Theysaid Review: strong, with clear gaps (75/100)

Theysaid is an AI-powered user testing platform for software and design teams.

Fact	Value
Domain	theysaid.io
Category	AI-powered user testing for software and design teams
Pricing	Price range not stated in the supplied crawl facts
Pages crawled	40
Crawl date	2026-09-28
Overall score	75/100

Theysaid presents a clear AI-powered user testing proposition for UX researchers, designers, and product managers, earning strong results for first impressions, audience messaging, usability, and design execution. Its technical base is also a major asset: risk and stability scores 96/100, while technical SEO scores 94/100 with open crawl paths, consistent self-referencing canonical signals, and perfect raw-to-rendered content parity across sampled templates.

The weaker areas are concentrated in the experience users encounter after the initial impression. Accessibility scores 65/100 because of critical form-label failures, missing skip links, and empty image alt text. Performance scores 78/100, with a 3.2-second mobile LCP compared with 0.7 seconds on desktop. Writing quality is the clearest content problem at 48/100: sampled pages place 1,000-plus words in a single paragraph, which makes the value harder to scan.

At 75/100, Theysaid has a credible product foundation. Better page structure, accessibility corrections, and visible self-serve help would make that foundation easier to use and trust.

Theysaid clearly presents an AI-powered user testing platform, but its hero copy does not name the category directly. The H1, ‘AI feedback fast enough to keep up with AI code and design’, signals speed while leaving the product type implicit. Navigation points to UX researchers, designers, and product managers, giving the B2B audience an implied rather than explicit fit. Category language remains readable, with jargon at 2.5 terms per 1,000 words. The brand appears as ‘TheySaid’ while the domain is ‘theysaid.io’, a minor consistency issue. Add a plain category phrase to the H1 or supporting line, and make the intended audience explicit.

Theysaid names UX researchers, designers, and product managers through its navigation, so the intended audience is clear enough for a B2B SaaS site. The messaging also answers feature questions, but it does not explicitly address the cost of using the platform; the pricing page only signals an enterprise contact-sales tier. Trust evidence is the larger gap: no testimonials or case studies were found. Add a visible pricing explanation and customer stories to reduce the questions a prospective buyer must resolve alone. Keep the technical vocabulary where it serves this audience, while adding a plain-language explanation around the product promise.

Theysaid communicates its core value clearly and places a ‘Get Started - Free’ CTA in the hero, but the conversion path becomes less direct after that first view. The main navigation has no sign-up or trial CTA, and the ‘Voice conversations’ feature page has no visible route to start a trial or demo. Visitors must return to the homepage or search for the next step. Labels such as ‘AI User Testing’, ‘Surveys’, and ‘AI Interviews’ are precise for specialists but may slow first-time visitors. Put a conversion CTA in the main navigation and on every feature page, then simplify labels around user goals.

Accessibility is the clearest functional weakness: the audit score is 65/100, with keyboard navigation, form context, image alternatives, and contrast all needing correction. The homepage has no skip link, so keyboard users cannot jump directly to main content. The blog contains 7 radio inputs without a label, aria-label, or aria-labelledby, leaving their purpose unclear to screen readers. The homepage also reports 79 images with empty alt text; each should be confirmed as decorative or given meaningful text. Address these structural issues first, then resolve the reported color-contrast failures.

Theysaid’s visual system is consistent and responsive: 72 colors and 59 font sizes are managed through a token system, and mobile checks found no horizontal scroll. Two hero CTA treatments are the material design defect. Pink text values #fb7185 and #f94b7e on white produce contrast ratios of 2.9:1 and 2.8:1, below the 4.5:1 body-text requirement. A navigation link also measures 42×35px on mobile, below the 44×44px target. Darken the CTA text to the recommended #4B5563, enlarge the navigation target, and reduce the roughly 92ch paragraph measure.

Performance is acceptable on desktop but needs targeted mobile work: homepage LCP is 3.2 seconds on mobile versus 0.7 seconds on desktop. The large hero image has no explicit dimensions and is 1,440px wide while rendered at 1,224px in the mobile capture, increasing work without matching the viewport. Blocking requests from cdn.prod.website-files.com total 2,137KB across 6 requests. Add image dimensions, serve responsive sizes, and defer non-critical requests. The two font families, Inter and Archivo, also lack unicode-range subsetting; limiting the character set would reduce font payload.

Writing quality scores 48/100 because the page structure makes otherwise clear SaaS promises difficult to scan. The homepage reports paragraph_count: 1 with an average paragraph length of 1,355 words; /features/templates repeats the pattern at 1,159 words. Break each block into 40–80 word paragraphs under descriptive H2s. The homepage H1 also contains the typo ‘keep upwith’, which weakens a claim about speed. Internal anchors repeat ‘ai interviews’ and ‘voice conversations’ 77 times each, giving navigation a templated feel. Preserve the stronger benefit-led feature H1s, including the clips-highlight-reels example, and vary surrounding anchor language.

Theysaid has a strong stability posture, with no accidental noindex, redirect loops, migration artifacts, or broad structural breaks found in the supplied evidence. Raw-to-rendered parity is 1.0 across sampled pages, so metadata and content remain available without client-side injection. Sitemap URLs align with the crawl inventory, and Wayback snapshot density is consistent from January 2023 through August 2026. The remaining exposure is concentrated in oversized assets and missing alt text, which can affect performance and image discovery. Maintain the current canonical, redirect, and noindex configuration; use GSC to establish an ongoing traffic baseline and monitor search changes.

Public self-serve help is the largest discoverability gap: the score is 15/100. No public help path was discovered in the 40-page crawl, so common setup and feature questions have no visible public destination in the reviewed pages. Add a /help or /docs section to the main navigation and publish a Getting Started guide first. A dedicated FAQ can answer recurring buyer questions; only after that foundation exists should the site add FAQPage schema or an /llms.txt resource.

Technical SEO is a major strength: robots.txt permits all user-agents, the sitemap contains 492 URLs, and every crawled page returns 200 with a self-referencing canonical. Raw HTML matches rendered output at a 1.0 ratio across sampled templates, while TLS 1.3, HSTS with max-age=31536000, and CSP frame-ancestors ‘self’ are present. The /features/ask-ai page correctly uses noindex and stays out of the sitemap and main navigation. The polish work is mostly asset and image metadata: sampled pages report 79, 35, and 26 images without alt text, while SVGs reach 1,436KB, 870KB, and 822KB. Add context-aware alt text and compress oversized assets.

Theysaid is clearest when it states what the product is and who it serves. The site earns 85/100 for first impressions, 85/100 for usability, and 87/100 for design execution, supported by clear visuals, a clear product proposition, and a responsive visual system using CSS custom-property tokens. Its risk and stability score is 96/100, and its technical SEO score is 94/100, showing that the underlying site is dependable.

The next pass should rank the work: first make pages scannable and correct the confirmed form-label, skip-link, and empty-alt-text issues; then add public self-serve documentation and address the 3.2-second mobile LCP.

Theysaid is best suited to software and design teams evaluating AI-powered user testing, especially UX researchers, designers, and product managers who value a clear product proposition and stable technical delivery.

This is a 13-dimension review based on a crawl of 40 pages completed on 2026-09-28. The review combines the supplied public dimension score table with crawl evidence, including page content, accessibility checks, performance measurements, design-system observations, stability signals, and technical SEO signals.

Three dimensions were excluded from the article's dimension sections: 12 · Decision-support surfaces was not applicable; 13 · Review-content integrity was not applicable; and 19 · Editorial QA of content was missing. Google Search Console was not connected, so no GSC enrichment was used.

Read the full scoring approach in How SiteList scores. The byline identifies this as a 13-dimension review; excluded and missing dimensions are disclosed here so the 75/100 score is read in context.