A complete analysis synthesized from Semrush data across domain authority, backlink audit, AI search visibility, organic keyword rankings, position tracking, traffic analytics, on-page audit, and competitor gap analysis.
Authority Score 32 is a solid baseline for a growing AI-tooling product. The key divergence: India-specific organic traffic is declining (−0.83%) while worldwide traffic is growing strongly (+11.04%). This signals seomaxim ai's content resonates better globally — likely reflecting a US-heavy user persona — while local Indian SERP relevance is contracting. Zero paid traffic in India means no safety net for organic dips.
A traffic cost of just $11 (IN database) signals extremely low commercial-intent keyword coverage. Confident AI, a direct competitor, has a traffic cost of $7,600 — a 690× difference in monetizable organic value. The keyword set is dominated by branded and navigational terms, not commercial AI evaluation queries. This is the most important strategic gap in this entire report.
Two black-hat Telegram link networks ("TG @BHS_LINKS" and "TELEGRAM @SEO_ANOMALY") have each injected exactly 100 backlinks from 100 separate domains. These 200 spam links are the primary driver of the HIGH toxicity classification and are actively suppressing rankings for competitive keywords. Disavowing these via Google Search Console's Disavow Tool is the single highest-ROI action available. Takes 1–2 hours; effects visible within 4–6 weeks.
Dozens of pages that have earned backlinks — especially old blog.seomaxim.ai posts — no longer exist, wasting the ranking power those links carry. Implementing 301 redirects to the nearest live equivalents recovers this equity at zero cost.
Referring domains have grown continuously from ~394 (Jun 2025) to ~1.5K (May 2026) — a 3.3× increase in unique linking domains in 12 months. This is the most important backlink metric (domain diversity) and it's trending strongly upward. New and lost domain counts since November 2025 consistently show more new referring domains acquired than lost each week. Peak acquisition was late Nov/early Dec 2025 (up to 69 new domains in one week). A March 2026 spike brought ~7.6K new backlinks in one week.
Total backlinks peaked at ~80K around Jan–Feb 2026 and have since dropped to 70.2K (−11%). The divergence between domain count (growing) and link count (falling) suggests quality is improving but volume is eroding — potentially healthy pruning of low-quality sources, but worth investigating to rule out indexation drops of legitimate links.
A 9% .ai TLD share signals peer AI-ecosystem products are linking in — strong topical authority signal. The 44% "other TLD" category suggests a highly diverse international link base, which is a positive quality signal for Google's domain diversity assessment.
| Country | Referring Domains | Share | Notes |
|---|---|---|---|
| 🇺🇸 United States | 426 | 32% | Dominant — confirms US-primary audience |
| 🇸🇬 Singapore | 80 | 6% | Strong APAC presence |
| 🇩🇪 Germany | 34 | 3% | European developer community |
| 🇫🇷 France | 28 | 2% | |
| 🇨🇳 China | 25 | 2% | |
| 🇲🇩 Moldova | 21 | 2% | ⚠️ Anomaly — likely link farm. Audit for toxicity. |
| URL | Referring Domains | Signal |
|---|---|---|
| https://www.seomaxim.ai/ | 362 | Homepage — core authority hub |
| https://seomaxim.ai/ | 97 | Non-www variant (canonical issue?) |
| https://www.seomaxim.ai/bifrost | 83 | Breakout product page — generating own link equity |
| https://www.seomaxim.ai/docs/introduction/overview | 29 | Docs-as-SEO is working |
"seomaxim ai" (345 links), "seomaxim" (193), and "Bifrost" (136) show strong brand recognition. Top anchors also include "visit website" (very large — directory/aggregator CTA links), "book a demo," "careers," "security," "privacy policy," "terms of service," and a long URL anchor. Empty/image anchors also significant. This protects brand SERPs but the keyword-intent anchor gap is critical.
The dominance of "visit website" and empty anchors (from directories/aggregators) means Google receives almost no topical anchor signals for "llm evaluation," "llm testing," or "ai agent testing." seomaxim ai currently has only 9 backlinks containing "llm evaluation" anchor text. Confident AI built its rankings on exactly this anchor strategy. Building even 50–100 editorial links with keyword-rich anchors from AI/dev publications would materially move rankings.
| Competitor Domain | Competition Level | Category |
|---|---|---|
| portkey.ai | 24% | LLM gateway / observability |
| langwatch.ai | 20% | LLM observability |
| futureagi.com | 20% | AI evaluation |
| getinboxzero.com | 19% | AI email |
| helicone.ai | 18% | LLM ops |
| Platform | Mentions (IN) | Cited Pages (IN) | Reading |
|---|---|---|---|
| Google AI Mode | 7 | 27 | Best performer in India |
| ChatGPT | 6 | 16 | Moderate presence |
| Gemini | 4 | 11 | Decent presence |
| Google AI Overview | 0 | 2 | Critical gap — 0 mentions in India |
| Platform | Mentions (WW) | Cited Pages (WW) |
|---|---|---|
| Google AI Mode | 129 | 389 |
| Gemini | 102 | 138 |
| ChatGPT | 66 | 254 |
| Google AI Overview | 32 | 90 |
Google AI Overview has 0 mentions in India with only 2 cited pages. AI Overviews appear at the very top of Google SERPs for informational queries — this is a significant missed opportunity for Indian traffic. AI Overview favors structured, authoritative, well-cited content with clear factual claims, FAQ schema, and structured headers. Key product pages need restructuring for this format.
Google AI Mode leads in India (7 mentions, 27 cited pages) and globally (129 mentions, 389 cited pages). AI Mode is Google's newer conversational search layer, and early strong presence here positions seomaxim.ai well as this channel grows. This is the most promising AI search channel to double down on.
The site audit shows an 87% AI Search Health score — "better optimized for AI search engines" than average. 165 AI-related issues exist, but the blocked-from-AI-search section confirms ChatGPT-User, OAI-SearchBot, Googlebot, and Google-Extended are all "All good" — crawling access is fully open to all AI crawlers. The remaining score gap is content structure, not access.
| Keyword | Position | Traffic | Traffic % | Volume | KD% | Type |
|---|---|---|---|---|---|---|
| bifrost | #1 | 377 | 52.72% | 2,900 | 63 | Product |
| maxim | #4 | 162 | 22.65% | 5,400 | 83 | Brand |
| clinc | #5 | 92 | 12.86% | 590 | 30 | Competitor |
| bifrost ai | #3 | — | — | — | — | Brand |
| ai prompt engineering course | #24 | 24 | 3.35% | 1,600 | 63 | Informational |
| deepseek r1 price | #4 | 5 | 0.69% | 590 | 36 | Pricing Tool |
| groq pricing | #8 | 5 | 0.69% | 880 | 49 | Pricing Tool |
| claude 3.7 pricing | #6 | — | — | — | — | Competitor Pricing |
| gpt-4o-mini pricing | #7 | — | — | — | — | Competitor Pricing |
| gpt-image-1 pricing | #7 | — | — | — | — | Competitor Pricing |
| llm evaluation | Not ranking | — | — | ~1,200 | ~45 | TARGET GAP |
| llm testing | Not ranking | — | — | ~210 | ~30 | TARGET GAP |
| llm evaluation metrics | Not ranking | — | — | ~800 | ~40 | TARGET GAP |
The top 2 ranked keywords are pure brand terms ("bifrost," "bifrost ai") — meaning seomaxim.ai primarily ranks for its own product name, not for broader problem-space queries that would drive new user discovery. Ranking for competitor pricing pages (Claude 3.7, GPT-4o-mini, GPT-image-1) is clever content marketing, but these are low-intent traffic unlikely to convert to seomaxim ai customers.
| Page URL | Traffic | Traffic Diff. | Keywords | Signal |
|---|---|---|---|---|
| /bifrost/llm-cost-calculator/xai/grok-3 | 958 | +958 | 18 | Surging ↑ |
| /bifrost | 458 | +146 | 4 | Growing ↑ |
| / (homepage) | 323 | −545 | 139 | Declining ↓ |
| /bifrost/llm-cost-calculator/llamagate/nomic-embed | 237 | +237 | 2 | New ↑ |
| /blog/best-llms-for-legal-ai-agents | 79 | −1 | 13 | Stable |
The LLM cost calculator sub-pages are the fastest-growing organic asset. The Grok-3 page alone earns 958 visits (+958 new). These pages rank for specific model pricing queries with low KD. This programmatic approach should be directly applied to "llm evaluation" use cases — create comparison/benchmark pages for evaluation metrics, testing frameworks, and methodology guides. Bifrost's /oss-friends page has earned 83 referring domains independently, confirming product pages generate their own link equity.
The homepage, which ranks for 139 keywords, dropped 545 visits in the tracked period. This is a major signal — possibly keyword cannibalization from Bifrost sub-pages, a Google Core Update impact, or the HIGH toxicity score suppressing homepage rankings. Connect Google Search Console to identify which specific queries dropped and correlate with the Semrush trend chart dates.
At 83%, seomaxim.ai scores 9 percentage points below top-10% websites (92%). With only 12 fully healthy pages out of 100 crawled, the technical debt is significant. This gap likely translates to crawl efficiency losses and ranking suppression on competitive SERPs.
Markup (100%), Crawlability (99%), and HTTPS (98%) are excellent — core technical infrastructure is solid.
No hreflang or international targeting configured. With referrals from 6+ countries and a global AI developer audience, this is a missed geo-targeted traffic opportunity.
| Issue | Type | Pages | Priority |
|---|---|---|---|
| JavaScript & CSS total size too large (>3.3MB per page) | Warning | 65 pages (65%) | HIGH |
| Links with no anchor text (hurts AI Search indexing) | Info | 130 pages | HIGH |
| Low text-to-HTML ratio (thin content signal) | Warning | 74 pages | MED |
| Duplicate title tags | Error | 2 pages | MED |
| Content not optimized for AI Search | Info | 3 pages | LOW |
65 out of 100 crawled pages exceed 3.3MB of JavaScript and CSS per page. This is the most widespread and technically impactful issue. Bloated JS/CSS directly causes: (1) slower Core Web Vitals (LCP, FID, CLS), (2) poor mobile performance, (3) higher bounce rates, (4) lower Googlebot crawl budget efficiency. The docs pages (seomaxim CLI, PM Cookbook, Custom Logs Dashboard, etc.) are particularly affected. This is likely a build bundling/code-splitting issue in the Next.js or similar framework powering the docs site. Target: under 1MB total JS/CSS.
130 pages contain links with no descriptive anchor text. This hurts traditional SEO (Googlebot can't understand link context) and AI search (LLMs can't extract semantic relationships from empty anchors). This is the single highest page-count issue and disproportionately impacts AI search indexing quality.
74 pages have a low ratio of actual text content versus HTML markup — a common pattern in React/Next.js SPAs where content is dynamically rendered. Google may classify these as "thin content" and deprioritize them in rankings. The 2,890 warnings in the audit are almost certainly dominated by this issue at scale. Adding substantive text content (feature descriptions, use cases, FAQ sections) to key docs and product pages directly addresses this.
Average visit duration jumping from ~2 minutes to 7 minutes 46 seconds is the standout metric. This suggests users are deeply engaging with content — likely spending time in docs, product demos, or interactive features. Combined with +29.54% unique visitor growth, this indicates strong product-market fit signaling from user behavior.
A bounce rate of 70.51% (+8.09%) is high, but must be contextualised against the very long average session duration. This pattern — high bounce + long session — signals a single-purpose visit pattern: users land on a specific docs page or tool page, spend extended time reading, then leave. This is normal for developer tool sites. However, the rising bounce rate warrants monitoring to rule out landing page irrelevance for certain traffic segments.
| Channel | Trend | Est. Apr Volume | Notes |
|---|---|---|---|
| Direct | ▲ Dominant, growing | ~15K | Grew from ~5K (Nov) to ~15K (Apr) — brand recognition rising |
| Organic Search | ▲ then plateau | ~15K | Grew to ~11K by Jan, recovered near 15K in Apr |
| Referral | ▲ Surging Feb–Apr | ~15K | Dramatically rising since Feb 2026 — now co-equal with direct/organic |
| Organic Social | → Flat | ~2–3K | Minor contribution, no meaningful growth |
| Paid Search/Social | → Negligible | ~0 | No paid channel investment anywhere |
The referral channel has surged dramatically since February 2026, now competing with direct and organic as one of the top 3 traffic sources. This directly aligns with the growing referring domain count (1.3K–1.5K) and suggests link building efforts and/or product virality (links shared in developer communities, GitHub, forums) are materially translating into real traffic — not just domain authority.
No investment in paid search or paid social is visible across any data. For a B2B SaaS product in competitive AI tooling, this means 100% reliance on organic, direct, and referral channels. This is a defensible strategy if organic growth continues, but creates vulnerability — any algorithm change or competitive content surge could disproportionately impact traffic with no paid safety net.
The page /blog/what-are-online-evaluations-and-how-to-set-them-up-for-your-ai-system-using-seomaxim-ai/ — seomaxim ai's closest page to "llm evaluation" — has ZERO keyword usage of "evaluation/evaluate" in its Title, H1, and Meta Description. Body usage is 0.21% vs rivals' average of 0.42%. TF-IDF score is 0.0005 vs rivals' max of 0.003 — 6× underoptimized. This page should be ranking but sends no keyword signal.
The model-library compare page, gemini provider page, and the grok-4-0709 calculator page all triggered "The Semrush crawler couldn't find this page." This suggests noindex tags, JavaScript rendering issues, or robots.txt blocks on these URL patterns. If Google also can't crawl them, these pages earn zero organic value despite having backlinks pointing to them. Run a Google Search Console URL Inspection report immediately on all /bifrost/model-library/* and /bifrost/llm-cost-calculator/* patterns.
Confident AI built dedicated, deeply optimized blog posts for each evaluation concept — "llm evaluation metrics," "llm-as-a-judge," "g-eval," "llm testing" — each with exact-match keyword in title, H1, meta, and body at appropriate density. They then earned backlinks using those exact phrases as anchor text. Maxim has the product capability to match and exceed them; the gap is purely content depth and targeted anchor text acquisition strategy.
| Gap Category | Maxim AI Status | What's Needed | Priority |
|---|---|---|---|
| Spam Disavowal | 200 Telegram spam links actively suppressing rankings | Disavow via Google Search Console — 1–2 hours | Critical |
| Keyword-Rich Anchors | Only 9 backlinks with "llm evaluation" anchor text | 50–100 editorial links with exact "llm evaluation" / "llm testing" anchors | Critical |
| Dedicated Evaluation Page | No /llm-evaluation/ landing page exists | Build long-form hub: title/H1/meta all targeting "llm evaluation" — 3,000+ words | Critical |
| On-Page Signals (Eval Blog) | 0% keyword in title, H1, meta; body 0.21% vs rivals 0.42% | Rewrite title, H1, meta; double body keyword density to ~0.40% | Critical |
| Crawlability | 3 of 4 audited high-value pages uncrawlable by Semrush | Audit robots.txt, noindex tags, JS rendering on model-library and calculator pages | High |
| JS/CSS Bundle Bloat | 65 pages exceed 3.3MB — hurts Core Web Vitals | Code-splitting, lazy loading, tree shaking — target under 1MB total | High |
| Semantic Coverage | Missing "llm-as-a-judge", "g-eval", "evaluation metrics", "agent evaluation" | Write dedicated posts for each concept (2,000+ words each) | High |
| 404 Broken Backlinks | 49 URLs receiving backlinks return 404 | 301 redirects from old blog.getmaxim.ai URLs to current equivalents | High |
| Empty Anchor Links | 130 pages with no descriptive anchor text | Add descriptive anchor text to all internal links — critical for AI search | High |
| Google AI Overview (India) | 0 mentions in India — key SERP feature missed | Add FAQ schema, structured headers, factual claims to key pages | High |
| Domain Authority Gap | AS 32 vs competitors likely 45–65 | Earn links from high-AS domains: HuggingFace, dev.to, DigitalOcean, Medium | Medium |
| Duplicate Title Tags | 2 pages with confirmed duplicate titles | Fix immediately — directly suppresses rankings on affected pages | Medium |
| Thin Content (74 Pages) | Low text-to-HTML ratio — drives the 2,890 warnings | Add substantive text content (features, use cases, FAQ) to key pages | Medium |
| Schema Markup | No Product/Aggregate Rating schema on calculator pages | Add JSON-LD Product schema to all calculator pages — boosts SERP CTR 15–30% | Medium |
| Internal Linking | Bifrost pages not linking to evaluation content | Add keyword-rich internal links from 1,000+ visit Bifrost pages to evaluation hub | Medium |
| Moldova Link Audit | 21 referring domains from Moldova — anomalous for US AI product | Toxicity audit; disavow if spammy via Google Search Console | Medium |
| Homepage Traffic Drop | −545 traffic loss unexplained (139 keyword page) | Connect GSC + Analytics; identify which queries dropped and correlate with dates | Investigate |
| International SEO | No hreflang or geo-targeting configured | Implement hreflang for US, IN, SG, DE, FR markets | Low |
getmaxim.ai is on a strong growth trajectory — genuinely impressive momentum in referral traffic (surging since Feb 2026), link acquisition (3.3× domain growth in 12 months), and user engagement (+187% session duration). The product is clearly resonating. However, several structural technical and content issues are creating a ceiling on organic growth. The gap to Confident AI is not a product gap — it is a content depth, anchor strategy, and technical hygiene gap. All of it is fixable within 3 months.