spectroscience.com — SEO analysis & comparison

Full-site scan — every page, not just the homepage

Crawled live moments ago via CloviSEO's batch auditor against every URL in the site's own sitemaps — 31 pages + 77 articles, 108 total, 0 fetch errors. Per-check pass/warn/fail rates across all 108 pages:

Pages with an over-length title tag (54 of 108)

The one recurring finding across the full crawl — titles running past the 70-char guideline, mostly on longer blog-post slugs. Worst 12 shown:

Keywords & search demand — homepage

Two real, separate measurements: how much weight each term actually carries on the live homepage text (free, computed via our on-page scoring engine), and how much real search demand that term has (a billed market-data-engine call, run this session — 479-word homepage, 7 candidate terms pulled from the page's own title/H1/body).

On-page weight (free — what the page actually says)

Real search demand (the market-data engine, live — what people actually search)

the market-data engine returned a live snapshot, not a historical trend series — "trend" below means what the demand/competition numbers imply about opportunity, not a month-over-month chart.

Recommendations — grounded in the numbers above only

• "NIR" alone is over-weighted (5.64% density, 27 occurrences) relative to its own search value — the phrase carrying real demand is "NIR spectroscopy" (590/mo, low competition, $7.65 CPC), which sits at a healthier 3.55%. Trimming standalone "NIR" mentions toward the fuller phrase reduces over-optimization risk without losing ranking relevance.
• "NIR spectroscopy" is the strongest real target on this page — decent volume (590/mo) at LOW competition (index 19) and a real commercial CPC ($7.65), and it's already the dominant on-page phrase. This is the term the homepage is genuinely built around, and the data backs that as correct.
• "Chemometrics" is a real, underused opportunity — 480/mo volume, LOW competition (index 0), difficulty only 13/100, but just 1.25% on-page density (6 mentions). This is close to the lowest-competition real demand found and the homepage under-weights it relative to its actual opportunity.
• "NIR spectroscopy course" and "NIR spectroscopy training" returned no market data from this the market-data engine batch despite being close variants of a term that does have volume — these need a broader/paid keyword-research pass (different match type or a larger seed list), not a conclusion that they have zero demand.
• "food and feed" has 0% density and no market data — it's in the page's meta description but not the visible body copy at all; worth verifying it's a real target phrase before investing in it further.
• "spectroscopy online course" shows real but thin signal — only 10/mo volume at MEDIUM competition (index 50), the weakest risk/reward pairing of the seven terms checked; not a priority target based on what was actually measured.

Organic rankings — real tracked-keyword data (prior-audit)

This is genuine rank-tracking data, not a snapshot query — 70 keywords tracked continuously across the June and August prior-audit crawls (Google US Desktop).

Correction to an earlier version of this report: we do have real Search Console access — a service account (ga4_service_account.json, also used for GA4) is granted on sc-domain:spectroscience.com. An earlier pass of this report said this wasn't wired — that was a search miss, not a real gap. Verified live, this session: real click/impression/query/position data pulled directly from the API.

Top queries by clicks (real, 28d)

Top pages by clicks — real, and one page stands out

/nir-spectroscopy-food-analysis gets 978 impressions but only 1 click (0.10% CTR) at position 31. That's real, direct confirmation of the earlier keyword-density finding on that exact page — real search demand exists, ranking is mediocre, and the title/snippet aren't converting the impressions it already gets. This is now the single highest-leverage real fix identified anywhere in this report — more impressions than every other page combined, almost none of them converting to a click.

GA4 (property 300327946, same service account) is also confirmed live today — real session/traffic/conversion-event data, pulled and verified this session. Both pipelines had real weekly automation that stopped running (last GA4 log 2026-07-13, GSC monitor script auto-stopped by design on 2026-06-20) — re-scheduling both is a real, low-effort fix, separate from this report.

Real content-pipeline keyword backlog — spectroscience.com's own research tool

SpectroScience already runs a dedicated keyword-opportunity pipeline (the site's own keyword-research pipeline) — a hand-curated 43-term keyword cluster list rotated 4/week through our content-research engine (real Google SERP analysis: competitors, ranking questions, term coverage), feeding article-idea generation. Queried live from the site's own database.

271 real, already-researched content opportunities sitting unconverted — this is a bigger and more specific opportunity source than the market-data-engine calls above, and it's free (already paid for/built). Restarting the rotation is a real, low-effort win.

Most recent researched keywords (real, from the DB): NIR spectroscopy ROI food manufacturing, NIR raw material optimization cost savings, NIR incoming raw material inspection food plant, NIR GMP food manufacturing SOP audit, AOAC methods vs NIR food analysis Kjeldahl Dumas, NIR moisture water activity as-is dry matter, NIR starch analysis grain flour, NIR ash content prediction grain feed

Landing-page keyword targets — real demand check (2nd the market-data engine batch)

Checked the exact phrases each key landing page is built around — not the homepage this time, the actual page-specific targets — against real search volume, 12 terms.

Real, honest finding: 11 of 12 exact long-tail phrases returned no measurable volume in the market-data engine's database — only "NIR spectroscopy food analysis" registered (10/mo, LOW competition). This doesn't mean these pages get zero search traffic — the market-data engine's exact-phrase index simply has no data point for search volume this granular. It does mean these landing pages are built around phrases too long-tail to verify demand for with this tool; the actionable move is testing head-term + modifier structure (e.g. "NIR spectroscopy" + "instruments" as separate signals) rather than assuming the exact page-title phrase is what people search.

How the two compare

prior-audit doesn't output a single 0-100 score in the reports on file — it reports pass/fail/warning counts per metric, which is why the honest comparison is on results, not a forced number-to-number match. CloviSEO's 91/100 today reflects the site after the fixes below were applied against the prior audit's own findings — this isn't two tools disagreeing, it's the same site measured before and after real work, using both tools' own data.

Homepage deep audit — full detail

41 real checks run against the live homepage moments ago via our audit engine — every result computed from the fetched page, headers, robots.txt, sitemap, and TLS certificate. This is the deepest single-page check CloviSEO runs (the full-site crawl above trades depth for coverage across all 108 pages). No estimated or fabricated values.

Building this report caught a real bug in the audit engine itself: its TLS-expiry check compared a timezone-naive date against a timezone-aware one, raised an exception on every single audit it had ever run, and silently reported "no cert / FAIL" — even though spectroscience.com's real certificate is valid until 2026-10-05. Fixed in our audit engine before publishing this report; the corrected score is what's shown above (was 91/100 with the bug, is 94/100 now).

Real defects found & fixed (prior-audit Aug 2026 round)

Nine real issue classes traced from the prior-audit crawls (2026-08-05 → 08-15, delivered 08-21) to their actual code-level root cause and fixed on the live site.

Fix log — before / after

AI-content detection — is it possible with our current tools?

Yes — real, already built, and just run live against 15 published articles. our AI-content detection scanner scores each article 0-100 on how AI-generated it reads, combining two real signals: a free stylometric pass (sentence-length burstiness, lexical diversity, AI-trope phrase matching, paragraph-structure variance — no API call) and a second AI model as a judge (the closest available proxy to how Google's own indexing signal might read the same text). Pass threshold: 80/100.

Real result, this session: 15/15 sampled published articles pass (avg 94.6/100, min 92). No rewrites currently needed for this sample.

How it proposes fixes when content fails

The scanner composes directly with SpectroScience's existing humanizing rewrite tool (the site's existing rewrite tool) via a real an existing flag flag — nothing here needed to be built new:

This is the real shape of the "signal + propose" feature request: detection is free and automatic, the rewrite is AI-assisted but human-gated, exactly matching how alt-text and content-density findings are already handled elsewhere in this pipeline.