Heylillian Review: Clear tools, uneven execution (54/100)

Heylillian is a web-based AI influencer marketing product reviewed from a 39-page crawl completed on 2026-10-14.

Fact	Value
Domain	heylillian.com
Category	AI influencer marketing research, creator discovery, and campaign intelligence
Platform	Web
Pricing	SaaS subscription; price range unknown
Pages crawled	39
Crawl date	2026-10-14

Heylillian earns 54/100 because clear navigation and a stable sampled surface are offset by weak design execution, limited self-serve guidance, and uneven content quality. Usability is the strongest buyer-facing result at 77.9/100: the supplied walkthroughs found a clear proposition and short paths to tools, pricing, the blog, and the Chrome Web Store handoff. Risk and stability also scores 81/100, with all 39 observed URLs returning 200 and public pages described as indexable.

The main weakness is design execution at 31.5/100. The supplied responsive evidence records a 1,400px document width at a 390px viewport on the homepage, making the current mobile implementation the clearest fix. Performance scores 55.2/100; real-user evidence reports LCP p75 of 3,746 ms and TTFB p75 of 2,975 ms.

The product category is understandable, but the intended buyer, cost, and proof behind the core promise require more inference. First impressions and positioning scores 60.5/100, while audience and messaging scores 52.4/100. Writing quality is 51.8/100, editorial QA is 41/100, and docs and self-serve help is 0/100. Together, those results describe a product that prospects can find and start evaluating, but not yet assess with equal confidence across messaging, mobile presentation, guidance, and editorial consistency.

The homepage names an "AI Influencer Marketing Assistant," but it leaves the intended buyer and proof case broad; Heylillian scores 60.5/100 on first impressions and positioning. The homepage promise to turn "every marketer" into an influencer-marketing expert establishes the job quickly, yet it does not specify a role, company type, spend level, or vertical. The hero’s "1/50th competitor cost" claim offers a measurable point of difference, but the supplied copy gives no comparison basis or attributable customer result. Make the primary buyer explicit, define how the cost comparison is calculated, and place one sourced case study beside the promise. That would turn understandable positioning into a credible reason to choose the product.

Heylillian’s messaging makes the product and next step visible, but its 52.4/100 score reflects unresolved buyer, pricing, and trust questions. The homepage exposes "Try Lillian Free," "View Pricing," Log in, and a Chrome Web Store download, so a high-intent visitor can start evaluating quickly. The phrase "every marketer" still asks brand-side teams, agencies, and creator-side users to infer whether the product fits them. The "1/50th competitor cost" claim also omits the unit, baseline, limits, and worked example in the supplied evidence. Name the primary buyer in the hero, show one transparent campaign-cost example, and pair the low-friction start path with an attributable customer outcome and methodology.

Heylillian’s 77.9/100 usability score comes from clear navigation and short paths, while mobile speed creates the main obstacle. The homepage links directly to pricing, the blog, free analysis tools, and the Chrome Web Store. However, the mobile lab run records 7.7 s FCP, 10.4 s LCP, 3,770 ms total blocking time, and 13.8 s interaction time. The free-start journey also uses several labels—"Try Lillian Free," "Start 7-day free trial," "Free Download," "Download," and "Search for free"—for the same external destination. Improve the mobile critical path first, then consolidate those labels into one primary action that states it opens the Chrome Web Store and explains when the trial begins.

Heylillian’s 74/100 accessibility score indicates a fair baseline with verified structural and perceivable issues. Lighthouse reports failures for the main landmark, form labels, image alt attributes, and color contrast on the homepage and pricing page. Rendered samples report no landmarks or skip link on the homepage, /tools/brand, /blog, and /tools/portrait. Missing-alt candidates total 5 of 90 homepage images, 12 of 18 on /tools/brand, and 12 of 19 on /tools/portrait; the blog heading outline also begins at h2. Add one semantic main region and a visible-on-focus skip link, provide programmatic names for every control, then fix image text alternatives and contrast by template. Confirm the result with keyboard and assistive-technology testing.

Heylillian’s 31.5/100 design-execution score is driven by responsive failures that outweigh its clear desktop hierarchy. At a 390px viewport, the homepage reaches 1,400px document width and the blog reaches 541px, pushing both templates beyond the mobile canvas. The mobile facts also report 18 small targets on the homepage and 6 on the blog, including 24px-high navigation links and 37×37px pagination buttons. Fix the shared shell, navigation breakpoint, and fixed-width hero children until both pages fit a 390px viewport. Then raise interactive targets to at least 44×44px with adequate separation, correct the Lighthouse contrast and label failures, and reserve image space with explicit dimensions or aspect ratios.

Heylillian’s 55.2/100 performance score reflects slow rendering and a heavy critical path, especially on mobile. Origin-level field data records LCP p75 of 3,746 ms and TTFB p75 of 2,975 ms, while CLS is 0. The harsher lab evidence shows homepage mobile LCP of 10,426 ms and total blocking time of 3,770 ms; pricing mobile reaches 8,701 ms LCP. PSI estimates 581 KiB of unused JavaScript on the homepage desktop run and 613 KiB on pricing mobile. Trace the LCP candidate and remove render-blocking work first, then split route code and defer non-critical modules. Re-test against LCP ≤2.5 s p75 and measure field INP rather than treating lab total blocking time as a substitute.

Heylillian’s 51.8/100 writing score reflects a clear commercial premise weakened by dense and repeated copy. The homepage extraction contains 614 words and 21 sentences in one paragraph, while sampled creator profiles also use one-paragraph blocks and a repeated scaffold. On Kyliekeolani’s page, specific figures—85.8K subscribers, 9.5K median views, and a 9.84 engagement-rate figure—sit beside the generic label "Strong Performance in Entertainment Niche." The blog compounds the editorial friction with the title "Home Page," no meta description, and no H1. Break long blocks into short evidence-led sections, replace reusable category labels with creator-specific interpretation, and give /blog a descriptive title, H1, and concise promise before its article cards.

Heylillian’s 81/100 risk-and-stability score reflects a stable sampled surface without a reported traffic decline. All 39 observed URLs returned 200, listed public pages were indexable, raw and rendered word counts matched across the sample, and HTTP plus non-www roots consolidated to the preferred HTTPS www host in one 301. The main resilience concern is concentration: 35 of 39 observed URLs are influencer profiles, tying much of the sampled footprint to one page class. Build internally linked campaign methods, platform comparisons, audience-fit frameworks, and evidence-rich brand cases around those profiles. Also redirect the live non-preferred /blog variant to /blog/ in one hop. Search Console data is unavailable, so clicks, impressions, coverage, and traffic magnitude are not measured here.

Heylillian’s 41/100 editorial-QA score reflects a functional content system that needs a stronger page-specific review pass. Sampled creator profiles contain 210–299 words in one paragraph and repeat "Creator Profile," "Audience Analysis," "Engagement Performance," and "Commercial Value" across distinct creators. The blog hub also carries the title "Home Page," no meta description, and no H1. Large claims such as a "120M+ Resource Library" and "100,000+ brand collaboration cases" have no adjacent source or definition in the supplied copy. Repair the blog’s metadata and heading order, add coverage dates and counting methods beside those figures, and require each profile to explain one creator-specific partnership implication plus uncertainty in audience inferences before publication.

Docs and self-serve help scores 0/100. Publish a public help hub and link it from the main navigation or footer. Start with a getting-started guide, task guides for analysis and creator discovery, Chrome extension and subscription questions, and a clear support escalation path. The llms.txt probe also returned 404; after the documentation exists, semantic headings and appropriate structured data can make the guidance easier for people and machines to navigate.

Heylillian appears best suited to influencer marketers who value a clear path into research, creator-discovery, and campaign-intelligence tools and are prepared to validate fit and cost for themselves. The 77.9/100 usability result and 81/100 risk and stability result show that the reviewed surface has a workable foundation.

Three weaknesses limit the case. First, the 31.5/100 design execution score reflects responsive-width problems in the supplied mobile evidence. Second, the 55.2/100 performance score includes LCP p75 of 3,746 ms and TTFB p75 of 2,975 ms. Third, docs and self-serve help scores 0/100.

Make the mobile layout production-safe, improve rendering speed, and publish a visible getting-started path. Clearer audience and pricing evidence would then make the product easier to assess without inference.

This 13-dimension review uses the supplied 39-page crawl from 2026-10-14, the public dimension score table, simulated cognitive walkthroughs, supplied Lighthouse samples, responsive captures, and supplied real-user performance data. SiteList assembled the frame from those inputs without changing the scores or adding unsupported findings. Read the full SiteList methodology.

Three dimensions were excluded. Decision-support surfaces and review-content integrity were marked not applicable. Technical SEO was excluded because no usable result was available, so this article makes no claim from that gap. Google Search Console was not connected and access was unavailable. No enrichment-pending items were supplied for this frame. The findings therefore describe the reviewed public surface and the evidence available on the crawl date.