still needed for focus order, dynamic errors, focus trapping, announcements, and captions.

Hilm’s visual system is coherent, but mobile interaction sizing is the clearest design defect. The desktop and mobile captures show a restrained white and pale-lilac canvas, saturated purple actions, warm orange accents, rounded cards, and deliberate section hierarchy. At 390px, however, the layout facts report 38 small tap targets; the homepage example chips measure 129×38px, 182×38px, and 98×38px, below the 44px guidance. Raise actionable controls to at least 44×44px with separation, or render illustrative examples as labels. The pricing comparison also becomes dense on mobile, where two plan columns are squeezed beside feature labels. A stacked mode or horizontally scrollable table would improve scanning. The stylesheet inventory lists 48 colors, 35 font sizes, 32 radii, and 19 spacing values; consolidate active tokens after the interaction fixes.

Hilm is fast on desktop but needs a focused mobile performance pass. PageSpeed Insights measured homepage mobile LCP at 3,384.5 ms, with FCP at 2,401 ms, Speed Index at 4,427.6 ms, TBT at 23 ms, and CLS at 0.001. The same mobile runs recorded LCP of 3.22 s, 3.31 s, and 3.50 s on the representative Arabic AI website-builder, free-website, and how-it-works pages; desktop homepage LCP was 0.56 s. The first fix is the mobile LCP request: use the trace to shorten its critical chain. Transfer weight is also material, at 2,661 KB across 67 homepage requests and 2,601 KB across 116 template requests. Review the estimated 891 KiB homepage image savings, 58 KiB unused JavaScript, and 150 ms render-blocking opportunity. These are lab results; no CrUX field data was supplied.

Hilm’s marketing voice is specific and trustworthy until its public café demo exposes unfinished fields. The homepage promise, "Your website, ready in minutes. Edit it just by asking," is concrete, and the AI page says that anything not provided stays a visible placeholder until filled. The demo-cafe page contradicts that discipline with "SAR [price]", "[Number] seats", "[free / SAR —]", and "[Street name], Al-Rawdah, Jeddah". Replace every public placeholder with complete demo data or remove the block before linking it from the funnel. The vertical pages are useful but reuse the same acquisition block; add one audience-specific proof point before the shared CTA. Mechanical polish also remains: the company-profile title is 71 characters and the homepage has 24 images without alt text. Fix the demo first, then automate title and alt checks.

Hilm gives visitors a real plan recommendation, but the comparison table carries more detail than the decision needs. The pricing page says Free is enough to publish today and points to Pro when a buyer needs up to three bilingual sites without the hilm label. The concrete tradeoffs are useful: one versus up to three sites, one versus Arabic and English, and own domain plus SSL availability. The table then expands to 26 rows, including capabilities marked Included or checked in both plans, which dilutes the opening guidance. On mobile, the full three-column grid compresses plan labels, values, and checkmarks into a difficult scan. Keep the recommendation and the highest-value 8–12 axes visible, then move secondary details into grouped sections or use stacked plan cards. The Arabic pricing surface mirrors the same two-plan logic.

The supplied evidence shows no active traffic-suppression mechanism, but it cannot measure traffic health. Robots allows crawling and declares the sitemap; sampled public pages return 200 with self-canonicals; HTTP and www normalize in one hop; the nonexistent probe returns a real 404; and raw and rendered word counts match across all 40 samples. App login and account-creation URLs are intentionally noindex, nofollow and outside the sitemap. The remaining exposure is measurement and concentration: Search Console and analytics are unavailable, no SERP sample was supplied, and the commercial site centers on one website-builder theme. Connect GSC and analytics before attributing a change, then compare clicks, impressions, queries, landing pages, and sessions. Supporting content can hedge demand concentration if it answers real customer questions; it is not evidence of a current decline.

Hilm’s public content is coherent, but the release gate is not yet strict enough for a polished product surface. Both privacy and terms pages display "Draft — to be reviewed before launch. Not in effect yet," while remaining in public navigation; publish final text or remove those links until the policies are effective. The public café demo also contains "SAR [price]", "[Number] seats", and "[Street name], Al-Rawdah, Jeddah", making unresolved operational fields the clearest publish blocker. The marketing system itself is sound: the AI page gives a concrete promise and a six-step path from description to publish. Programmatic landing pages still reuse CTA wording and benefit bullets, so require one audience-specific example, objection answer, and proof section per vertical. Add automated checks for bracketed placeholders, title length, and alt coverage before release.

Hilm’s largest self-serve gap is the absence of a public help path. Useful raw material already exists: the restaurant page answers questions about menus, WhatsApp orders, Arabic/English availability, and changing prices. Turn those four topics into linked task articles with steps, a verification point, and troubleshooting, rather than leaving them as isolated conversion FAQs. Publish a navigation- or footer-linked Getting Started hub first, then organize the articles into searchable categories and add visible last-updated dates. The crawl also found hilm.ai/llms.txt absent with HTTP 404; that is secondary and should follow the human-facing help content, not substitute for it.

Hilm has a clean technical SEO base across the sampled public surface, with mobile LCP as the main practical opportunity. Verified robots allows crawling and declares the sitemap, which contains 52 URLs. Sampled public pages return 200 with self-canonicals; HTTP and www consolidate in one 301; the deliberate /templates trailing-slash normalization is one 307; and the nonexistent probe returns 404. Raw and rendered word counts match on all 40 samples, including 1,451 homepage words and 6,347 template words, while English/Arabic hreflang pairs are present. Review whether visible catalog structured data is appropriate for /templates/, where no JSON-LD or microdata was captured. Mobile LCP is 3.22–3.50 seconds on key pages, but no CrUX field metrics were supplied. Keep the intentional noindex treatment on app account flows.

Hilm is a strong choice for Arabic- and English-speaking small-business owners, professionals and small teams that need a publishable site without design or code expertise. Its clearest strengths are a direct product promise, 100/100 Usability and Accessibility scores, and 97.3/100 Technical SEO.

The priority fixes are clear. Build a public docs or help surface to address the 6.3/100 Docs & self-serve help score. Correct the tap-target defect behind the 73.3/100 Design execution score. Improve mobile delivery: the homepage's largest content arrives at 3.38 seconds, while representative Arabic landing pages are at 3.22 seconds and above. The legal pages also remain marked as drafts awaiting review, an editorial release issue identified in the crawl.

This is a 13-dimension review based on a crawl of 40 pages on 2026-10-21. The article uses the supplied public dimension score table and summaries, product facts extracted from the site's JSON-LD, table and text evidence, and the supplied accessibility and lab performance evidence.

Dimension 13, Review-content integrity, was excluded as not applicable. Google Search Console enrichment is pending: it is not connected and access is not available for this review. No competitor comparison is included because no competitors were provided in the review inputs.

Read How SiteList scores for the full methodology and interpretation notes.