Therly Review: strong support, slow first load (76.4/100) — SiteList

Fact	Value
Domain	therly.ai
Category	AI mental health support for private use
Pages crawled	40
Crawl date	2026-08-30

Therly presents a clear proposition: private AI emotional support, available 24/7. Usability scores 85/100, design execution 82/100, risk and stability 85/100, and technical SEO 88/100. Editorial QA scores 76/100, supported by expert review bylines, source references, and prominent medical disclaimers.

The main constraints are performance and decision support. Performance scores 58/100 because the homepage has a 4.1s mobile LCP; the 137KB hero image is rendered as a background without preload or fetch priority. Decision-support surfaces score 65/100, with a three-plan grid that has more than 12 feature axes and no defended recommendation.

Therly clearly presents a private AI emotional support tool, available 24/7. The proposition is easy to identify, but the category is inferred rather than expressed in conventional search language. Its audience remains broad: people seeking balance, clarity, and deeper self-understanding, rated L3 for implied specificity. Proof points are weak-grade; testimonials include names and roles but not specific outcomes. The site also uses terms such as ‘AI therapist’ and ‘psychologist’ at a jargon density of 2.5 per 1,000 words. Use a more familiar category description, name the intended segment directly, and add outcome-based proof so the first screen explains both fit and credibility.

Therly speaks to people seeking balance, clarity, and deeper self-understanding, but it does not explicitly name a target segment. That broad framing makes the service approachable while leaving prospective users to decide whether it fits their situation. The site answers some practical questions, yet does not address ‘What happens if it goes wrong?’ or ‘Can I trust them?’ in the reviewed material. Vocabulary alignment is low, with jargon density measured at 2.5 per 1,000 words. Testimonials provide names and roles, but not specific outcomes. Identify the primary audience, explain unfamiliar terms in everyday language, and add measurable or concrete customer outcomes to make the message more persuasive.

Therly’s usability is strong at 85/100: prominent ‘Start a chat’ and ‘Pricing’ calls to action lead users toward core tasks, and the header links to pricing, contact, and blog sections. The main friction is on pricing, where a privacy choices modal covers 30% of the screen and must be handled before plan details are accessible. The contact page places ‘User support’ and ‘Business inquiries’ in side-by-side boxes with identical styling, so labels do the work of differentiation. Footer links also vary in weight and color, such as bold ‘Pricing’ beside lighter ‘About service’. Make the consent prompt immediately dismissible, distinguish contact paths visually, and standardize footer links.

Therly’s design execution is sound, with strong hierarchy, consistent spacing, and mobile layouts that showed no horizontal scroll or tap-target failures. Two contrast defects materially reduce readability. Hero heading and subheading text rendered in white over the bright sky background at approximately 1.8:1, below the 4.5:1 WCAG AA body-text requirement. Pricing-page copy under ‘Three steps to clarity’ used #e8e8e8 on white at approximately 2.1:1. Both are straightforward color or background treatments. The pricing cards also use 16px and 24px internal padding, creating misalignment. Set a compliant dark text color or overlay, then standardize card padding and refine template-level font scaling.

Performance is Therly’s clearest technical weakness: mobile Largest Contentful Paint is 4.1s, associated with a 137KB, 1537×900px hero image served as a CSS background without preload or fetch priority. It is rendered at 1440px and lacks width and height attributes, leaving layout dimensions implicit. The network log also records a 188KB Google Tag Manager request in the head without async or defer attributes. Make the hero an appropriately sized image with explicit dimensions, preload it with high fetch priority, and defer non-critical analytics. These changes target the measured LCP path and reduce rendering contention.

Therly’s writing combines a compassionate voice with practical article structure, but its most visible marketing language conflicts with its stated boundaries. The homepage H1 and meta description say ‘Start a private therapy with an AI therapist & psychologist’, while /about describes the product as a digital self-support tool rather than a medical service or replacement for in-person work. Articles such as /blog/how-to-manage-anxiety provide actionable Box Breathing and 5-4-3-2-1 Grounding steps with safety disclaimers. Rendered headings also contain missing spaces: ‘AI therapistonlinefor private support 24/7’ and ‘When Therlycan help’. Align clinical terminology with the disclaimers and fix the template spacing before expanding the editorial voice.

Therly’s pricing page shows Basic, Unlimited Pro, and Therly Ultra, but the grid does not explain which plan fits which user. Therly Ultra carries a prominent ‘RECOMMENDED’ label without reasoning or a stated tradeoff. The feature presentation spans more than 12 axes and repeats overlapping items, including unlimited text chat across plans and differently worded memory features. Checkmark-heavy rows further obscure meaningful differences; Basic includes items such as ‘Trial session’ and ‘Basic state check-in’ rather than a clearly comparable entitlement. Reduce the grid to 8–12 decision-relevant axes, state the price and practical tradeoffs, and add an overridable recommendation tied to verifiable needs such as voice, memory, or support level.

Therly is technically stable, while public pages are indexable and /chat and /es/chat are intentionally noindex, nofollow, with no accidental noindex, clean redirects, valid HTTPS, and application routes correctly excluded from the index. Its principal exposure is strategic: AI mental health support is a sensitive YMYL vertical, so content authority and visibility in AI-generated search answers matter more than a current crawl blocker. The site has /authors pages, which provides a useful trust foundation. Every blog article should link to a verified expert profile with credentials. Five homepage images and one image on /es/terapia-cognitivo-conductual lack alt text, creating a direct accessibility and image-search gap. Historical migration patterns could not be verified because Wayback data was unavailable.

Therly shows good editorial QA discipline: articles identify Mark Root, MSc, as Lead Psychology Expert, display review dates such as ‘Updated June 7, 2026’, link author profiles, and place medical disclaimers near the top of sensitive articles. Source sections establish clear boundaries, including on English and Spanish localized pages. The key mismatch is between that editorial discipline and metadata claiming ‘AI therapist & psychologist’, despite /editorial-policy and /terms stating that Therly does not provide medical services or replace a clinical specialist. Metadata also needs routine cleanup: one title is 79 characters, and internal links repeat generic anchors 48 times for ‘blog’, 46 for ‘pricing’, and 27 for ‘try it’. Align claims, shorten titles below 60–65 characters, and add contextual in-body anchors.

Therly’s technical SEO foundation is strong at 88/100. Public pages return HTTP 200 with index, follow directives, while /chat and /es/chat use noindex, nofollow to limit application-route index bloat. HTTPS redirects both http and www variants in a single 308 hop, canonicals self-reference, and hreflang connects English and Spanish equivalents such as /blog and /es/blog. Raw-to-rendered text is near 1.0 across sampled templates, indicating that content is available without waiting for client-side hydration. The remaining polish is concrete: six images are flagged without alt text across key pages, and blog titles reach 66, 79, and 77 characters. Add descriptive alt text and trim those titles toward 50–60 characters.

Therly earns a strong 76.4/100 because its positioning is direct, its interface is intuitive, and its long-form content shows visible editorial care. The evidence supports it as a clearly positioned private AI emotional support tool, while the 4.1s mobile LCP and unclear three-plan choice limit the experience.

Reduce the homepage’s 4.1s mobile LCP and make the three-plan pricing choice clearer through stronger differentiation and a defended recommendation. These are fixable product and communication issues.

This 13-dimension review is based on a crawl of 40 pages completed on 2026-08-30, together with the public dimension score table and its one-line findings. Accessibility and Review-content integrity were excluded as not applicable. Google Search Console was not connected, so no GSC enrichment was available. See How SiteList scores for the framework.